madVR Options Explained [Archive]

View Full Version : madVR Options Explained

Asmodian

15th February 2015, 05:00

Current as of madVR v0.92.17 (http://forum.doom9.org/showthread.php?t=146228)

Please ask questions or offer corrections and suggestions. Thanks. :)

devices

device name: [EDID Product Name] A name for the display (customizable).
device type: [unknown] Sets the icon for the display. When set to projector the 'screen config' section is available.

identification
EDID information, some displays are detected with different identifications depending on how they are attached or detected. You can drag and drop these identification entries to combine displays.

properties

the display expects the following RGB output levels: [PC levels (0-255)] This is what your display expects to receive. 'PC levels (0-255)' means “use a 16-235 YCbCr to 0-255 RGB conversion matrix” and 'TV levels (16-235)' means “use a 0-255 YCbCr to 0-255 RGB conversion matrix”. If the source is flagged full range either a 0-255 YCbCr to 0-255 RGB or a 0-255 YCbCr to 16-235 RGB conversion matrix is used instead.
the native display bitdepth is: [8 bit] This is the bit depth madVR will dither to. madVR offers very high quality dithering so if you are using a low bit depth display (e.g. a 6-bit TN) you may get better quality when setting this to 6-bit. Due to limited support for high bit depths, madVR dithers to 8-bit even when this is set to 9 or 10-bit except when using D3D11 'full screen exclusive' (or D3D11 fullscreen windowed in Windows 10 with AMD or Nvidia GPUs). It can be interesting to turn this down to a very low value (2-4) while testing the dithering options but make sure to have the 'trade quality for performance' option 'don’t use linear light for dithering' unchecked even if you leave it checked for normal use.
3D format: [auto, disabled] Set the 3D format to: auto, side-by-side, top-and-bottom, row alternative, or column alternative. There is also an option to swap the left and right eyes.

calibration [disable calibration controls for this display]

disable calibration controls for this display: The display is assumed to have a BT.709 gamut with a pure power gamma of 2.2 (probably). If outputting HDR madVR assumes the display is expecting BT.2020 with the PQ transfer function.
this display is already calibrated: This sets the current display characteristics. The gamut option affects the conversions used by madVR when playing video with different gamuts. The gamma options only have an effect when using 'tone map HDR using pixel shaders' for HDR content or when using the 'enable gamma processing' option in 'color & gamma'.
[B]calibrate this display by using yCMS: yCMS offers a simple gamut and gamma calibration method. It requires a few measurements of the display that can be done with HCFR (http://sourceforge.net/projects/hcfr/). If gamut or grayscale measurements are not supplied only the other correction is performed. Usually bad results (mostly banding and odd effects in shadows) are obtained if gamma measurements below ~20% IRE are used so only enter values above 15-30% IRE.
calibrate this display by using external 3DLUT files: Use a calibration package such as Argyllcms, Calman, or LightSpace to generate a 256x256x256 3DLUT to perform a full calibration. For Argyllcms see this AVS Forum thread (http://tinyurl.com/nx87y7c), for Calman (http://tinyurl.com/map94jl), or for LightSpace (http://www.lightillusion.com/madvr_manual.html). If the provided 3DLUT file contains calibration curves (e.g. in the case of ArgyllCMS, the collink "-a" or "-H" options were used), madVR will load the calibration curves into the system-wide GPU hardware gamma ramps, which affects everything displayed, not only madVR. madVR will restore the original gamma ramps (if any) when it is closed.
disable GPU gamma ramps: [Disabled] Clears system-wide GPU gamma ramps (monitor calibration), if any, replacing them with linear (no-op) curves. This affects everything displayed, not only madVR. madVR will restore the original GPU gamma ramps (if any) when it is closed. This option changes its behavior when using the 'rendering -> general settings' option 'enable windowed overlay' to only apply to madVR because madVR is emulating the gamma ramps in that case. This option has no effect if a 3DLUT file that contains calibration curves is loaded (see above).

display modes [None]
Allows setting refresh rate and resolution automatically depending on the source video.
list all display modes madVR may switch to: An example for a modern 4K display: 2160p23, 2160p24, 2160p25, 2160p29, 2160p30, 2160p50, 2160p59, 2160p60
treat 25p movies as 24p (requires Reclock or VideoClock): [Disabled] Uses 24 Hz modes to display 25 fps content, it requires Reclock or VideoClock to slow down the audio to match.
hack Direct3D to make 24.000Hz and 60.000Hz work: [Disabled] A hack to Direct3D that enables true 24 and 60 Hz display modes in Windows 8.1 or 10 which are usually locked to 23.976 Hz and 59.940 Hz. May cause presentation queues to not fill.

custom modes
A second tab in display modes. Allows configuring custom refresh rates and can assist in optimizing them for the purpose of avoiding dropped or repeated frames without frame blending (smooth motion) or resampling the audio (ReClock, VideoClock). Only custom timings can be optimized but simply editing a mode and applying the "EDID / CTA" timing parameters creates a custom mode, and is the recommended way to start optimizing a refresh rate. New timing parameters must be tested before they can be applied. Delete replaces the add button when selecting a custom mode. It uses each of GPU vendor's private APIs to add these modes and does not work with EDID override methods like CRU; supports AMD, Intel, and Nvidia GPUs. With Nvidia these custom modes can only be set to 8-bit.

color & gamma [Disabled]

pure power curve: [Default] use the standard pure power gamma function
BT.709/601 curve: Use the inverse of a meant for camera gamma function. It has a linear section near black so shadows lighten faster as you move away from black. This can be helpful if your display has crushed shadows.
2.20, 2.40, etc.: [2.20] changes the display's gamma, based on the display's settings in 'calibration', to this target gamma. Lower brightens mid range values which can be nice in a brightly lit room. Higher darkens mid range values which might look better in a darker room.

hdr [let madVR decide]

let madVR decide: Sends HDR data to displays capable of HDR, otherwise coverts to SDR using pixel shader math.
passthrough HDR to display: Passes the HDR video to the display as is, along with the metadata if 'send HDR metadata to the display' is enabled.
tone map HDR using pixel shaders: Enables the options below-
-target peak nits: [200] Set the target display's peak white brightness. Above this value is used as the highlights. A lower setting increases the brightness of mid range values.
-tone mapping curve: The curve used when compressing the range.
[B]-color tweaks for fire & explosions:: Keeps the color approximately the same while de-saturating and/or reducing luminance to bring a pixel to a value the display can represent.
[B]-highlight recovery strength: [None] Sharpening the highlights after range compression.
-measure each frame's peak luminance: [Disabled] Allows madVR to only compress highlights as much as necessary on a per frame basis resulting in more detail in highlights.
-output video in HDR format: [Disabled] Add HDR metadata to the output video.
convert or process HDR content by using an external 3DLUT: Requires display calibration hardware and software. Possible with DisplayCal (Argyllcms) or Lightspace.

screen config [Disabled]
This section is hidden unless the device is set as a projector.

define visible screen area by cropping masked borders: Allows scaling to a smaller resolution and adding black pixels to the edges. Useful if the projector is overshooting the edges of the screen. Only active when full screen.
activate lens memory number: Sends a command to a JVC or Sony projector to activate an on-projector lens memory number.
anamorphic lens: Allows output to non-square pixels.

processing

deinterlacing

if in doubt, activate deinterlacing: [Disabled] deinterlace video not flagged as progressive
if in doubt, deactivate deinterlacing: [Enabled] deinterlace video flagged as interlaced
force film mode: [Disabled] force IVTC, reconstructing the original progressive frames from video encoded as interlaced, decimating duplicate frames if necessary. madVR's IVTC is not functional when using DXVA2 or D3D11 native GPU decoding because madVR's IVTC algorithm runs on the CPU and the decoded video data is never copied to system memory.
force video mode: [Disabled]force DXVA deinterlacing, uses the GPU's native deinterlacing. This is the same as automatic and cannot decimate frames.
only look at pixels in the frame center: [Enabled] As described; good for broadcast (with its messy edges) but bad for Anime.

artifact removal

reduce banding artifacts: [Disabled, low/high] Fade detection does not run if both options are set to the same value for a small increase in performance. Five frames are removed from madVR's rendering buffer every time a fade is detected if the trade quality for performance option 'don't rerender frames when fade in/out is detected' is disabled.
reduce ringing artifacts: [Disabled] Attempts to reduce any ringing/halos present in the source caused by downscaling or sharpening before madVR, e.g. during mastering.
reduce compression artifacts: [Disabled] A deblocking and deringing algorithm. high quality is more significant at higher strengths and is much slower. Chroma often has artifacts but they are usually less obvious and it takes a lot of power to process the chroma channels too.
activate only if it comes for free (as part of NGU): [Disabled] Only runs RCA if it can be run as part of NGU image upscaling. Requires NGU Sharp high for medium quality and NGU Sharp very high for high quality.
reduce random noise: [Disabled] A denoising or grain reduction algorithm. Chroma often has less random noise than the luma.

image enhancements [Disabled]
These are applied before scaling. See 'upscaling refinement'

zoom control

disable scaling if image size changes by only: [Enabled, ... 2 lines or less] If the resolution needs scaling by the number of pixels set or less no image scaling is done, black pixels are added to the right and/or bottom edge instead.
move subtitles: [Enabled, ... to bottom of the screen/window] Move subtitles, this is important when removing black bars or subtitles can be displayed outside the visible screen area. Cannot move PGS subtitles if they are encoded as fullscreen images or Advanced Substation Alpha (ASS) subtitles which have a specific location defined.
automatically detect hard coded black bars: [Disabled] This enables the options below. madVR can detect and remove black bars on all sides of the video. Black bars cannot be detected when using DXVA2 or D3D11 Native decoding.
if black bars change pick one zoom factor: [Disabled, ... which doesn't lose any image content] Using this setting madVR will avoid changing the zoom or crop. e.g. It will not zoom or crop a 16:9 section in a normally 4:3 film when set to "which doesn't lose any image content" and will zoom or crop all of the 4:3 footage the amount needed to remove the black bars from any 16:9 sections when set to "which doesn't show any black bars".
if black bars change quickly back and forth: [Disabled, ... once every 2 sec] If not using the above option this option allows a limit on how often madVR can change the zoom or crop while playing to remove black bars as they are detected. Without either of these options madVR will always change the crop or zoom to only remove the black bars.
notify media player about cropped black bars: [Disabled, ... no more than once every 2 sec] How often to notify the media player of changes in the black bars. Some media players use this information to resize the window.
always shift the image: [Disabled, ... to the top of the screen] This allows setting whether the top or bottom of the video is cropped when zooming.
keep bars visible if they contain subtitles: [Disabled, ... forever] Disables zooming or cropping any black bar in which subtitles are detected. Either forever or for a set time.
cleanup image borders by cropping: [Disabled, ... 1 line from all black bar borders] Crop extra non-black pixels away from black bars or all edges. When set to crop all edges they are cropped even when no black bars are detected.
if there are big black bars: [Disabled, ... reduce bar size by 25%] Allows setting separate handling of large black bars.
zoom small black bars away: [Disabled] This zooms into the video to remove the black bars, cropping a small amount of video off of one edge to maintain the original aspect ratio and resizing to the original display resolution. e.g. The bottom is cropped when cropping small black bars on the left and right and the video scaled to the original resolution.
crop black bars: [Disabled] This crops black bars, changing the display aspect ratio and resolution. Cropping black bars increases performance; the pixels no longer need to be processed. Profile rules referencing resolution will use the after crop resolution.

Asmodian

16th February 2015, 05:39

scaling algorithms

Nearest Neighbor: {super fast} A low quality method, the value of the closest pixel is used. This looks very bad when not doing integer scaling (1x, 2x, 3x, etc.). Configure a custom keyboard shortcut to enable it.
Bilinear: {super fast} A low quality method. It is as fast as Nearest Neighbor because GPUs have hardware to do bilinear filtering in a single operation.
DXVA2: {very fast} Uses the GPU drivers to access hardware scaling. Transfer to and from hardware is done at the video bit depth so it may introduce banding or cause scaled dither noise. It can be very fast and low power. The quality varies between GPU brands and models but none are particularly good. Can do chroma upscaling as well so this option might override 'chroma upscaling'. Requires a restart of madVR to take effect. Does not bypass the graphics card's video (damage) algorithms so make sure they are off.
Cubic: {fast} Bicubic is b = 0, c = value / 100. SoftCubic is b = value / 100, c = 0. Mitchell-Netravali is b = 1/3, c = 1/3. Bicubic125 and 150 (c > 1) are special cases for downscaling because they have nasty artifacts when used for upscaling.
[B]Lanczos: {medium} [3] A sharp sinc type
Spline: {medium} [3] A different sinc type
Jinc: {slow} A more advanced sinc type, uses 3-taps (based on an EWA LanczosSharp from ImageMagick)
Bilateral: {slow} A chroma scaler that uses the luma channel as reference. This can be very good but it can fail too.
Reconstruction: {very slow} [soft] Similar to Bilateral in that it uses luma channel information to aid chroma scaling. Soft is less prone to artifacts.
SSIM: {very slow} [1D-100%] Downscaling only. It is very sharp and detailed, especially useful with high downscaling ratios and fine details. with Shiandow (http://tinyurl.com/hat4kbq)
super-xbr: {very slow} [100] xBR means scale By Rules, it works by detecting edges and interpolating pixels along them. Can only perform a 2x upscale.
NGU Family: {variable super slow} [medium] madshi's own Next Generation Upscaling. A sharp doubling algorithm with minimal artifacts. NGU Soft, Standard, and Sharp can also quadruple the luma in one step. The Anti-Alias version is softer but reduces aliasing and is more tolerant of source artifacts. Works in Windows XP.

activate anti-ringing filter: [Disabled, relaxed] madshi's method to reduce ringing (https://forum.doom9.org/showthread.php?t=145358), the major artifact caused by sharp scalers. Designed to only remove the ringing introduced by the filter it accompanies.

A comparison of scaling algorithms made by Warner306 (https://forum.doom9.org/showthread.php?t=175429)

chroma upscaling
Cb and Cr are stored at half the resolution of Y in a standard YCbCr 4:2:0 source and this controls how madVR doubles their resolution in both dimensions to create YCbCr 4:4:4 for the conversion to RGB. When downscaling a lot this setting is used to bring the chroma to the target resolution, not the luma resolution, so the luma is downscaled before the conversion to RGB.

[B]activate SuperRes filter, strength: [Disabled, 3] A special version of SuperRes designed for the chroma upscaling step. Not recommended with NGU.

image downscaling
How madVR scales to the target resolution if it is smaller than the source.

[B]scale in linear light: [Disabled] Using linear light when downscaling produces a result more similar to what one would get with optical scaling; bright specks such as stars in a night sky are preserved much better. The chroma must be upscaled to match the luma before downscaling when using this option because linear light is only possible in RGB.
activate anti-bloating filter: [Disabled, 100%] A post process method to reduce the fatness created by downscaling with SSIM.

image upscaling [Lanczos 3]
How madVR scales to the target resolution if it is larger than the source.

scale in sigmoidal light: [Disabled] Linear light but for upscaling, applies a sigmoidal curve after the conversion to linear light. This reduces the dark halo artifacts around sharp edges caused by upscaling in linear light.

NGU or super-xbr: When choosing a doubling scaler the 'algorithm quality', 'activate doubling / quadrupling...', and 'if any (more) scaling needs to be done' options become available. Algorithm quality is much more important for luma than chroma. "let madVR decide" chooses options appropriate for the performance requirements of the chosen luma algorithm quality. It is recommended to turn the luma quality up as high as possible before adjusting other options.

algorithm quality:
<-- luma: The option used for doubling or quadrupling the resolution of the colorless component of the video (Y).
<-- chroma: The option used for doubling the resolution of color components of the video (Cb & Cr).
activate doubling / quadrupling...: Changes the scaling ratio required before doubling or quadrupling activates. When not active the option set in 'if any (more) scaling needs to be done' is used to scale to the display resolution, if required. 'let madVR decide' uses 1.2x and 2.4x respectively (when using direct 4x NGU quadrupling "if any upscaling past 2.0x is needed" is used instead) so set these to 2.0x and 4.0x if you want to avoid doubling or quadrupling to a resolution larger than the display and downscaling.
if any (more) scaling needs to be done:
<-- upscaling algo: The option used for upscaling, if required, after doubling or quadrupling.
<-- downscaling algo: The option used for downscaling if the display resolution is lower than the doubled or quadrupled resolution.

upscaling refinement [All Disabled]
These are applied as part of upscaling or supersampling.

sharpen edges: [1.0] Only sharpens edges instead of textures like skin or cloth. Avoids sharpening artifacts too much.
crispen edges: [1.0] A tamed version of FineSharp, a sharpener originally by Didée that attempts to keep local energy close to the original. Better used with higher quality sources because it sharpens artifacts.
thin edges: [1.0] As its name implies. Good for SD Anime or cartoons.
enhance detail: [1.0] Sharpens textures like skin or cloth, also sharpens artifacts.
soften edges: [1] Applied as part of NGU. Useful for large upscaling ratios with this very sharp upscaler.
add grain: [1] Applied as part of NGU. Useful for large upscaling ratios with this very sharp upscaler. Combats the artificial flat look caused by the combination of a lack of detail in non-edge textures with very sharp upscaling of edges.
LumaSharpen: [0.65] Blurs the original pixel with the surrounding pixels and then subtracts the blur. Avoids sharpening artifacts.
AdaptiveSharpen: [0.5] Tries to sharpen medium sharp edges the most, it avoids sharpening near flat areas and very sharp edges. From bacondither (http://tinyurl.com/pmmn6gv)
activate anti-bloating filter: A post process method to reduce the fatness created by line sharpeners. Not needed for crispen edges, which is already "skinny."
activate anti-ringing filter: A post process method to reduce ringing due to sharpening.
SuperRes: [3] A post process method from Shiandow (http://tinyurl.com/kh8lta5). Not recommended in combination with NGU.
----use linear light: Ideally use the same method that the source was originally downscaled with when it was mastered. This is usually not known but gamma light might be more likely. When enabled the image may be slightly darker and dark lines slightly fatter.

Calculate an initial guess (using the configured upscaler)
Downscale and calculate differences with original image.
Scale those differences to the final size (SoftCubic50)
Improve guess by:

Softening the image
Subtracting differences with the original image
Sharpening
Removing aliasing
Removing ringing

Repeat steps 2-4 several times.

rendering

general settings

delay playback start until render queue is full: [Disabled] Pauses playback until all the buffers are full. With this option disabled only the present buffer needs to fill before playback starts.
enable windowed overlay (Windows 7 and newer): [Disabled] Only available on Nvidia and Intel GPUs. Uses a low level overlay method which bypasses the GPU LUT (monitor profile) so madVR emulates it when using this option, this is done in 16-bit so madVR can provide better quality than the GPU. Overlay also bypasses the OS to a large extent; screen-shots are not possible. D3D9 Only. Incompatible with 'fullscreen exclusive mode'.
enable automatic fullscreen exclusive mode: [Enabled] madVR has exclusive access to the display, nothing else can draw to the display without dropping out of this mode. This is the most stable mode for madVR because it has the most control over when and how video frames are displayed. There is a slight flicker and delay as madVR enters and exits this mode. Required for 10 bit in Windows 7 or 8.
disable desktop composition (Vista and Windows 7): [Disabled] Disables Aero which has its own v-sync that can conflict with madVR’s. This option cannot disable desktop composition on Windows 8 or 10 but desktop composition also works better on 8 and 10.
use Direct3D 11 for presentation (Windows 7 and newer): [Disabled] Use Direct3D 11 instead of D3D9. Requires a restart of madVR to take effect. Required for >8 bit output. Slightly more responsive than D3D9. Overrides windowed overlay.
present a frame for every VSync: [Enabled] when disabled madVR only presents new frames when needed, relying on D3D11 to repeat frames for each VSync. Disabled may improve performance but may also cause presentation glitches.
use a separate device for presentation (Vista and newer): [Disabled] Might improve performance slightly, or it might hurt it slightly.
use a separate device for DXVA processing (Vista and newer): [Disabled] Might improve performance slightly, or it might hurt it slightly.
CPU queue size: [16] The buffer in system memory for decoded video and xySubFilter subtitles.
GPU queue size: [8] The buffer in the graphics memory for madVR’s rendering.

windowed and exclusive mode settings

present several frames in advance: [Enabled] Uses a new method for presenting frames in advance instead of using a back-buffer queue.
how many video frames shall be presented in advance or how many backbuffers shall be used: [8 or 3] The present queue, a final buffer in DirectX, after madVR sends completed frames to Windows.
when and how shall the GPU be flushed: [flush, flush & wait (sleep), don't flush, don't flush] To be able to display rendering times in the OSD the '... after last render step' must be one of the 'flush & wait' options.

stereo 3d
madVR only supports frame packed 3D with the LAV video decoder.

enable stereo 3d playback: [Enabled] Allow madVR to output stereoscopic 3D content. 3D output is currently limited to 1080p23 with smooth motion automatically disabled.
when playing 2d content: [Disabled, disable os stereo 3d support for all displays] Change the stereo 3D setting in the OS when playing 2D content.
when playing 3d content: [Disabled, enable] Change the stereo 3D setting in the OS when playing 3D content. Depending on the display and GPU, 3D support may not need to be enabled in the OS to display 3D content.
restore os stereo 3d settings when media player is closed: [Disabled]

smooth motion [Disabled]
Uses frame blending to perfectly display the video’s frame rate at the display’s refresh rate without any dropped or repeated frames and while staying in sync with the audio. This eliminates frame time judder in exchange for slightly increased blurring due to the frame blending. The blurring is more visible when the refresh rate matches the source frame rate because all the frames will be blended frames. The blurring is less visible on higher refresh rate displays because fewer frames will be blended frames. Smooth motion will harm motion interpolation, do not use them at the same time.

only if there would be motion judder without it...: [Default] Smooth motion will activate when the refresh rate is not close to a multiple of the frame rate, e.g. it activates for 23.976 fps at 60Hz but not for 29.976 fps at 60Hz.
... or if the display refresh rate is an exact multiple of the movie frame rate: Smooth motion will always activate unless the frame rate matches the refresh rate, e.g. it will also activate for 29.976 fps at 60Hz but will still not activate for 23.976 fps at 24Hz.
always: Smooth motion is always active. This allows no dropped or repeated frames even when the source frame rate is very close, but not quite equal to, the real refresh rate of the display. Note that smooth motion has less blurring when displaying 23.976 fps at 60 Hz than 23.976 fps at 24 Hz so this option is not recommended unless you cannot increase the display's refresh rate.

dithering
Dithering (http://en.wikipedia.org/wiki/Dither) is performed as the very last step in madVR to convert its internal 16 bit data to the bit depth set for the display. Any time madVR does anything to the video high bit depth information is created and dithering allows much of this information to be preserved when displayed at a lower bit depth. For example, the conversion of 8-bit YCbCr to RGB generates >8-bit RGB data. The higher the output bit-depth, as set in madVR's 'devices' -> 'properties', the less important dithering quality is.

None: Only use for testing, it is never useful for normal use, banding.
Random Dithering: Highest noise, but it is also the fastest option.
Ordered Dithering: [Default] This option has the lowest visible dithering noise. It uses a fixed dither texture that was optimized for low visible noise and offers high quality dithering with a low performance hit.
Error Diffusion: Uses DirectCompute to perform very high quality error diffusion dithering, similar to the Floyd–Steinberg or Sierra dithering algorithms. Test option 1 and 2 for yourself (or pick one randomly). They are both very good and my preference seems to change depending on the display.
use colored noise: [Enabled] Uses an inverted dither pattern for green which reduces luma noise but adds chroma noise. I like this option disabled, but test it for yourself.
change dither for every frame: [Enabled] Uses a new dither seed for every frame or, for ordered dithering, adds a random offset and rotates the dither texture 90° between every frame.

Asmodian

16th February 2015, 14:56

screenshots
When taking screenshots madVR re-renders the current video frame based on the settings in this tab.

create screenshots with a custom zoom factor: 100% view means the native video resolution viewed at 100%, only resizing to square pixels. Disabled captures the current video as displayed.
don't perform color & gamma processing, calibration etc: It is recommended to leave this enabled if you plan to share or view your screenshots on any other display.

trade quality for performance
'scale chroma separately, if it saves performance'[/I], are enabled]

[B]optimize subtitle quality for performance instead of quality: This does not hurt subtitle quality very much for a reasonable performance gain.
use DXVA chroma upscaling when doing native DXVA decoding: DXVA chroma upscaling returns incorrect colors on some systems but it is fast. DX9 DXVA2 Native hardware decoding returns bad quality (blurred chroma) on Nvidia GPUs, which this option does not change.
use DXVA chroma upscaling when doing DXVA deinterlacing: DXVA chroma upscaling returns incorrect colors on some systems but it is fast.
store custom pixel shader results in 16bit buffer instead of 32bit: Does not seem to improve memory use or performance, at least on my Titan X (Pascal).
don't use linear light for dithering: Especially significant at lower output bit depths. Usually not visually significant at 8+ bits.
don't analyze gradient angles for debanding: More detail is lost for the amount of banding removed.
don't rerender frames when fade in/out is detected: Without this option debanding causes madVR to rerender five frames every time it detects a fade, but only when it is set to a higher strength for fades. This can cause dropped frames or presentation glitches on some systems.
scale chroma separately, if it saves performance: Scaling chroma separately is much faster when downscaling from large resolutions. It is probably lower visual quality but it is not obviously mathematically worse and it can be very important for watching 4K content on lower resolution displays with weaker GPUs.

The rest are probably not a good idea and can cause visual errors and/or offer little performance gain.

More Advanced Topics

On Screen Display
Toggle the On Screen Display by pressing Ctrl-J.

To avoid dropped frames and/or presentation glitches the sum of the average stats "deinterlace" (if present), "rendering", and "present" must be a bit below the frame time, Y, in the line "v-sync [X] ms, frame [Y] ms". For example, progressive 29.97 fps video has a new frame every 33.37 ms so "rendering"+"present" needs to be a few ms below 33.37 ms. However, 23.976 fps video only has a new frame every 41.71 ms so you can use more demanding settings with lower frame rate video. Exactly how far below the frame time is required for glitch free playback is dependent on the system but a few milliseconds is usually sufficient.

Profiles
Right click on a device in 'devices' or the 'processing', 'scaling algorithms', or 'rendering' settings folder to create a profile group. You can only add settings pages to a group when creating it. There can only be one profile in a profile group when deleting it. If madVR hits the end of a profile rule script without picking a profile it uses the top profile in the group.
configure profile rules (http://tinyurl.com/k9tgtcr)

HDR Support
madVR supports the Windows API, introduced with the Windows 10 Creator's update, as well as Nvidia's and AMD's APIs to pass HDR meta data to a display. The display must report HDR support. Applications cannot use the Windows API to turn HDR mode on or off, instead content is converted by Windows when needed, at a lower quality than madVR is capable of. This means that AMD's or Nvidia's API is required for maximum quality when viewing both SDR and HDR content, without manually changing a Windows setting. The Windows setting should be off when using another API. HDR usually does not work with fullscreen exclusive mode. As of Nvidia driver version 390.65 madVR can use Nvidia's API for good HDR behavior in Windows 10 1709 (Fall Creators Update).
madshi's detailed description of HDR support in Windows 10. (https://forum.doom9.org/showthread.php?p=1810891#post1810891)

Rotation Support
madVR will automatically rotate the video if rotation is specified in the video container, rotation can also be controlled manually.
The default keyboard shortcuts for rotation are:
Control + Shift + Alt + Right Arrow | Clockwise
Control + Shift + Alt + Left Arrow | Counterclockwise

IVTC and Deinterlacing
Shortcut keys to force auto deinterlacing, DXVA2 deinterlacing, or CPU based IVTC. Forcing a deinterlacing mode also forces deinterlacing to 'on' instead of 'auto'.
Control + Shift + Alt + D | Toggle Deinterlacing: [Auto], On, Off
Control + Shift + Alt + T | Toggle Mode: [Auto], DXVA2 Deinterlacing, madVR's CPU IVTC

madVR can decimate progressive streams
e.g. 720p60/720p50 broadcast with 24p/25p content. Press control+alt+shift+d to enable deinterlacing and press control+alt+shift+t to force decimation on. It has the same limitations as madVR's IVTC or film mode.

File Name Tags
madVR will look at the folder and file names for tokens or "magic words" that tell it to treat a particular file differently.
example: "D:\Movies\Broken PAL matrix=PAL\my broken rip from the 90s deband=high deint=Video.mkv"

Possible file name tags include:
frame rate: frameRate=24p, 24i, 24fps, 24Hz, etc. or refreshRate=
deinterlacing: deint=On|Off|Video|Film|IVTC
debanding: deband=off|low|medium|high
deringing: dering=on|off|half|full
profile: profile='profile name'
matrix: matrix=2020|709|601|NTSC|PAL|YCgCo|240M
primaries: primaries=2020|DCI|709|SmpteC|EBU|sRGB|NTSC|PAL|470M|240M|170M
levels: levels=PC|TV|fullrange|limited|doubleExp|tripleExp or range=
blacklevel: blacklevel=-50 to +50
whitelevel: whitelevel=-50 to +50
contrast: contrast=-100 to +100
brightness: brightness=-100 to +100
saturation: saturation=-100 to +100
hue: hue=-180 to +180
chroma deblocking: deblockChroma or denoiseChroma
black bars: detectBlackBars=on|off" or "blackBarDetection=on|off"
hdr: hdr=on|off or transfer=hdr|sdr|2084|709

Hidden Options
An empty folder or file in the madVR program directory with a specific name will trigger a hidden option, these are options you probably do not want to use.

BilinearSuperRes: Use a lower quality, but faster, mode for SuperRes.
DisableHdrBrightnessTweak: Disables raising the brightness of shadows which normally occurs when 'this display's peak luminance capability' is set to 120 or 180 nits.
DontRenderAfterStop: Stops madVR from rendering for 7 seconds after playback is stopped or paused. It may stop black screen flashing during seeks and play rate changes with specific players.
force full range input: Forces madVR to treat all input as full range.
KeepLastFrameOnStop: Prevents a black screen when stopping playback.
ShowBlackBars: Draws a green rectangle around the active video area, as detected by zoom control.
ShowRenderSteps: Displays every shader pass in the OSD with the rendering time for each one. Will slow down the overall rendering time due to needing to flush the GPU and wait for the rendering to succeed after every shader pass. Depending on the number of steps and the GPU this effect can be significant or relatively minor. Only active when the OSD is being displayed.
YCbCr: Outputs YCbCr data instead of RGB, as if it was RGB.
ShowHdrMode

Known Limitations or Complications

-madVR does not like f.lux, some GPU monitoring, or other similar programs.
-DXVA2 Native DX9 hardware decoding returns bad quality (blurred chroma) on Nvidia GPUs. The DX11 hardware decoding API (D3D11VA) that is used when LAV Video is set to D3D11 Automatic (so native decoding is used) avoids these problems. So do all copy-back hardware or software decoding methods.
-All Native hardware decoding methods result in some functional ramifications. It is impossible to use force film (IVTC) and the detection of black bars does not work. madVR can still shift the entire video (including any hard coded black bars) to the top or the bottom of the screen, as well as crop pixels off all sides, but it cannot detect black bars. Again, copy-back hardware or software decoding avoids these problems.
-SLI reduces performance when enabled, even if it is not enabled for madVR or the video player.
-When using a 3DLUT madVR can generate out of gamut values to display WtW information unless using 'enable windowed overlay', in which case WtW information is clipped.
-Windowed Overlay mode won't linear out the Windows calibration when loading a 3DLUT with a linear calibration attached, it will simply stop emulating the calibration loaded in Windows. This way Windows can remain calibrated while madVR uses a 3DLUT with a linear calibration attached.
-madVR's IVTC can detect a lot of cadences but can "only" handle some of them correctly. For example the rare cadence 3:2:3:2:2 (25 fps in 30i 60 fields) does not work well.
-when using display mode changing madVR always switches to 23Hz, if available, when having forced film mode on while playing 59i content. This is often optimal, but not always. So, with no good way to know what would be best in advance, madVR goes with 23Hz.
-madVR's 3D support is currently limited to 1080p23.

Example rendering path
https://res.cloudinary.com/asmodian/image/upload/v1570099552/madVR/madVR_UHD_HDR_Example.png

settings.bin
madVR's settings can be backed up or shared via a settings.bin file. My settings are optimized for the Nvidia 2080 Ti, Windows 10, and on my LG CX 48" 4K OLED TV with profiles for SD, 720, 1080, UHD, and above. 3D is disabled. I use Zoom Player Max (http://www.inmatrix.com/files/zoomplayer_download.shtml) for the video player. Note that "low" still requires a decent GPU (~1050 Ti).

To use simply restore madVR to default settings and place the file in the madVR program directory. madVR will automatically create or update the file with new settings if it has write access to its directory (e.g. not in "program files"). madVR will still use the registry to store these settings, along with any changes, without write access. :)

For madVR v0.92.17 or later:
Max Quality settings.bin (https://www.mediafire.com/file/aq8bevrirokiry2/madVR_MaxQuality.zip/file)
Low Power settings.bin (http://www.mediafire.com/file/k28zu6228oxpocw/madVR_LowPower.zip/file)

Feedback is appreciated, especially for the low power settings.

noee

16th February 2015, 15:41

Nice so far. Maybe put a link in for madshi's write-up of profiles. You might also add a section for filename tokens.

huhn

16th February 2015, 17:05

the calibration is not limited to 3d LUT. and you completely skipped color & gamma. the gamma setting is very useful for rooms with changing light conditions.

Asmodian

16th February 2015, 22:37

Nice so far. Maybe put a link in for madshi's write-up of profiles. You might also add a section for filename tokens.

A Profiles link is easy, a second link to madshi's "(click here for online help)" couldn't hurt. ;)

filename tokens really should be there, they need more documentation. I am having trouble finding a good list of possible tokens though. Does anyone have such a list?

the calibration is not limited to 3d LUT. and you completely skipped color & gamma. the gamma setting is very useful for rooms with changing light conditions.

I figured those did not need further explanation. madshi's descriptions seem sufficient if you know what color gamut and gamma are or are used to messing with them. I did not want to have to explain all these concepts for those who do not. I actually had a few lines about color & gamma but deleted the section as it did not say anything helpful (i.e. "you can use this section to mess with color and gamma").

I will add a note about this display is already calibrated and disable GPU gamma ramps. If there is anything in particular you think I should mention please let me know! :)

edit: I thought of something to write about for color & gamma, e.g. your mention of gamma setting being useful for lighting conditions.

Warner306

16th February 2015, 22:54

You should explain what dithering is and why it is performed. Currently, there is one sentence. It could be a small paragraph. That is a topic that most newbies will not understand. I am still not clear on this. Is it not that the conversion of YCbCr to RGB requires dithering due to mathematical rounding errors present is the conversion calculation? Also, is dithering not used to take high bit-depth output of madVR and dither it down to the bit-depth of the display? A dither is an introduction of noise (simulated color information) to smooth out color bands.

Most people overlook dithering settings because they don't understand what they are adjusting.

Asmodian

16th February 2015, 23:26

Sounds reasonable :)

huhn

17th February 2015, 11:48

I figured those did not need further explanation. madshi's descriptions seem sufficient if you know what color gamut and gamma are or are used to messing with them. I did not want to have to explain all these concepts for those who do not. I actually had a few lines about color & gamma but deleted the section as it did not say anything helpful (i.e. "you can use this section to mess with color and gamma").

I will add a note about this display is already calibrated and disable GPU gamma ramps. If there is anything in particular you think I should mention please let me know! :)

edit: I thought of something to write about for color & gamma, e.g. your mention of gamma setting being useful for lighting conditions.

2.20, 2.40, etc.: changes the display's gamma based on what is set in calibration to this target gamma. Lower brightens mid range values which can be nice in a brightly lit room. Higher darkens mid range values which might look better in a darker room.

well done.

Warner306

18th February 2015, 00:11

Not everything is clear, but I appreciate the effort. You have taught me several things I didn't know.

My wish is to have an introduction that summarizes the video rendering process. My curiosity surrounds how the whole process fits together, which helps me decide what settings are important and what aren't. Because, for most of us, madVR settings are a compromise based on the limitations of your hardware. You can't just turn everything on at the maximum settings.

I'm thinking something like:

The Video Rendering Process Summarized:

Chroma Upscaling to YCbCr 4:4:4 to match native luma resolution.
YCbCr to RGB conversion for both luma and chroma.
Image Scaling to target resolution. This may include luma/chroma doubling.
Introduction of dithering and debanding to reach desired RBG bit depth.
Picture appears on the screen.

Something like that. I have no idea if the order is right. It puts the whole guide into context and helps people understand what it all means.

I have been putting together a beginners guide to using madVR with an MPC with Kodi Entertainment Center and I'm constantly updating it to be more accurate/useful. Of course, I linked to this post in the madVR section because you know more than I do.

huhn

18th February 2015, 01:07

first of all this is not a guide this just explains settings.

debanding is done at the source resolution before scaling at least in madVR.

about your "The Video Rendering Process Summarized:" it is not complete.

there are colorspace corrections changing the DVD bt 601 to bt 701 and you don't have to get 4:4:4 frist you can frist scale to the target resolution using 4:2:0 and than scale the chroma. you can even scale both unaffected from each other direct to the file resolution MPDN does it this way.

even if you write up how it works that doesn't mean people will understand what is important.

let's have a look at chroma upsampling.

when chroma is upsampled from a BD source you change the resolution from 960x540 to 1920x1080. this is a huge resize but it's not that important you don't learn that from this. is the TV capable of 4:4:4 and setup properly if yes it is more important compared to a 4:2:2 screen but still not nearly as important as luma scaling.

I don't think it is a good idea to add all these complicated things it will create more questions than answer.

Warner306

18th February 2015, 01:36

Asmodian

18th February 2015, 01:51

Not everything is clear, but I appreciate the effort. You have taught me several things I didn't know.

Please let me know anything that doesn't seem clear! I would be very happy to reword or elaborate as necessary. :)

The Video Rendering Process Summarized:

This is an interesting topic but I agree with huhn that it might be too complex for this post, definitely not as an introduction. I am thinking this post is for people looking at the madVR settings window and wondering what a particular option does.

That said I would like a flow chart describing the rendering steps, perhaps in the more advanced topics section. I am not sure I know enough to build a complete and accurate flow chart but I will give it a shot.

Warner306

18th February 2015, 05:08

Please let me know anything that doesn't seem clear! I would be very happy to reword or elaborate as necessary. :)

This is an interesting topic but I agree with huhn that it might be too complex for this post, definitely not as an introduction. I am thinking this post is for people looking at the madVR settings window and wondering what a particular option does.

That said I would like a flow chart describing the rendering steps, perhaps in the more advanced topics section. I am not sure I know enough to build a complete and accurate flow chart but I will give it a shot.

Regardless of the outcome, I appreciate your efforts to provide some education. It took me a long time to collect all of the information necessary to set-up and understand how to use madVR and its supporting software. The AVS Forum post is a great starting point, but this topic is too large for one post or one guide.

I am thrilled with my set-up and credit an understanding of the software with getting the last 10th of power from my graphics card.

Warner306

18th February 2015, 05:14

Please let me know anything that doesn't seem clear! I would be very happy to reword or elaborate as necessary. :)

I don't have time for a complete readthrough, but the description of NNEDI3 confused me. You mention 2x scaling. Is this describing luma/chroma doubling or how NNEDI3 works in general? By that, I mean does NNEDI3 always double the image even if I choose it as the chroma upscaler over Jinc?

Asmodian

18th February 2015, 06:06

I don't have time for a complete readthrough, but the description of NNEDI3 confused me. You mention 2x scaling. Is this describing luma/chroma doubling or how NNEDI3 works in general? By that, I mean does NNEDI3 always double the image even if I choose it as the chroma upscaler over Jinc?

Yes it does, NNEDI3 can only do 2x scaling. Note: chroma upscaling is always a 2x resize.

edit: Thanks for the feedback, these kinds of questions are great for figuring out what confuses people. :)

toniash

18th February 2015, 08:33

In my case (windows XP and AMD 5450 with 13.12) lowers GPU consumption.

Asmodian

18th February 2015, 10:20

In my case (windows XP and AMD 5450 with 13.12) lowers GPU consumption.

Thanks, noted. :)

Warner306

18th February 2015, 23:50

huhn

19th February 2015, 00:13

Asmodian

19th February 2015, 00:35

I had time for another readthrough. I felt as though the following definitions required further elaboration/clarification:

color&gamma
2.20, 2.40
"changes the display's gamma based on what is set in calibration to this target gamma"

What does "what is set in calibration" mean?

It seems like you do not notice the italics? I used italics to indicate I was referring to a specific name in madVR's settings. All the references to calibration are talking about the calibration page in madVR's settings.

artifact removal
"If default debanding strength is lower than strength during fade in/out 5 frames are rerendered using the higher strength on a fade detection (it takes 5 frames to detect a fade)."

This sentence is very confusing. I understand what you mean (the higher strength is used during fades) but it should be simplified.

I gave it a shot, default debanding strength and strength during fade in/out are the specific names for the settings used in madVR. The information I was adding was the 5 frames, that the higher strength is used during a fade should be obvious based on the setting names. ;)

image doubling
"If this is used with any scaling factor below 2.0x image downscaling is used to interpolate to the target resolution."

A comma after 2.0x would help this sentence, but it is the phrase "interpolate to the target resolution that is unclear". This is technical jargon.

"Interpolate" is not part of everyone's vocabulary? I can change "interpolate" to "scale". Commas are good sometimes.

enable windowed overlay (Windows 7 and newer):
"If not using the calibration option disable GPU gamma ramps madVR will emulate the GPU LUT. "

What does the "calibration option" refer to?

How about reordered, "If not using the option disable GPU gamma ramps in calibration, madVR will emulate the GPU LUT."?

I find these to be the most offensive definitions. Other users may agree/disagree.

Thanks for the feedback.

Warner306

19th February 2015, 01:29

It seems like you do not notice the italics? I used italics to indicate I was referring to a specific name in madVR's settings. All the references to calibration are talking about the calibration page in madVR's settings.

I understand what you mean. I think what you want to say is "This setting is dependent on the display calibration chosen in calibration. Any setting other than disable calibration controls for this display will be used as the baseline to produce the chosen gamma. Otherwise, this setting does nothing."

I gave it a shot, default debanding strength and strength during fade in/out are the specific names for the settings used in madVR. The information I was adding was the 5 frames, that the higher strength is used during a fade should be obvious based on the setting names. ;)

"Interpolate" is not part of everyone's vocabulary? I can change "interpolate" to "scale". Commas are good sometimes.

I would say "If the scaling factor is not an exact duplicate of the original, downscaling to the target resolution is done by interpolation."

How about reordered, "If not using the option disable GPU gamma ramps in calibration, madVR will emulate the GPU LUT."?

That's an improvement.

Asmodian

19th February 2015, 05:44

I think I will stick with the current phrasing, it has improved due to your feedback and I feel that your phrasing is not precise enough. :)

Warner306

19th February 2015, 06:02

I think I will stick with the current phrasing, it has improved due to your feedback and I feel that your phrasing is not precise enough. :)

Sure. In either case, the matter is clarified in this thread. I still wish more threads like this would appear. Much more appealing to aggregate information than having to piece through threads.

huhn

19th February 2015, 12:33

I would say "If the scaling factor is not an exact duplicate of the original, downscaling of the target resolution is done by interpolation."

but it may need upscaling haver doubling too.

Warner306

20th February 2015, 00:47

but it may need upscaling haver doubling too.

Thanks, I enjoy how many here are allergic to English.

namaiki

20th February 2015, 01:43

DXVA2: Uses the GPU drivers or fixed function hardware to perform scaling. It is done in 8-bit so it may introduce banding or cause scaled dither noise. It may use separate scaling hardware so it can be very fast and low power. It is good for lower end GPUs or mobile, especially Intel embedded GPUs. Quality varies between GPU brands and models. Image doubling cannot be used at the same time as DXVA2 scaling. Does chroma upscaling as well so this option overrides chroma upscaling. May require a restart of madVR to take effect.

Just to say that my laptop with Intel HD4000 graphics can use madVR default scaling settings even with smooth motion enabled. DXVA2 scaling also enables Intel's video enhancements (enabled by default, at least on my laptop) which include aggressive sharpening and dynamic contrast which produces terrible banding in some situations.

Asmodian

20th February 2015, 04:28

Thanks, noted. :)

Asmodian

20th February 2015, 05:41

What errors have I made in this flow chart?
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Flow%20Chart_zpsujwiegkf.png

I know it is over simplified, any details you want added?

Thanks! :)

huhn

20th February 2015, 23:50

are you sure debanding is done at RGB not at the untouched input frame?

Asmodian

21st February 2015, 01:44

Based on everything I can find about f3kdb I would say you are probably right, that makes DXVA2 scaling make more sense too.

But then, when using image doubling, it converts to RGB and then immediately converts to YCbCr? I suppose that would make sense, especially given that there is probably more going on between them that is not on my chart. :p

Arm3nian

21st February 2015, 02:38

Can you expand on image upscaling?

Let's say watching a 1080 video on a 1440 screen would look messy on youtube or maybe a mobile device, but it would look better with the advanced codecs in madvr?

Also, for watching a 1920x1080 video on a 3840x2160 screen, 4x the res, would using the other codecs do anything compared to nearest neighbor? Wouldn't just adding the same pixels horizontally and vertically be all that is needed.

Warner306

21st February 2015, 02:59

What errors have I made in this flow chart?
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Flow%20Chart_zpskpknmc8l.png

Where is deinterlacing, can I simply add it to DXVA2 scaling? Does deinterlacing return RGB or YCbCr?

I know it is over simplified, any details you want added?

Thanks! :)

Where are you planning on posting this? It is a good first point of contact before trying to talk about madVR.

I think I'm a layman and it's not oversimplified to me. Each bubble is an entry point to a discussion.

Asmodian

21st February 2015, 04:15

I have updated the chart.

Can you expand on image upscaling?

Let's say watching a 1080 video on a 1440 screen would look messy on youtube or maybe a mobile device, but it would look better with the advanced codecs in madvr?

I do not want to recommend particular resize methods or go into the subjective differences between them. The bar graphs in madVR already do a good job of that. The reason it might look messy on a mobile device is only because it uses bilinear, nearest neighbor, or similar which are also options in madVR.

Resizing methods are not codecs. Codec is short for "coder decoder" and usually referrers to software which codes images, video, or audio in a different format like JPEG, H.264, or MP3 and can then decode it back to the original format.

Also, for watching a 1920x1080 video on a 3840x2160 screen, 4x the res, would using the other codecs do anything compared to nearest neighbor? Wouldn't just adding the same pixels horizontally and vertically be all that is needed.

No. Imagine you start with an image at 3840x2160. You scale it down to 1920x1080. Now you want to upscale it so it looks as much like the original 3840x2160 image as it can. Nearest neighbor does not get very close. You want it to look as if it was the same content but at 2x the resolution, you do not want to represent the lower resolution source perfectly.

Also it is 2x the resolution, not 4x. 4x would be 8K or 7680x4320. A scaling factor of 1.414213562393095 is not a 2x upscale.

Image doubling generates 4 times the number of pixels. Image quadrupling generates 16 times the number of pixels.

Where are you planning on posting this? It is a good first point of contact before trying to talk about madVR.

I think I'm a layman and it's not oversimplified to me. Each bubble is an entry point to a discussion.

I want to collect feedback on it for a bit before adding it to the end of the first post.

sheppaul

21st February 2015, 06:38

What errors have I made in this flow chart?
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Flow%20Chart_zpsrf3trlzj.png

But then, when using image doubling, it converts to RGB and then immediately converts to YCbCr?

It looks like nonsense to me. If image doubling is performed with YCbCr 4:4:4, the RGB conversion should not be occurred directly before image doubling. So the arrow might be from chroma upsampling to image doubling IMO. Your flow chart really helps to understand a glimpse of rendering stages.

:thanks:

ps. I guess deinterlacing is likely to be placed at the very first stage of the flow chart. It's hard to think deinterlacing after image processing: Lav - deinterlacing - deband. Just my guess :p

Asmodian

21st February 2015, 07:49

It looks like nonsense to me. If image doubling is performed with YCbCr 4:4:4, the RGB conversion should not be occurred directly before image doubling. So the arrow might be from chroma upsampling to image doubling IMO. Your flow chart really helps to understand a glimpse of rendering stages.

I know that image doubling will always do a conversion to YCbCr even if doubling both luma and chroma with the same number of neurons. I don't think madVR will skip the RGB conversion before scaling if doing image doubling but I could be wrong.
edit: I think this step also does all the needed color space conversions (e.g. BT.601 to RGB or BT.709 to RGB) while only one matrix is used in image doubling.

ps. I guess deinterlacing is likely to be placed at the very first stage of the flow chart. It's hard to think deinterlacing after image processing: Lav - deinterlacing - deband. Just my guess :p

I would assume this as well but does the video come out of deinterlacing as RGB or YCbCr 4:4:4? Maybe I will add it there and assume it returns as 4:2:0. :)
edit: after testing it is obvious chroma scaling is done after deinterlacing so I think it is safe to say the chroma is returned as 4:2:2 or 4:2:0. I think it might be 4:2:2 actually

edit: New chart with deinterlacing. Please let me know if I need to add / correct anything. :)
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Flow%20Chart_zpsziuu7mhl.png

sheppaul

21st February 2015, 09:11

Maybe I will add it there and assume it returns as 4:2:0. :)

http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Flow%20Chart_zpsziuu7mhl.png

I agree but we cannot confirm. :rolleyes:

Asmodian

21st February 2015, 10:52

I agree but we cannot confirm. :rolleyes:

I did a little testing and chroma scaling still has to be done after (or before?) deinterlacing. I also know deinterlacing uses DXVA for video mode and it would make sense for it to accept 4:2:0 as that is the most common. Doing chroma scaling before deinterlacing doesn't make sense. MadVR will not deinterlace RGB sources.

But to be honest I do not feel all that confident about the placement of deinterlacing and debanding. :p

nevcairiel

21st February 2015, 11:15

Deinterlacing has to be before chroma scaling, scaling interlaced chroma would be extra complexity and not make much sense, especially considering that GPUs prefer 4:2:0 in form of NV12 anyway.

Arm3nian

21st February 2015, 13:00

I have updated the chart.

I do not want to recommend particular resize methods or go into the subjective differences between them. The bar graphs in madVR already do a good job of that. The reason it might look messy on a mobile device is only because it uses bilinear, nearest neighbor, or similar which are also options in madVR.

Resizing methods are not codecs. Codec is short for "coder decoder" and usually referrers to software which codes images, video, or audio in a different format like JPEG, H.264, or MP3 and can then decode it back to the original format.

No. Imagine you start with an image at 3840x2160. You scale it down to 1920x1080. Now you want to upscale it so it looks as much like the original 3840x2160 image as it can. Nearest neighbor does not get very close. You want it to look as if it was the same content but at 2x the resolution, you do not want to represent the lower resolution source perfectly.

Also it is 2x the resolution, not 4x. 4x would be 8K or 7680x4320. A scaling factor of 1.414213562393095 is not a 2x upscale.

Image doubling generates 4 times the number of pixels. Image quadrupling generates 16 times the number of pixels.

I want to collect feedback on it for a bit before adding it to the end of the first post.

Yes, algorithm would be the correct term to use. However, your grammar lesson regarding resolution is incorrect. 4x the resolution of 1920x1080 is 3840x2160. Resolution implies the total number of the pixels. 7680x4320 is 4x horizontally and vertically, making it 8 times the resolution. It makes no sense to refer to resolution in only one dimension, there is 4x the amount of information, making it 4x the amount of resolution. Yes image doubling works in both directions, as Nev has stated to me in the past. Maybe madshi can explain why he calls it image doubling, I'm sure he has his reasons. Probably because to properly double an image, you would have to change the horizontal and vertical together, it would make no sense to say I doubled a 1920x1080 image to 3840x1080 even though the amount of pixels is 2x.

Anyway, a 1080p video looks quite good on my 1440p nexus 6, even though the renderer most likely uses crap interpolation. The 493ppi density hides artificats, not the same as a desktop monitor.

Shiandow

21st February 2015, 16:10

Maybe madshi can explain why he calls it image doubling, I'm sure he has his reasons. Probably because to properly double an image, you would have to change the horizontal and vertical together, it would make no sense to say I doubled a 1920x1080 image to 3840x1080 even though the amount of pixels is 2x.

To 'properly' double an image in that sense is impossible since √2 is irrational. Anyway, in almost all cases it is easier to talk about the scaling factor instead of the ratio of pixels. This also corresponds more closely to the technical meaning of resolution which is measured in pixels per inch.

huhn

21st February 2015, 16:19

I did a little testing and chroma scaling still has to be done after (or before?) deinterlacing. I also know deinterlacing uses DXVA for video mode and it would make sense for it to accept 4:2:0 as that is the most common. Doing chroma scaling before deinterlacing doesn't make sense. MadVR will not deinterlace RGB sources.

But to be honest I do not feel all that confident about the placement of deinterlacing and debanding. :p

debanding works best before scaling same goes for chroma scaling so it makes only sense that it is done before chroma scaling.

you could add IVTC to deinterlacing. and you my add an comment that force film mode doesn't work with native DXVA

sheppaul

21st February 2015, 16:38

Maybe madshi can explain why he calls it image doubling, I'm sure he has his reasons.

It's from intrinsic properties of Nnedi3 filter. The filter is designed for that purpose.

http://avisynth.nl/index.php/Nnedi3
nnedi3 is also very good for enlarging images by powers of 2,

Arm3nian

21st February 2015, 19:32

To 'properly' double an image in that sense is impossible since √2 is irrational. Anyway, in almost all cases it is easier to talk about the scaling factor instead of the ratio of pixels. This also corresponds more closely to the technical meaning of resolution which is measured in pixels per inch.
Ppi is pixel density. The definition of resolution is the amount of information/detail in an image. If one image has 4x the information as another, it is 4x the resolution. Why break up the relation between the horizontal and vertical and call it 2x the resolution, when doubling in both directions implies you get 4x the original. Makes no sense.

Also in games for example, 4xSSAA is 2x in the horizontal and 2x in the vertical. But it is still called 4x, because the result is 4x greater.
It's from intrinsic properties of Nnedi3 filter. The filter is designed for that purpose.
Well power of 2 can mean anything. Maybe 2^2 (4x) is reffered to as doubling and 2^4 (16x) is quadrupling, based on the amount of pixels they give.

iSunrise

21st February 2015, 20:42

No need to make this so complicated.

"Image doubling" doubles horizontal as well as vertical (width and height) pixels, while quadrubling is basically 2x "Image doubling" applied. And since this is directly connected to the way NNEDI3 works, you can 2x2x2x2.... as much as you would like with subsequent "doubles" (at least theoretically). But then again, even most of the current GPUs struggle with doubling (with at least NNEDI3 32 neurons). So, basically "to be continued"?

madshi always uses the "makes sense, why make it anymore complicated than it already is" approach, there is no rocket-science behind this, even though madshi usually comes up with pretty great ideas and surprises us.

Warner306

21st February 2015, 22:18

Your chart appears logical to me. But I also question why the output would be converted to RGB, immediately converted back to YCbCr only to be converted back to RGB once again when image doubling. Are you sure this is correct?

huhn

21st February 2015, 23:05

you should ask madshi. but he clearly said it is converted back to YCbCr using BT 709.

vivan

21st February 2015, 23:34

I remember reading about it too. Probably keeping rgb between shaders is easier (from development point of view), even though it's not effective.

Also in games for example, 4xSSAA is 2x in the horizontal and 2x in the vertical. But it is still called 4x, because the result is 4x greater.And some people call 4-tap filters (where 4 is radius) 8-tap (diameter) or even 64-tap (total pixels sampled). Cause bigger numbers are better :p

NNEDI is a 1-dimensional filter that doubles the resoulution.
In case on anamorphic video it's possible that it will run it in only one direction (e.g. 16:9 1440x1080 with doubling set to always). In other cases it runs it twice (1 in each direction).
So you should think not about resolution, but about directions. It's just that keeping separate settings for 2 directions doesn't make much sence, so they're combined.

About debanding.
RGB: if you check shader code, you'll see that it converts from rgb to ycbcr and then from ycbcr to rgb.

Subsampling: I believe that DX9 doesn't support writing into 16-bit subsampled textures, so... Emulating it with 2 textures is not only painful, but also will make it slower (+25% pixels to write, doesn't matter that they have less channels).
Actually I don't believe that it even supports subsampled 10/16-bit textures (this is probably the reason why we won't see 10-bit native DXVA in madVR).
So doing it after chroma upsampling is hell lot easier and doesn't impact performance.

Asmodian

21st February 2015, 23:57

Deinterlacing has to be before chroma scaling, scaling interlaced chroma would be extra complexity and not make much sense, especially considering that GPUs prefer 4:2:0 in form of NV12 anyway.

Thanks, that makes me feel more confident in the placement of deinterlacing. :)

Ppi is pixel density. The definition of resolution is the amount of information/detail in an image. If one image has 4x the information as another, it is 4x the resolution. Why break up the relation between the horizontal and vertical and call it 2x the resolution, when doubling in both directions implies you get 4x the original. Makes no sense.

Because we are talking about spacial resolution. If an image has 4x the information as another, it is 2x the resolution. This is based on the definition of resolution used by everyone except digital camera makers who want to be able to say they quadrupled the resolution when they really only doubled it.

Using your nomenclature 2x the resolution implies an irrational scaling factor of square root of 2 or only doubling one dimension. It sounds great for people trying to sell new models of digital cameras but no one else talks about resolution this way.

I am describing a standard so everyone uses the same words to mean the same thing. madshi obviously agrees given image doubling and as this thread is about madVR options let us stick to the terminology madVR uses. Everyone else who works with digital image processing thinks madshi got it right so this should not cause confusion. When you scale an image in Photoshop to "200%" you get 4 times the number of pixels.

Also in games for example, 4xSSAA is 2x in the horizontal and 2x in the vertical. But it is still called 4x, because the result is 4x greater.

No it isn't, that is 2x SSAA. 4x SSAA uses 16 times the pixels. 2x SSAA does not only filter the vertical or horizontal dimension and you cannot have square root of two pixel sampling.

I have not been able to find a statement from AMD or Nvidia but this is representative of what I was able to find:
SSAA, or Super-Sample Anti-Aliasing is a brute force method of anti-aliasing. It results in the best image quality but comes at a tremendous resource cost. SSAA works by rendering the scene at a higher resolution, 2x SSAA renders the scene at twice the resolution along each axis (4x the pixel count), 4x SSAA renders the scene at four times the resolution along each axis (16x the pixel) count, and 8x SSAA renders the scene at eight times the resolution along each axis (64x the pixel count). The final image is produced by downsampling the massive source image using an averaging filter. This acts as a low pass filter which removes the high frequency components that caused the jaggedness.

Your chart appears logical to me. But I also question why the output would be converted to RGB, immediately converted back to YCbCr only to be converted back to RGB once again when image doubling. Are you sure this is correct?

I am fairly sure it is correct.

you should ask madshi. but he clearly said it is converted back to YCbCr using BT 709.

It was looking for that quote but I couldn't find it. This is also relevant and changes image doubling in my chart. :p

And to make sure I'm no longer confused, what is the result of the following?

640x360 4:2:0 -> 1920x1080
Chroma Upscaling = NNEDI3
NNEDI3 double Luma = Enabled
NNEDI3 double Chroma = Enabled
NNEDI3 quadruple Luma = Enabled
NNEDI3 quadruple Chroma = Disabled

Conversion from 4:2:0 YcbCr (320x180) to 4:4:4 RGB (640x360) with 'chroma upscaling' setting NNEDI3

Conversion from 640x360 4:4:4 RGB -> 640x360 4:4:4 YCbCr.

Y & CbCr channels doubled to 1280x720 4:4:4 YCbCr with NNEDI3

Y channel only doubled to 2560x1440

Is it then:

CbCr channels upscaled 1280x720->1920x1080 with 'image upscaling' setting

Y channel downscaled 2560x1440->1920x1080 with 'image downscaling' setting

Conversion from 4:4:4 YCbCr to RGB

Or is it:

CbCr channels upscaled 1280x720->2560x1440 with 'image upscaling' setting

Conversion from 4:4:4 YCbCr to RGB

2560x1440 RGB -> 1920x1080 RGB with 'image downscaling' setting

It is the first ("CbCr channels upscaled 1280x720->1920x1080 with 'image upscaling' setting").

About debanding.
RGB: if you check shader code, you'll see that it converts from rgb to ycbcr and then from ycbcr to rgb.

Subsampling: I believe that DX9 doesn't support writing into 16-bit subsamped textures, so... Emulating it with 2 textures is not only painful, but also will make it slower (+25% pixels to write, doesn't matter that they have less channels).
Actually I don't believe that it even supports subsampled 10/16-bit textures (this is probably the reason why we won't see 10-bit native DXVA in madVR).
So doing it after chroma upsampling is hell lot easier and doesn't impact performance.

Ok thanks again, that also makes sense. Debanding happens after conversion to RGB as well since it is converting from RGB.

Arm3nian

22nd February 2015, 01:44

Because we are talking about spacial resolution. If an image has 4x the information as another, it is 2x the resolution. This is based on the definition of resolution used by everyone except digital camera makers who want to be able to say they quadrupled the resolution when they really only doubled it.

Using your nomenclature 2x the resolution implies an irrational scaling factor of square root of 2 or only doubling one dimension. It sounds great for people trying to sell new models of digital cameras but no one else talks about resolution this way.

I am describing a standard so everyone uses the same words to mean the same thing. madshi obviously agrees given image doubling and as this thread is about madVR options let us stick to the terminology madVR uses. Everyone else who works with digital image processing thinks madshi got it right so this should not cause confusion. When you scale an image in Photoshop to "200%" you get 4 times the number of pixels.

No it isn't, that is 2x SSAA. 4x SSAA uses 16 times the pixels. 2x SSAA does not only filter the vertical or horizontal dimension and you cannot have square root of two pixel sampling.

I have not been able to find a statement from AMD or Nvidia but this is representative of what I was able to find:

Replace resolution with 'information'. If something has 4x the information of another thing, saying it has 2x instead makes no sense. If you want to say 2x the resolution, then you would have to independantly define and state that resolution x is 2 times greater horizontally and 2 times greater vertically than resolution y.

You are correct that we are talking about spacial resolution. If an image is 3840x2160, but its spacial resolution is lower, meaning the pixels are much more spread out, it might look the same as the same image at 1920x1080 at the correct distance. But, the 3840x2160 image still has more information, it just displayed incorrectly/differently.

Photoshop refers to 200% as 4x the pixels because it scales the horizontal and vertical together. Just like nnedi3. If it only scaled in one dimension, what would you call it? What is 3840x1080 compared to 1920x1080... using your defintion? You can't describe it. But you can say it is 2x the resolution, and 3840x2160 is 4x the resolution. It makes sense to refer to "scaling" as 2x, because you scale both directions together. But it does not make sense to refer to the entire resolution as 2x.

In triple A games, 4xSSAA is 4x the resolution. If 4xSSAA was 16x the pixels, it wouldn't exist in any games, you would need 4 gtx 980s running at 2200mhz cooled by ln2, obviously not realistic to implement.

Also, check the dynamic super resolution in your nvidia control panel. I am currently on a 1920x1200 monitor, and 4x the resolution is referred to as 3840x2400. Same multiplier scale is used for all the other dynamic super resolutions. 2x the resolution is 2715x1697.
Picture here: http://i.imgur.com/ID3XoBc.png

So you should think not about resolution, but about directions. It's just that keeping separate settings for 2 directions doesn't make much sence, so they're combined.

Questionable. Nvidia inspector, as shown in my next post, allows you to configure the horizontal and vertical independantly of each other. This can provide better image quality without destroying your frame rate, as the picture rendered at the higher resolution is downsampled. Obviously this might not work properly in some games, or madvr for that matter, which is why it isn't an option in the applications itself, but rather the drivers. A 1920x1080 image scaled by 4x, is now 400% larger. Photoshop refers to it as 200% because Photoshop is an image editing tool. You can select to only the scale the horizontal or vertical by 200%, doesn't mean the amount of pixels increased fourfold. The 200% that results in 4x the pixels just means you increased the resolution by 2 in both directions.

Arm3nian

22nd February 2015, 02:07

Also check this: http://i.imgur.com/7VaN0Tc.png

Notice how SSAA is referred in both the horizontal and vertical resolution. 2*2=4, so 2x2SSAA is shortened as 4xSSAA. This can be confirmed by looking below at the combined SSAA and MSAA options. 32xS is 2x2SSAA + 8xMSAA, 2*2=4, 4*8=32. This is the most logical method of labeling the values.

Asmodian

22nd February 2015, 02:39

Replace resolution with 'information'. If something has 4x the information of another thing, saying it has 2x instead makes no sense. If you want to say 2x the resolution, then you would have to independantly define and state that resolution x is 2 times greater horizontally and 2 times greater vertically than resolution y.

This is exactly my point. The sort hand "2x the resolution" means 2 times greater horizontally and 2 times greater vertically. You cannot replace resolution with information; they are not synonymous.

To be precise "2x the resolution" is ambiguous, it could have either interpretation. It would be more precise to say "2x the resolution horizontally and vertically" or "twice the number of pixels".

There are reasonable reasons to talk about resolution as the total number of pixels but that is not the convention normally used when discussing digital images and video (unless you are a marketing department).

2x scaling gives 2x the resolution (Nvidia's DSR notwithstanding, they only want to describe the performance hit and do not care about conventions).

Anyway I am happy to agree to disagree, only know if I say "2x the resolution" I mean 4 times the number of pixels.

Arm3nian

22nd February 2015, 03:33

This is exactly my point. The sort hand "2x the resolution" means 2 times greater horizontally and 2 times greater vertically. You cannot replace resolution with information; they are not synonymous.

To be precise "2x the resolution" is ambiguous, it could have either interpretation. It would be more precise to say "2x the resolution horizontally and vertically" or "twice the number of pixels".

There are reasonable reasons to talk about resolution as the total number of pixels but that is not the convention normally used when discussing digital images and video (unless you are a marketing department).

2x scaling gives 2x the resolution (Nvidia's DSR notwithstanding, they only want to describe the performance hit and do not care about conventions).

Anyway I am happy to agree to disagree, only know if I say "2x the resolution" I mean 4 times the number of pixels.

I'm just going off the fact that the term resolution was created to describe information. You can say that 3840x2160 is 2x the resolution of 1920x1080, but what does that mean? It is a useless definition if resolution does not imply information. Maybe it does make sense when strictly talking about photography or video, but that is because we already know doubling the resolution in both directions leads to 4x the pixels, and are just trying to talk about the size. Resolution on its own is a meaningless value if not talking about information. In engineering, the resolution bandwidth of a spectrum analyzer for example implies more samples, therefore shows more information.

Asmodian

22nd February 2015, 11:04

I'm just going off the fact that the term resolution was created to describe information. You can say that 3840x2160 is 2x the resolution of 1920x1080, but what does that mean?

It means there are twice as many pixels, both horizontally and vertically; you can resolve details that are half the size. :p

It is a useless definition if resolution does not imply information. Maybe it does make sense when strictly talking about photography or video, but that is because we already know doubling the resolution in both directions leads to 4x the pixels, and are just trying to talk about the size. Resolution on its own is a meaningless value if not talking about information. In engineering, the resolution bandwidth of a spectrum analyzer for example implies more samples, therefore shows more information.

When analyzing information the purpose of using the term "resolution" instead of "samples" is to signify the relative spacing of samples in a data set. In a 1D set twice the resolution is twice the samples, in a 2D set twice the resolution is 4 times the samples, in a 3D set it is 8 times the samples, 4D 16, 5D 32, etc. If you want to discuss the total number of samples instead use "samples", "pixels", or similar.

If you have twice the resolution you expect to be able to resolve details at half the size. For a 2D space you need four times the number of pixels to resolve details at half the size. e.g. If the smallest filament you can resolve with an imaging system is 0.1 mm in diameter and you need to be able to resolve one with a diameter of 0.05 mm you need a sensor with double the resolution. It is not useful to raise the change in resolution required to the power of the number of possible axes when discussing a needed change in sampling. It simply removes meaning. Saying you need 8 times the resolution to resolve a 3D object at half the size is pointless, why not say you need 8 times the samples if that is how you want to express it?

Resolution is a term used to describe data sets which are both a type of information themselves and contain information. The relative change in resolution equals the relative change in the number of samples only when talking about a data set with one dimension.

At this point I think I have expressed my option on the matter as well as I can. :o

Arm3nian

22nd February 2015, 11:30

It means there are twice as many pixels, both horizontally and vertically; you can resolve details that are half the size. :p

When analyzing information the purpose of using the term "resolution" instead of "samples" is to signify the relative spacing of samples in a data set. In a 1D set twice the resolution is twice the samples, in a 2D set twice the resolution is 4 times the samples, in a 3D set it is 8 times the samples, 4D 16, 5D 32, etc. If you want to discuss the total number of samples instead use "samples", "pixels", or similar.

If you have twice the resolution you expect to be able to resolve details at half the size. For a 2D space you need four times the number of pixels to resolve details at half the size. e.g. If the smallest filament you can resolve with an imaging system is 0.1 mm in diameter and you need to be able to resolve one with a diameter of 0.05 mm you need a sensor with double the resolution. It is not useful to raise the change in resolution required to the power of the number of possible axes when discussing a needed change in sampling. It simply removes meaning. Saying you need 8 times the resolution to resolve a 3D object at half the size is pointless, why not say you need 8 times the samples if that is how you want to express it?

Resolution is a term used to describe data sets which are both a type of information themselves and contain information. The relative change in resolution equals the relative change in the number of samples only when talking about a data set with one dimension.

At this point I think I have expressed my option on the matter as well as I can. :o

I agree on all points. I think the problem we have is that 'resolution' on its own is a very ambiguous term. Are we talking about image resolution, pixel resolution, spatial resolution, or even resolution that has nothing to do with digital pixels like my spectrum analyzer example.

Nvidia most likely uses 4x in the DSR to show performance impacts, as the gpu will render 4x the amount of pixels. For photography and video I see why 2x would make sense. 4x when talking about scaling might give the wrong impretion on expected quality. The size of the image is important to consider. It would be to the benefit of all if everyone was more descriptive, but as you already know, it is hard to get an entire industry to agree on something.

newguy1

23rd February 2015, 17:20

Isn't the whole 2x or 4x resolution an industry standard that the 4k tv decided to overlook?

You know, a standard? Like from a textbook?

Warner306

24th February 2015, 03:02

Isn't the whole 2x or 4x resolution an industry standard that the 4k tv decided to overlook?

You know, a standard? Like from a textbook?

The marketing always seems to skew towards selling more televisions as opposed to providing technical information, but I don't know why this simple topic has to cover so many consecutive posts.

A 720p source has 2.25x the number of pixels of a 1080p source. However, the actual increase in vertical resolution or (pixels per inch) is only 1.5x. The amount of dots per inch is a better indication of quality than the total number of pixels.

Moving from 1080p to 4K, the number of pixels increases to 4x, but the scaling factor only improves to a mere 2x.

A better question is whether anyone can resolve the extra pixels per inch from their seating distance, which they probably cannot.

Arm3nian

24th February 2015, 05:36

The marketing always seems to skew towards selling more televisions as opposed to providing technical information, but I don't know why this simple topic has to cover so many consecutive posts.

A 720p source has 2.25x the number of pixels of a 1080p source. However, the actual increase in vertical resolution or (pixels per inch) is only 1.5x. The amount of dots per inch is a better indication of quality than the total number of pixels.

Moving from 1080p to 4K, the number of pixels increases to 4x, but the scaling factor only improves to a mere 2x.

A better question is whether anyone can resolve the extra pixels per inch from their seating distance, which they probably cannot.

Everyone in this thread understands 4x vs 2x. We were just debating on the correct terminology. In my first post I mentioned 1920x1080 is 4x the resolution of 3840x2160. Saying 2x horizontal and 2x vertical resolution is the same. 2x scaling just makes it confusing. 3840x2160 is 4x more demanding than 1920x1080, so we should refer to it as 4x unless otherwise specified that you are talking about the pixel count in 1 direction.

Think about a water bottle. If I have a water bottle that is 1ft tall, and I increase it to 2ft, I now have a water bottle that is 2x bigger, because it holds 2x the water. If I increase it to 4ft, then I have a watter bottle that holds 4x as much as water as the initial bottle, making it 4x bigger. Now lets say I invent a gun that makes objects 2x bigger and I shoot my water bottle. Is it going to make it only 2x taller? No, it is going to multiply ALL dimensions by a factor of 2. Now the water bottle does not hold 2x as much water, it holds 8x as much water. So in theory, increasing something by a factor of 2 is relative on the amount of dimensions. In 1d, a factor of 2 increases the size by 2, in 2d, a factor of 2 increases the size by 4, and in 3d, a factor of 2 increases the size by 8. In other words, it more precise to refer to these things by what matters. In resolution, the pixel count, which directly relates to the performance. And in the case of the water bottle, to the volume, which directly reflects the amount of water it can hold. Making something 2x bigger just implies that the actual value did not increase linearly, so why not just refer to the actual value directly... makes it more simple, and more logical.

Resolution is just another word for pixel count. A resolution of 2 means 2 pixels. A resolution of 3840x2160 means 8294400 pixels. The image is just the object. The resolution is the quantity being defined. Just like a water bottle is the object and the volume is being defined. This is why madshi refers to it as image doubling, because making an image 2x bigger creates 4x the pixels. Saying the resolution in 1920x1080 to 3840x2160 doubled makes no sense, because it quadrupled. The image itself doubled.

4k might provide an increase in perceived image quality on a TV, but it is more noticeable on a monitor to me since I sit close.

resides

25th February 2015, 05:40

Thanks for the write up.

Definitely helped me and clarified things too.

Asmodian

26th February 2015, 03:52

you could add IVTC to deinterlacing. and you my add an comment that force film mode doesn't work with native DXVA

I never use DXVA decoding but I feel that adding notes on its ramifications where appropriate could be helpful. Are there any other details or issues you know of with DXVA native decoding? Thanks.

Thanks for the write up.

Definitely helped me and clarified things too.

Thanks, I am glad it helped a bit. :)

huhn

26th February 2015, 04:47

I never use DXVA decoding but I feel that adding notes on its ramifications where appropriate could be helpful. Are there any other details or issues you know of with DXVA native decoding? Thanks.

nothing limited to madVR. just the general. impossible to use the xy/vsfilter, some renderer can't display subtitle. nothing that should be part of this in my eyes. just basic stuff like this.

maybe this:
madVR can't accept p010 from DXVA native. i'm not even sure if there is a stable software that is trying to do this.

MSL_DK

28th February 2015, 10:41

image up or downscaling isn't used when displaying 1920x1080 content on a 1080p display, correct? I'm asking because i've read that upscaling is used on everything, perhaps i've misunderstood that article.

Qaq

28th February 2015, 11:04

Its chroma upsampling that is used always, cause 4:2:0 needs to be upsampled to 4:4:4.
1080@1080 doesn't need any image upscaling, but keep in mind that some *1080* videos are not exactly 1080 (1077 or so).

Asmodian

8th March 2015, 23:00

I did some minor updates and corrections and I expanded the calibration section some. yCMS deserves a mention at least.

maybe this:
madVR can't accept p010 from DXVA native. i'm not even sure if there is a stable software that is trying to do this.

This is good to know but I am not sure where to put it. Maybe simply as a note near the end?

sheppaul

9th March 2015, 09:17

I did some minor updates and corrections and I expanded the calibration section some. yCMS deserves a mention at least.

This is good to know but I am not sure where to put it. Maybe simply as a note near the end?

What about creating the section like "known problems" or "limitations"?

huhn

9th March 2015, 15:26

This is good to know but I am not sure where to put it. Maybe simply as a note near the end?

this is only affecting the gtx 960 at the moment no other card can hardware decode 10 bit sources at least AFAIK.

and it was not 100% correct from me to say it can accept it. it simply shows a green/black screen when native DXVA feeding P010 in madVR.

some other limitations i know are:
madVR IVTC can detect a lot of cadences but can "only" handle a couple of them correctly.
it can handle at least 3:2, 5:5 and 8:7 from a 60i source. very rare cadences like 3:2:3:2:2 (25 fps in 30i 60 fields) 4:2:2:2 doesn't work

madVR currently always switches to 23Hz when having forced film mode on while playing 59i content. Yeah, with some cadences 23Hz is not the optimal refresh rate. But that's not so easy to solve, especially because cadence detection can be unstable. E.g. what happens if cadence detection switches between 3:2 and between 3:2:3:2:2 all the time? Should madVR then always switch between 23Hz and 25Hz all the time, in the middle of playback? This is a problem I'll have to revisit later, but I don't consider it a "bug" right now. It's as intended, even it might not be the best (or even the correct) solution.
source: http://bugs.madshi.net/view.php?id=150

the decimation feature is currently not working the way it should be. see here: http://bugs.madshi.net/view.php?id=199&PHPSESSID=af06a17857165747bc51e1f417cb10e6

madVR can't do frame packed 3D

and this from the madVR thread
known problems / limitations:

- rotated displays (portrait vs landscape orientation) don't work yet

michkrol

14th March 2015, 11:59

First of all, great guide. It must have taken a lot of work to write such extensive explanations.

You've asked for suggestions, so here are mine:
File name tags:
madVR will look at the file name for "magic words" or tokens that tell it to treat a particular file differently.
Technical nuance: madVR looks for tags in the whole path, so both folder(s) and file names.
Pro Tip: put tags in folder(s) name(s) if you have a whole collection of broken rips (files), to save a lot of work (time).
You can even put tags in folders' names on different level for maximum flexibility, for example to get a monster like this:
D:\Movies\TV Rips levels=tv\Broken PAL Rips matrix=PAL\my broken rip from the 90s.mkv
Possible file name tags include:
(...)
What you've listed is of course correct, but you're missing a few tags and possible values.
Tags with missing values or missing altogether (all taken directly from changelog.txt included with madVR):

* added tag "matrix=2020|709|601|NTSC|PAL|YCgCo|240M"
* added tag "primaries=2020|DCI|709|SmpteC|EBU|sRGB|NTSC|PAL|470M|240M|170M"
* added tag "levels=PC|TV|fullrange|limited|doubleExp|tripleExp"
* added tag "blacklevel=%value%", value range [-50, +50]
* added tag "whitelevel=%value%", value range [-50, +50]
* added tag "contrast=%value%", value range [-100, +100]
* added tag "brightness=%value%", value range [-100, +100]
* added tag "saturation=%value%", value range [-100, +100]
* added tag "hue=%value%", value range [-180, +180]
* added tag "frameRate=%value%", e.g. 23.976, 24.000, 23, 24, ...
As a side note you can use either range=... or levels=... tag for the same effect.

Thanks for the guide and keep up the great work :thanks:

Asmodian

18th March 2015, 05:29

What about creating the section like "known problems" or "limitations"?

I agree, we have gathered enough items for this section. :)

this is only affecting the gtx 960 at the moment no other card can hardware decode 10 bit sources at least AFAIK.

and it was not 100% correct from me to say it can accept it. it simply shows a green/black screen when native DXVA feeding P010 in madVR.

some other limitations i know are:
madVR IVTC can detect a lot of cadences but can "only" handle a couple of them correctly.
it can handle at least 3:2, 5:5 and 8:7 from a 60i source. very rare cadences like 3:2:3:2:2 (25 fps in 30i 60 fields) 4:2:2:2 doesn't work

source: http://bugs.madshi.net/view.php?id=150

the decimation feature is currently not working the way it should be. see here: http://bugs.madshi.net/view.php?id=199&PHPSESSID=af06a17857165747bc51e1f417cb10e6

madVR can't do frame packed 3D

Good info, Thanks!
These kinds of things could take a long time to debug on your own. :p

First of all, great guide. It must have taken a lot of work to write such extensive explanations.

You've asked for suggestions, so here are mine:

Technical nuance: madVR looks for tags in the whole path, so both folder(s) and file names.
Pro Tip: put tags in folder(s) name(s) if you have a whole collection of broken rips (files), to save a lot of work (time).
You can even put tags in folders' names on different level for maximum flexibility, for example to get a monster like this:
D:\Movies\TV Rips levels=tv\Broken PAL Rips matrix=PAL\my broken rip from the 90s.mkv

What you've listed is of course correct, but you're missing a few tags and possible values.
Tags with missing values or missing altogether (all taken directly from changelog.txt included with madVR):

As a side note you can use either range=... or levels=... tag for the same effect.

Thanks for the guide and keep up the great work :thanks:

Great info! Thanks a lot. Good pro tip, I hadn't thought of using them in folder names. :o

huhn

18th March 2015, 10:34

Asmodian

18th March 2015, 18:42

here is my source for the DXVA native P010 problem:
http://forum.doom9.org/showpost.php?p=1710491&postcount=305
it is a green screen. i guess this information is crucial because i can't reproduce this problem on my own. i don't have a hardware decoder for 10 bit HEVC to output p010.

and it looks like you skipped the decimation feature from madVR created for 720p60/720p50 broadcast with 24p/25p film content.
madVR IVTC can decimate progressive streams. just press control + alt + shift+ t to force it on. it has the same limitation like madVR IVTC. in the end it's the same program working.

here the part out of the change log. it is one of the newest feature.
* added support for decimating 50p/60p movies to 25p/24p

Thanks again, I had missed that decimation feature. I seem to have hit the limit for characters per post so I was unable to add the section. I may edit for reduced character count. :(

Other Features:
-madVR IVTC can decimate progressive streams such as 720p60/720p50 broadcast with 24p/25p content. Press control + alt + shift+ t to force it on. It has the same limitations as madVR IVTC.

edit: some minor edits for length allowed me to add the Other Features section. :)

Asmodian

31st March 2015, 06:57

Updated for madVR v0.87.15

Some big fixes. Portrait mode displays (rotated) work! Also Optimus black screen fixes.

There is a bug when picking the color matrix so BT.601 is selected for HD resolution video in some situations. Video flagged correctly is handled correctly as far as I have tested. Most likely a new build soon.

edit: fixed, up to madVR v0.87.17 now.

Asmodian

30th April 2015, 07:10

Updated to v0.87.21

A 64-bit build is now available. :)

No major changes in the options.

Karl78

1st May 2015, 23:29

Great job Asmodian. Thanks!

Magik Mark

10th May 2015, 00:06

Hey Asmodian,

Maybe we can update the guide for v881? Thanks for the awesome work

huhn

10th May 2015, 00:21

i recommend to wait for some time X-)

Asmodian

11th May 2015, 00:06

Yes, I hope and expect the options change a lot in the near future but I will still attempt an update for v0.88.2 soon. However, it will not cover the fine tuning options for the new enhancement shaders or similar details.

Asmodian

12th May 2015, 02:14

Updated for madVR 0.88.4, as always please offer corrections and suggestions. :)

D3D11 support allowing 10-bit output and a collection of sharpeners. Another major update!

tobindac

12th May 2015, 02:40

"present a frame for every VSync: when disabled madVR only presents new frames when needed, relying on D3D11 to repeat frames for each VSync."

Search the last week's pages of the madvr thread about this. He said that we should disable it for more speed if we have certain setups. I don't recall how but I'm in that group, I have an R9 290 on Windows 8.1 64.

Asmodian

12th May 2015, 02:49

Yes, are you suggesting I add a note that it might help performance? I don't seem to have a change in performance with it off or on.

tobindac

12th May 2015, 02:54

I doubt it's going to be noticeable to most. I guess it's some kind of minor improvement of the kind of 0.1-0.2% or something. He said the option solves problems in some setups but in others it's only some latency.

Warner306

12th May 2015, 08:11

If Upscaling Refinement shaders are added to the post-resized image, why do they only impact videos that are sub 1080p?

Asmodian

12th May 2015, 08:28

If Upscaling Refinement shaders are added to the post-resized image, why do they only impact videos that are sub 1080p?

I assume you are watching on a 1920x1080 display? They apply for me watching 1080p on a 2560x1440 screen. Think of them as part of upscaling and upscaling doesn't run when viewing 1080p at 1920x1080.

Warner306

12th May 2015, 09:35

Ok, one more question. Is SuperRes a form of image doubling, or can it do irregular upscales?

Shiandow

12th May 2015, 09:46

It can handle arbitrary upscales.

Asmodian

12th May 2015, 10:26

I understand SuperRes is more of a post process method than an independent scaler; it runs along with or on top of the other scaling option chosen, it doesn't replace it.

Shiandow

12th May 2015, 10:38

I understand SuperRes is more of a post process method than an independent scaler; it runs along with or on top of the other scaling option chosen, it doesn't replace it.

That's more or less correct, although it will still give reasonable results even if you use a really simple scaler such as nearest neighbour, or even if you just feed it a black image. But if you start with a better image you need less iterations, so it's a bit of a trade off.

nevcairiel

12th May 2015, 10:42

Personally I'm still not sure about SuperRes. On high quality content, the difference it makes is very minimal, and on low quality content it tends to sharpen the artifacts more than I would like.

Asmodian

12th May 2015, 11:21

I do think I like SuperRes for chroma. The minimal difference on high quality content seems like a good difference as well, at least with animated content.

leeperry

12th May 2015, 15:33

SuperRes: This is a post process method. From Shiandow (http://forum.doom9.org/showthread.php?p=1685124#post1685124)
Sweeet, thanks for digging the link for us as I was really looking forward to explanations regarding this feature because to my eyes NEDI+SuperRes can really match/exceed NNEDI3's PQ but I'm still finetuning settings atm :)

Shiandow

12th May 2015, 15:58

There have been some changes to the algorithm though, although the general idea has remained more or less the same. The steps now look more like the following:

Calculate an initial guess (using a different upscaler)
Downscale and calculate differences with original image.
Scale those differences to the final size
Improve guess by:

Softening the image
Subtracting differences with the original image
Sharpening (optional)
Removing aliasing (optional)
Removing ringing (optional)

Repeat steps 2-4 several times.

Warner306

12th May 2015, 20:46

I do think I like SuperRes for chroma. The minimal difference on high quality content seems like a good difference as well, at least with animated content.

How can you tell it's impacting the chroma? I thought it only applied to the luma, or the luma and chroma together.

huhn

12th May 2015, 20:52

How can you tell it's impacting the chroma? I thought it only applied to the luma, or the luma and chroma together.

because superres can be used for chroma only.

leeperry

13th May 2015, 03:35

SuperRes is chroma only?

Anyway, major kudos to Shiandow once again for the NEDI+SuperRes combo because to my eyes it doesn't have the artificial look of NNEDI3, doesn't look oversharpened and looks very natural/realistic AFAICS. motionDSP's fluff (http://www.motiondsp.com/products/IkenaForensic) doesn't look all that impressive anymore.

Would be fun to know how the combo fares with the samples in this link (http://www.infognition.com/articles/video_resize_shootout.html) but some of those SuperRes apps are far from realtime so this might be quite unfair.

Warner306

13th May 2015, 03:58

Warner306

13th May 2015, 03:59

huhn

13th May 2015, 04:09

SuperRes is chroma only?

you can use it as the chroma resizer too and don't have to resize luma at the same time. so it can be used for chroma only.

i guess this is more accurate.

Asmodian

13th May 2015, 04:29

SuperRes is chroma only?

SuperRes can be chroma only (check the "activate SuperRes filter" box on the chroma upscaling page).

This should be added verbatim to the guide.

Yes, I really wish I had more characters to use so I could. :(

What sections should/could be removed or pared back?
edit: I squeezed it in, I hope I didn't remove anything important. ;)

There have been some changes to the algorithm though, although the general idea has remained more or less the same. The steps now look more like the following:

Calculate an initial guess (using a different upscaler)
Downscale and calculate differences with original image.
Scale those differences to the final size
Improve guess by:

Softening the image
Subtracting differences with the original image
Sharpening (optional)
Removing aliasing (optional)
Removing ringing (optional)

Repeat steps 2-4 several times.

Thanks for the excellent and succinct explanation of SuperRes, I am sure many people with appreciate it. :)

...and all the work you put into the method itself, of course.

:thanks:

Warner306

14th May 2015, 07:29

Can you read the following post? I think your rendering diagram is wrong. In the post below, madshi explains the input for image doubling is "always RGB." This is not reflected in your diagram: http://forum.doom9.org/showpost.php?p=1721413&postcount=29697.

Asmodian

14th May 2015, 07:43

But my diagram does have the input for image doubling as always RGB? The "image doubling" madshi is talking about is the dashed orange box.

I leave out that madVR will use any of the Bicubic opitons if you have it set for image upscaling but I assume most users will not have a bicubic set for image upscaling. Otherwise that post agrees with my diagram.

Warner306

14th May 2015, 20:39

huhn

14th May 2015, 20:58

So you are saying a color space conversion from RGB to YCbCr is a necessary part of image doubling? This is what I questioned based on that post.

yes of cause you can't scale luma from RGB.
i wonder if a RGB -> YCbCr conversation is used when chroma and luma is doubled at the same time.

Asmodian

14th May 2015, 21:27

yes of cause you can't scale luma from RGB.
i wonder if a RGB -> YCbCr conversation is used when chroma and luma is doubled at the same time.

Madshi has confirmed that it is, around when doubling was first added. There is no special case that skips coversion to YCbCr for doubling both luma and chroma with the same number of neurons. it would be odd (rare) to use the same number of neurons for both anyway.

So you are saying a color space conversion from RGB to YCbCr is a necessary part of image doubling? This is what I questioned based on that post.

Luma is Y and chroma is Cb and Cr.

Warner306

14th May 2015, 22:32

Madshi has confirmed that it is, around when doubling was first added. There is no special case that skips coversion to YCbCr for doubling both luma and chroma with the same number of neurons. it would be odd (rare) to use the same number of neurons for both anyway.

Luma is Y and chroma is Cb and Cr.

Ok, I understand. Now that chroma doubling has returned, your chart will still have to be updated, or reverted back to its previous state.

Asmodian

15th May 2015, 03:31

Updated for v0.88.7

Ok, I understand. Now that chroma doubling has returned, your chart will still have to be updated, or reverted back to its previous state.

Yes, more modifications needed. I should have listened to huhn. ;)

Thanks for the feedback, I hope this new chart makes sense. Like any other feature, chroma doubling doesn't do anything if it isn't enabled so I felt the explicit path for chroma doubling disabled wasn't helpful.

Warner306

16th May 2015, 03:24

Updated for v0.88.7

Yes, more modifications needed. I should have listened to huhn. ;)

Thanks for the feedback, I hope this new chart makes sense. Like any other feature, chroma doubling doesn't do anything if it isn't enabled so I felt the explicit path for chroma doubling disabled wasn't helpful.

Not meaning to be a pest, but have you considered factoring in NEDI image doubling? It appears to double with RGB values because you can only scale both luma and chroma together.

Asmodian

16th May 2015, 04:10

Not meaning to be a pest, but have you considered factoring in NEDI image doubling? It appears to double with RGB values because you can only scale both luma and chroma together.

I added a note about it to the NEDI description but I don't think I will include it in the diagram. It would simply replace the entire "image doubling" box and I leave out a lot of details in the diagram already. I am also not sure how to add it in a way that makes sense.

If I did include it I suppose I would simply add NEDI image doubling to the black "image upscaling or downscaling" box and call the current "image doubling" box "NNEDI3 image doubling". :o

Asmodian

28th May 2015, 01:36

Small updates for 88.10.

still more to do.

Neeto

5th June 2015, 01:57

Good write up - thanks.
Any chance of adding in the registry entries - particularly the ones to turn off driver dithering in AMD

Asmodian

13th June 2015, 10:15

Good write up - thanks.
Any chance of adding in the registry entries - particularly the ones to turn off driver dithering in AMD

I am actually at the character limit already. :(

I also don't know those AMD registry entries. :o

Don't you only want to use them when testing if a display is really 10-bit? The drivers dithering any conversion they do seems good.

Asmodian

13th June 2015, 10:17

Updated for madVR v0.88.11.

I also moved Known Limitations or Complication out of the first post for space. Most of these are somewhat obscure so hopefully not much is lost.

Known Limitations or Complications:
-madVR can't do frame packed 3D
-When using a 3DLUT madVR can generate out of gamut values when displaying WtW unless it is emulating the GPU LUT when using 'enable windowed overlay'.
-Native DXVA decoding has some negative ramifications such as it is impossible to use force film (IVTC) or xy/vsfilter for subtitles.
-madVR cannot handle p010 (10-bit) input from DXVA native.
-madVR IVTC can detect a lot of cadences but can "only" handle some of them correctly. For example the rare cadence 3:2:3:2:2 (25 fps in 30i 60 fields) does not work well.
-madVR currently always switches to 23Hz when having forced film mode on while playing 59i content.

RainyDog

14th June 2015, 08:38

Updated for madVR v0.88.11.

I also moved Known Limitations or Complication out of the first post for space. Most of these are somewhat obscure so hopefully not much is lost.

Known Limitations or Complications:
-Native DXVA decoding has some negative ramifications such as it is impossible to use force film (IVTC) or xy/vsfilter for subtitles.
-madVR cannot handle p010 (10-bit) from DXVA native.

So we shouldn't use DXVA native in LAV if using madVR in 10bit output mode?

I don't see any issues with it at my end (though I don't have any material that needs IVTC nor do I use xy/vsfilter) and DXVA native seems to have a lower overhead than copyback or Cuvid.

nevcairiel

14th June 2015, 08:48

IMHO, You shouldn't use DXVA native with madVR at all really, using DXVA2-CopyBack is probably a better combination.

Qaq

14th June 2015, 09:09

So we shouldn't use DXVA native in LAV if using madVR in 10bit output mode?
Nope. Asmodian meant madVR's input, not output. 10 bit from madVR's output works just fine for me with DXVA native.
IMHO, You shouldn't use DXVA native with madVR at all really, using DXVA2-CopyBack is probably a better combination.CopyBack shares GPU bus with madVR, right? Sometimes GPU is too busy with madVR to route additional LAV's decoding traffic there.

nevcairiel

14th June 2015, 09:18

At least if you're using NVIDIA (and maybe on Intel? I forget), madVR will perform a "copy-back" anyway as thats the only way to preserve full chroma quality. So in that case, its better to just use Copy-Back from the get-go and avoid any limitations in madVR.

madVR used to have a trade quality option to disable this copy-back and risk bluring chroma a bit, but I think madshi removed it recently. Not sure if he worked-around the chroma limitation differently without costing much performance, or just decided to not offer the trade option anymore...

Qaq

14th June 2015, 09:33

I have Intel g1830 on workstation and AMD 7750 on HTPC. After upgrading from 5450 (64 bit bus) to 7750 (128 bit bus) I started with DXVA CB and had no issues untill I tried full 1920x1080 (with no borders) video. It looked stuttering with perfect madVR OSD stats in same time. I switched LAV from DXVA CB to DXVA native and playback turned to smooth again.
You see now why I'm very carefull with DXVA CB. Though CB is still the best solution for havy bitrate telecined videos.

Asmodian, sorry if it looks like off-topic, lets just aware AMD users do not use havy madVR settings with LAV DXVA copy-back )

huhn

14th June 2015, 12:07

Qaq

14th June 2015, 13:14

No, I don't have that movie anymore. And whats the point anyway? You don't want to test it totally different environment and say "everything is alright", do you?
My point is that DXVA copy-back puts additional load on GPU (GPU bus at least), which BTW is already too busy with madVR. Do you agree or not?

huhn

14th June 2015, 13:33

no i don't agree at all.
the copyback load is nothing at all. we have CB-D now on top of it.

why would madVR stutter with totally fine OSD stats that means it is a clear bug.

Qaq

14th June 2015, 13:41

Bug or not, it was ages ago. Next, not everyone has SSE4 capable CPU for CB direct. I don't.
Also, it would be nice to see GPU-Z screenshots for DXVA N and DXVA CB.

huhn

14th June 2015, 14:00

i tried that but with no really findings.

the load difference is so small it's hard to measure. but i don't have a weak GPU currently the weakest i have is a HD 4000 and a HD 4400 both a IGPU with ddr3 speed and not with gddr5 which is worlds faster.

Qaq

14th June 2015, 15:22

huhn

14th June 2015, 15:42

Well, for IGPUs/APUs it may be a totally different story of course. Different architectures, different traffic routing. I use myself CBd with madVR on G1830 just fine for almost any video now.
Let alone GPU-Z for AMD without screenshots. GPU-Z doesn't show much for my 7750 anyway. With AMD we have to "copyback" OpenCL data and here comes PCIe bottleneck. Lets route there DXVA CB traffic as well?

the openCL copyback issue use the same bandwidth yes. but do you notice a difference in speed there?

AMD d3d <-> openCL intreop looks like a terrible inefficient copyback while DXVA CP/D is very efficient.

Telion

14th June 2015, 16:02

Do AMD APUs suffer from this d3d <-> openCL copyback issue? Isn't hUMA supposed to overcome exactly this kind of problems? Should madVR explicitly implement some kind of hUMA support, or it is a driver's task? If it's the former, would madVR implement it some day?

huhn

14th June 2015, 16:27

madVR uses openCL cl_khr_dx9_media_sharing and it looks like it is designed to copyback.

AMD APU usually have a to weak iGPU anyway so the buttleneck is not the bandwidth and in the end they don't use PCIe for copyback in the first place

Soul

30th July 2015, 11:20

hi.. what is the best option for GPU queue size and CPU queue size? I have GTX750 NVidia.

Thanks

huhn

30th July 2015, 12:33

default is usually optimal.

XMonarchY

30th July 2015, 14:47

Is there truly no benefit to using Chroma Upscaling and/or Image Doubling/Quadrupling on TV's that are set to use 0-255 range, RGB, but limited to 4:2:2. In my case, 4:4:4 simply looks really bad (too dim) on my TV, leaving 4:2:2 as the only real choice.

Asmodian

30th July 2015, 17:09

The high cost Chroma Upscaling and Chroma Image Doubling options will not provide as much benefit (and their benefit is already minor) when using 4:2:2 but they do still provide a benefit. Luma Image Doubling is still very helpful.

However, I have never seen a TV that changes brightness or contrast due to 4:4:4 or 4:2:2, are you sure you are not confusing limited range (16-235) with color sub-sampling? Usually 4:4:4 or 4:2:2 is not something you can easily change unless your TV has a PC Mode. Also the difference is somewhat subtle, hence the need for an image to test for 4:2:2 or 4:4:4.

vivan

30th July 2015, 17:49

However, I have never seen a TV that changes brightness or contrast due to 4:4:4 or 4:2:2, are you sure you are not confusing limited range (16-235) with color sub-sampling? Usually 4:4:4 or 4:2:2 is not something you can easily change unless your TV has a PC Mode. Also the difference is somewhat subtle, hence the need for an image to test for 4:2:2 or 4:4:4.No, he is just used to LSD colors of his Samsung and that extreme postprocessing is disabled in PC mode or something.

pirlouy

30th July 2015, 18:02

I have a 2009 Samsung TV and indeed, when using PC mode, you lose all post-processing, and image looks insipid.
In fact, image is surely normal, but when you compare to Samsung post-processing, it looks insipid.
My solution has been to increase saturation in madVR, and I manage to have the same result.
I don't know if I should blame Samsung for that, but I prefer a saturated image now. xD

XMonarchY

30th July 2015, 18:51

No, he is just used to LSD colors of his Samsung and that extreme postprocessing is disabled in PC mode or something.

Might want to think before you post or at least ask. ALL post-processing is completely turned off and I calibrated each mode using both manual settings and 3DLUT with i1Display Pro. I could calibrate 4:2:2 mode color gamut, but not 4:4:4 PC mode because color gamut calibration does not apply to PC mode. This, however, should make no difference since I applied 3DLUT on top of each calibration and the results were very accurate.

I use Standard mode (with all post-processing disabled in User Menu and in Service Menu) + Game mode (4:2:2), which reduces input lag by further bypassing post-processing. It looks identical to fully calibrated Movie mode, but Standard + Game mode simply provides better/lower input lag.

4:4:4 PC mode looks bland, pale and kind of dim, even with 3DLUT. Its not just me - other Samsung VA owners reported similar findings. I also own Samsung PNF5350 Plasma TV and PC mode (4:4:4) looks perfectly fine, better than 4:2:2 modes. It seems like the problem is with specific Samsung VA screens.

Asmodian

30th July 2015, 19:04

XMonarchY

30th July 2015, 21:55

Obviously all post-processing is not off if it looks very different? All post-processing you can turn off is off. It sounds like Samsung has a correction/calibration tied to a post-processing chip that operates in 4:2:2 and changes the way the panel runs outside the 0-255 RGB input values (if a 3DLUT cannot fix it).

Have you tried creating a 3DLUT with a saturation intent when in PC mode?

What do you mean 3DLUT cannot fix it? In both 4:2:2 mode and 4:4:4 mode 3DLUT accuracy is just as high. Everything is within dE 1 with proper Rec.709 gamut (not wide gamut or anything). I even had it re-checked with i1Pro. Post-processing in Movie mode 4:2:2 is completely turned off by default, but it looks the same as Standard mode, which requires manual turning off of all post-processing. I am saying that all 4:2:2 modes look better than 4:4:4 mode, which to me looks dimmer and duller. I wonder if the difference is in luminance. Colors depends on saturation, hue, and luminance. There are controls for saturation and hue, but not luminance. I mean I even visited the hidden Service Menu and made asbsolutely sure no post-processing was used. Again, Game mode by itself further removes post-processing. When I switch to PC 4:4:4 mode, the same post-processing options are grayed out as the ones I manually turned off.

Let's say I calibrate 4:2:2 color gamut to read exactly as 4:4:4 color gamut for some specific color (fullscreen pattern). Then I can toggle between 4:4:4 PC mode and 4:2:2 non-PC mode (be it Standard or Movie). i1Display Pro and i1Pro will read it as the same color in both modes within the same dE, but 4:2:2 will look brighter/more vivid/vibrant (luminance)? Yet, if it was some post-processing affecting the color, then i1Display Pro or i1Pro would pick it up, wouldn't they? More vibrant or vivid or brighter color than the other color = different dE reading, but such is not the case. It also baffles me! I mean what kind of post-processing could this be?

Maybe it is possible that contrast ratio plays a role because the max I can squeeze out of 4:2:2 modes is 2600:1 and out of 4:4:4 only about 2200:1. Don't worry - all dynamic contrast and such junk is turned off in all modes, but contrast ratio's differ by about 400:1. However, would such a small difference in contrast ratio have any effect on a fullscreen color pattern looking different? The reason why PC mode has lower contrast ratio is because it is impossible for me to max out white balance gains due to locked controls. I am quite certain if I could max out white balance gains in PC mode, then I could get identical contrast ratio.

Bah, it could be simply because PC 4:$:4 mode is set at 90cd/m^2 and 4:2:2 modes are set at 110cd/m^2 using identical brightness and backlight settings.

huhn

30th July 2015, 22:05

by calibration you should first set white 255 to 100-120 cm² before 3d lut calibration.

a colorimeter is reading the all over color not from some pixel so post processing can still be used.

Asmodian

30th July 2015, 22:38

I wonder if the difference is in luminance. Colors depends on saturation, hue, and luminance. There are controls for saturation and hue, but not luminance. I mean I even visited the hidden Service Menu and made asbsolutely sure no post-processing was used.

Bah, it could be simply because PC 4:$:4 mode is set at 90cd/m^2 and 4:2:2 modes are set at 110cd/m^2 using identical brightness and backlight settings.

luminance = brightness. It looks like you found your answer. :)

XMonarchY

30th July 2015, 22:39

by calibration you should first set white 255 to 100-120 cm² before 3d lut calibration.

a colorimeter is reading the all over color not from some pixel so post processing can still be used.

So what kind of processing can affect the way pixels look or how vibrant or bright they look without affecting the actual color properties? I just can't imagine something like that.

huhn

30th July 2015, 23:17

first check if both modes have a white point of d65 at 120 cm².

XMonarchY

31st July 2015, 02:55

first check if both modes have a white point of d65 at 120 cm².

They do (all 3) have D65 white point, but cd/m^2 for each mode differs slightly. Movie mode is darker than Standard mode with identical values for brightness, contrast, and backlight. PC mode is dimmer than Standard mode, but brighter than Movie mode using those same identical settings and appropriate 1DLUT (each mode needs its own 1DLUT for games and 3DLUT for films). Again, this happens due to lack of RGB White Balance controls in PC mode.

ANYWAY, it doesn't really matter because all I wanted to know is whether there is still some benefit from upscaling chroma for 4:2:0 content to 4:2:2 mode and the answer is YES, but it is mild.

huhn

31st July 2015, 11:53

numbers please...

XMonarchY

7th August 2015, 00:15

OK, I verified - white point cd/m^2 are the same just like contrast ratio.

Here's a thought - can a slight bit of sharpness affect colors, as in full-screen single-color patterns? I disabled sharpness everywhere in User Menu, but also dug deep into Service Menu and loosened sharpness there, Edge Enhancement was also disabled. There is a SLIGHT big of sharpness compared to Movie or PC mode, but it is there. Can sharpness somehow affect pixel vibrance or color? I am talking about PURE single-color fullscreen patterns here, not some complex-color image.

huhn

7th August 2015, 05:14

on a full screen it shouldn't have an effect.
just ask your colorimeter it's a fullscreen effect so easy to read.

the YCbCr <-> RGB conversation may do that.

XMonarchY

7th August 2015, 12:54

on a full screen it shouldn't have an effect.
just ask your colorimeter it's a fullscreen effect so easy to read.

the YCbCr <-> RGB conversation may do that.

Colorimeter reads both colors as being identical. Both modes are set to RGB.

I decided to stick to PC Mode 4:4:4 this time because you can now use ReShade 3DLUT's with dispcalGUI which allow full grayscale and colorspace calibration with extreme accuracy to be applied in games without the need for CPKeeper or using Borderless Window modes AND you can apply 3DLUT's to madVR of course.

I figured 4:4:4 PC mode is just a SAFER bet of seeing proper colors (& have lower input lag) than using Standard mode with some kind of post-processing.

XMonarchY

8th August 2015, 00:12

the YCbCr <-> RGB conversation may do that.

So the conversion itself can create colors that colorimeters read as identical, but people perceive as different?

huhn

8th August 2015, 02:23

So the conversion itself can create colors that colorimeters read as identical, but people perceive as different?

no your meter must be broken.

XMonarchY

8th August 2015, 13:46

no your meter must be broken.

It isn't. I verify everything with i1Pro.

huhn

8th August 2015, 14:00

if your meter doesn't see the difference between two colors it is broken.

XMonarchY

8th August 2015, 14:13

if your meter doesn't see the difference between two colors it is broken.

i1Pro also broken?

huhn

8th August 2015, 15:02

if it doesn't see the difference between two colors what else?

XMonarchY

8th August 2015, 18:35

if it doesn't see the difference between two colors what else?

I think it may be perception because the chances of both i1Display Pro and i1Pro both going bad at the same time is slimmer than the problem being with something else, maybe even in my head...

Akeno

9th August 2015, 07:22

This is being discussed in the madVR thread but I thought it would be better to ask here: What exactly does overlay mode do? Also, is it fine to use gamma processing to something that looks pleasing to me even though I haven't calibrated my display or am I deviating too far from the source?

vivan

9th August 2015, 09:11

This is being discussed in the madVR thread but I thought it would be better to ask here: What exactly does overlay mode do?It utilizes hardware overlay, which is supposed to be the fastest way to draw anything. It also bypasses OS. You can read about it on wiki:
https://en.wikipedia.org/wiki/Desktop_Window_Manager
https://en.wikipedia.org/wiki/Hardware_overlay

Akeno

9th August 2015, 09:47

It utilizes hardware overlay, which is supposed to be the fastest way to draw anything. It also bypasses OS. You can read about it on wiki:
https://en.wikipedia.org/wiki/Desktop_Window_Manager
https://en.wikipedia.org/wiki/Hardware_overlay

Thanks. Other than enabling it in madVR, do I need to do anything else to utilize windowed overlay? Wikipedia and other users say that overlay mode prevents screenshots from capturing the video area but all of mine are working just fine.

vivan

9th August 2015, 10:06

I've just checked and overlay doesn't wotk in D3D11 mode, only in D3D9. I don't think any other settings could affect it (but I don't like this mode so I don't really know).
When enabled it should say "D3D9 overlay" in OSD (Ctrl+J).

Screenshots of the whole screen using PrtScr or build-in into media player (Alt+I in MPC-HC)?
In overlay mode PrtScr screenshots should be like this: http://i.imgur.com/29dK9q0.png

Akeno

9th August 2015, 10:33

I'm using D3D11 so that explains why the video still appears in screenshots. Are there any distinct advantages to overlay mode over D3D11? I got faster render times with the latter.

aufkrawall

9th August 2015, 11:28

Don't take render times too seriously.
Better try some processing that is almost or slightly too much for your GPU and check which render path gives no/fewest dropped frames/glitches.

Asmodian

10th August 2015, 23:35

I'm using D3D11 so that explains why the video still appears in screenshots. Are there any distinct advantages to overlay mode over D3D11? I got faster render times with the latter.

The advantages to overlay are what was already stated, they do not change when compared to DX11 or DX9 non-overlay modes. The biggest advantage for me is that it bypasses the calibration hardware in the video card; madVR loads the calibration from the video card, runs it at higher precision, and before dithering to 8-bit. It also doesn't need to switch into and out of FSE to provide its benefits, this mode switch is very fast now but still obvious.

Overlay is also very robust and tends to avoid V-sync issues, e.g. with multi-monitor setups in Windows 7.

Overlay does not support 10-bit. Overlay is also a bit less responsive when resizing the video player's window.

David

13th August 2015, 11:48

Question answered.

Asmodian

27th August 2015, 03:32

Finally an update for madVR v0.88.21

Does anyone have a good short description for Bilateral as is used in madVR for chroma upscaling? Thanks. :)

huhn

27th August 2015, 04:39

Bilateral is a chroma scaler that uses the luma channel as reference. this can be very good but it can fail too.

Asmodian

27th August 2015, 05:44

Bilateral is a chroma scaler that uses the luma channel as reference. this can be very good but it can fail too.

Excellent! Thanks. :)

Dlget

19th September 2015, 13:46

can anyone suggest setting for anime & normal video.
like that deinterlacing setting.-look pixel in frame center (good for broadcast,bad for anime)

aufkrawall

19th September 2015, 14:02

Main difference is that chroma scaling is much more important for colored cartoons than for any filmed content.
But luma still is very important, as it's for the contoures.
If you don't like high deband strength for filmed content, you may like it for cartoons, as it hardly kills details there.

Jinc AR for chroma is relatively soft, but this often doesn't become apparent. If you also watch bad sources which usually have lots of ringing in them, you may want to use NNEDI3 32 for chroma. But this is quite luxury, considering the high performance costs.

For luma, NNEDI3 is a no-brainer for cartoons, if you have enough performance and upscaling factor is 1.5x or higher.
64 neurons should be used, but also 32 are still better than super-xbr. super-xbr boosts ringing, makes lines fat and is also more aliased than NNEDI3.
However, you can use it for quadrupling after NNEDI3 if you lack performance for NNEDI3 quadrupling (but only important for very low-res content).

What's good for cartoons is mostly also good for filmed content.

amayra

24th September 2015, 20:36

how about new zoom settings ?

Asmodian

24th September 2015, 20:46

how about new zoom settings ?

I will add them soon. I am currently exactly at the character limit so it requires some refactoring, I will need to move advanced topics to another post and link to it. :(

Sparktank

24th September 2015, 22:02

I will add them soon. I am currently exactly at the character limit so it requires some refactoring, I will need to move advanced topics to another post and link to it. :(

Why not ask a mod to copy the first post a few times? They should be copied right after each other, in chronological order.

vivan

25th September 2015, 00:11

I don't think post "copying" is even possible (each post has an id and owner, they can't be changed).
Usually such problems are solved by deleting first few posts from other users or by moving other posts from the same user (if it's possible from chronological point of view).

Asmodian

25th September 2015, 04:49

Why not ask a mod to copy the first post a few times? They should be copied right after each other, in chronological order.

I am thinking I should probably make a new post, "Things to know about madVR" or similar, and link to that. It could grow quite large as well. I could also add known driver issues, tips and tricks, etc.

huhn

25th September 2015, 13:55

you should just ask an admin/mod to get the first couple of post and wait what they say.

the number of feature will grow too.

Asmodian

2nd October 2015, 08:40

Admin was contacted and was very helpful. :)

Updated for madVR v0.89.5.

Asmodian

16th October 2015, 03:07

Updated for madVR v0.89.11

Sparktank

16th October 2015, 04:02

Oh, cool. Looks it all worked out nicely. :)

markanini

19th October 2015, 20:39

Are deint=Film and deint=IVTC file tags different from each other?

Asmodian

19th October 2015, 21:52

Are deint=Film and deint=IVTC file tags different from each other?

Good catch, no they are not.

edit: A lot of those tags are synonymous.

Asmodian

1st November 2015, 03:58

Updated for madVR v0.89.13.

Automatic, file container level, rotation support!

Other good things as well, like improved DXVA chroma quality on Intel (and maybe Nvidia), but automatic rotation stands out for me. :)

magnusr

9th November 2015, 14:35

Got a new upscaling problem with my new computer. Using mphc with madvr. New computer has windows 10 and latest mphc with madvr. Latest Intel graphics driver. Intel i7 6700 using builtin HD530 graphics outputs to LG oled tv who is set for hdmi high (0-255). My previous Windows 10 computer had an ATI card and didnt have this problem.

Intel graphics driver: RGB to 0-255 (quantization full range). Also had to set brightness in the driver itself to +8 to get it right. YCbCr deactivated, IT-content activated. Video color settings: Program Settings. And deactivated sking tone and contrast enhancment.
madvr is set to 0-255 8 bit for the display and in lav i have tried both untouched and 0-255.

Playing 1080p content everything is perfect, but the problem occurs when watching 720p content who gets upscaled to 1080p. Brightness increases a lot. Black is gray. If I set the tv to rgb limited it fixes it. Anyone got a clue? Why do I have to set the tv to rgb limited for upscaled movies but not 1080p movies?

huhn

9th November 2015, 17:14

what display mode do you use?

magnusr

9th November 2015, 19:51

1080p23, 1080p24, 1080p25, 1080p29, 1080p30, 1080p60

In this case 1080p23

Asmodian

9th November 2015, 20:36

That really does not make sense to me. There is no difference, to the Intel drivers, after madVR gets done with 720p or 1080p. madVR resizes to the output resolution and then hands it to the Intel drivers either way.

Are you using and DXVA or QS decoding features? I would test with them all disabled (software decoding in LAV and no DXVA resizing). Are you sure the test file is not the issue?

magnusr

9th November 2015, 23:03

Seems it might be 720p file I tested. Tested another movie and it seemed fine. Thanks

Asmodian

4th December 2015, 01:56

Finally updated for madVR v0.89.17. Suggestions for better descriptions for sharpen edges, thin edges, and enhance detail are welcome. :)

Asmodian

18th December 2015, 01:18

Updated for madVR v0.89.18

Nullack

23rd December 2015, 01:59

Thanks for the guide Asmodian

Can I suggest adding that Nvidia 3D Stereoscopic vision isnt compatible with madVR. This isnt just packed frames like you mention. madVR needs 3D vision disabled in the nvidia control panel.

Perhaps also stating what are the default settings for them all. Like auto FSE is default.

Thanks

SpoCk0nd0pe

23rd December 2015, 21:26

Thank you very much for the guide!

Are you o.k. with recommended options questions in this guide or should I go somewhere else?

Asmodian

23rd December 2015, 23:22

Asmodian

23rd December 2015, 23:25

SpoCk0nd0pe

27th December 2015, 06:46

I am happy to answer questions for recommended options but I cannot put them in the guide without a lot of discussion of specific hardware.

Anything that is definite I do try to add but that is only a few things like do not turn off dithering.

Thank you!

I've read through a lot of pages on the main madvr thread but the thread is really a little too ridiculous to catch up :/

I got some questions regarding the upscaling algorithms and filters:
For Luma:
The way I read it, NNedi3 and Jinc are still the overall best quality filters.
SuperRes got me some additional non-destructive sharpness, especially with SD sources. Is there any harm in doing it 4 times (highest option)? What are the advantages/disadvantages in doing it in LL?

For Chroma:
I've read a lot of people using Jinc or even NNEDI3 for chroma. Are those people talking anime? Because as far as I know the artifact hiding properties of SoftCubic (I use 80 softness) is desirable for chroma, as too sharp chroma can look unnatural (especially for SD sources, doubling the chroma with NNEDI first looks better though). Is SuperRes for chroma desirable?
Afaik chroma for anime/cartoon is a different story and behaves more like Luma.

Asmodian

27th December 2015, 11:54

For Luma: you are right about Jinc and NNEDI3 in my opinion but comparing Jinc to super-xbr is hard. Important only if you don't have the performance for NNEDI3. Some do like xbr more than NEEDI3 though so best is hard to say for sure. Lanczos is sharper than Jinc too. :p

SuperRes has diminishing returns with more passes, less benefit for the extra performance hit. As I understand linear light for SuperRes emphasizes erorr in small light details while gamma light emphasizes small dark details. There difference is fairly subtle in my experience.

For Chroma: I agree, madshi added SoftCubic for chroma and it was the default at first. A very reasonable choice for filmed content. I also agree that anime is different and sharper chroma scaling is more beneficial. People do like sharp though. ;)

SpoCk0nd0pe

27th December 2015, 16:34

Thank you for your quick response!

SuperRes has diminishing returns with more passes, less benefit for the extra performance hit. As I understand linear light for SuperRes emphasizes erorr in small light details while gamma light emphasizes small dark details. There difference is fairly subtle in my experience.

Just to make sure I got you right: You are saying that LL on is better for darker scenes while LL off is better for brighter ones? Which one do you prefer?
An SuperRes is mostly a case of more helps more? SuperRes seems to hide some SVP artifacts btw. Especially the halos on panning camera movements and bright background seem less agressive with SuperRes.
In what cases is SuperRes harming the image? Are those cases negligible?

For Chroma: I agree, madshi added SoftCubic for chroma and it was the default at first. A very reasonable choice for filmed content. I also agree that anime is different and sharper chroma scaling is more beneficial. People do like sharp though. ;)

I do think that NNEDI64 chroma doubling helps image quality if you also do luma doubling. Before doing SoftCubic upscaling that is.

What do you think about SuperRes for chroma upscaling refinement (for film content)? Does it improve the picture without the overly sharp look of Jinc? Do you use it and if yes, what settings do you use?
I would bet it is desirable for anime.

aufkrawall

27th December 2015, 19:37

Gamma light SuperRes is more correct when the image has been downscaled before in gamma light. But as we usually can't know how the material we feed madVR with has been downscaled, this doesn't really matter.
So, in practice you can say: LL SR is darker, GL brighter. LL SR often makes dark lines fat, that's why GL probably can be considered as better or at least the better default setting.

Asmodian

30th December 2015, 01:29

Just to make sure I got you right: You are saying that LL on is better for darker scenes while LL off is better for brighter ones? Which one do you prefer?
An SuperRes is mostly a case of more helps more? SuperRes seems to hide some SVP artifacts btw. Especially the halos on panning camera movements and bright background seem less agressive with SuperRes.
In what cases is SuperRes harming the image? Are those cases negligible?

To be honest I do not use SuperRes or SVP (I cannot handle SVP's artifacts). I find SuperRes adds back minor aliasing which was present in the source and which NNEDI3 doubling removed.

I do think that NNEDI64 chroma doubling helps image quality if you also do luma doubling. Before doing SoftCubic upscaling that is.

What do you think about SuperRes for chroma upscaling refinement (for film content)? Does it improve the picture without the overly sharp look of Jinc? Do you use it and if yes, what settings do you use?
I would bet it is desirable for anime.

I agree that there is something nice about using NNEDI3 chroma doubling when using luma doubling but I generally do not like NNED3 chroma upscaling. I always use Jinc AR for image upscaling and I do not consider Jinc very sharp. :)

I normally use Jinc AR for chroma upscaling except for very low resolution sources when I sometimes use NNEDI3 64. I never use SuperRes for chroma. I also quite like Mitchell-Netravali without AR for chroma upscaling, when testing I tend to pick one or the other depending on the source (and mood? :p). SoftCubic is a bit too soft for me unless the source is very low quality.

I have never found universal settings for upscaling refinement, my optimal settings are entirely source dependent.

ikarad

31st December 2015, 14:41

Thanks for the guide Asmodian

Where is the guide? Is it the first page?

Asmodian

1st January 2016, 00:10

The first two posts. It is not really a guide, more of a reference to make it easier to pick your own settings.

MadVR does not lend itself to a complete guide in my opinion. To much is dependent on personal preference, the particular display, the computer's hardware, the content, etc.

Anima123

1st January 2016, 05:14

How can I use IVTC to decimate a video fps 59.940 to fps 24.970?

As I remembered, I've done it with success once.

Asmodian

1st January 2016, 05:35

All you should need is to enable IVTC.

Anima123

1st January 2016, 15:30

All you should need is to enable IVTC.

Well, problem is, I cannot find the shortcut to enable IVTC now.

Could someone confirm the ways of enable it has been changed for latest version of madVR?

huhn

1st January 2016, 15:34

control +shift + alternative +d = deinterlancing
than control +shift + alternative +t to switch to film mode.

SpoCk0nd0pe

2nd January 2016, 16:41

I find SuperRes adds back minor aliasing which was present in the source and which NNEDI3 doubling removed.
Yes, I see that too. It is really very minor though and imho worth the overall improvements in image quality.

I do not like overly sharp chroma, even with HD content. SoftCubic 80 is the sweet spot for me personally. But scaling too much with just soft cubic makes the chroma too burred imho, so doing doubling first is nice.

Thanks a lot for the input!

Nullack

8th January 2016, 09:44

rendering

general settings:
enable windowed overlay (Windows 7 and newer): Only available on Nvidia and Intel GPUs. Uses a low level overlay method which bypasses the GPU LUT (monitor profile) so madVR emulates it when using this option, this is done in 16-bit so madVR can provide better quality than the GPU. Overlay also bypasses the OS to a large extent; screen-shots are not possible.
use Direct3D 11 for presentation (Windows 7 and newer): Use Direct3D 11 instead of D3D9. Requires a restart of madVR to take effect. Required for >8 bit output. Overrides windowed overlay.
present a frame for every VSync: when disabled madVR only presents new frames when needed, relying on D3D11 to repeat frames for each VSync. Disabled may improve performance but may also cause presentation glitches.

Hi

So even if I have enable window overlay set to on, if I have DX11 also set to on, the window overlay is disabled? The UI settings allow both to be on.

Since DX11 is required for >8bit output thats the option Im interested in. Is there any general commentary on the DX11 setting in practical use?

Testing on Win 10 x64 is showing many presentation glitches if the present frame for each Vysnc is not enabled. This is with my setup being Nvidia GTX 960, UHD 4K display through HDMI 60 FPS.

Thanks

Nullack

8th January 2016, 09:50

disable desktop composition (Vista and Windows 7): Disables Aero which has its own v-sync that can conflict with madVR’s.

Is this setting not applicable for Windows 10 installations?

Asmodian

8th January 2016, 09:54

Hi

So even if I have enable window overlay set to on, if I have DX11 also set to on, the window overlay is disabled? The UI settings allow both to be on.

Since DX11 is required for >8bit output thats the option Im interested in. Is there any general commentary on the DX11 setting in practical use?

Testing on Win 10 x64 is showing many presentation glitches if the present frame for each Vysnc is not enabled. This is with my setup being Nvidia GTX 960, UHD 4K display through HDMI 60 FPS.

Thanks

Yes, if you enable DX11 overlay is never used. You can verify this in the OSD.

DX11 works well on my system, slightly more responsive than DX9 or overlay with similar performance (I am also using Win10 x64 but with a Nvidia GTX 980 Ti). I also need present frame for each Vsync enabled. Both DX11 and full screen exclusive are required for 10 bit output. Don't forget to change the bit-depth in the display's properties as well.

Is this setting not applicable for Windows 10 installations?

Yes, it does not apply for Windows 8 or 10. Desktop composition cannot be disabled on Windows 8 or 10.

Nullack

8th January 2016, 11:10

Thanks mate I appreciate your words of wisdom

CarlosCaco

13th January 2016, 17:08

What the new option "use Sigmoidal light" on image upscaling does?

Asmodian

24th January 2016, 01:17

What the new option "use Sigmoidal light" on image upscaling does?

Finally added a description; description borrowed from the DitherTools (http://avisynth.nl/index.php/Dither_tools) manual.

Updated for madVR v0.90.3

Asmodian

6th February 2016, 08:34

Updated for madVR v0.90.4

Asmodian

23rd February 2016, 09:21

Updated for madVR v0.90.12.

A lot of improvements in madVR but the only major change for this post was removal of all mention of Jinc only being for upscaling and 3D doesn't crash anymore. :)

Siso

13th March 2016, 11:55

Excellent explanation to the madvr options, will there be an info for the stats in madvr etc. "average rendering time, present" and more?

For example I'm kinda confused which of the two modes are good for my card gtx 550 ti, the "present" stats are different and I don't know what they mean exactly.

http://t1.someimage.com/UYsBa8K.jpg (https://someimage.com/UYsBa8K) http://t1.someimage.com/3bSCuNP.jpg (https://someimage.com/3bSCuNP)

huhn

13th March 2016, 22:16

you can usually ignore present times unless you have an issue.

present times has "nothing" to do with performance. of cause this is over simplified.

JarrettH

15th March 2016, 04:45

I have a 550 Ti too. It's still capable of quite a lot. :cool:

Siso

15th March 2016, 09:21

I have a 550 Ti too. It's still capable of quite a lot. :cool:

Yeah, it has some sort of good potential :D

CountryBumkin

23rd March 2016, 21:40

I think it would be great if you could add some information on the first page (or so) that explains a little about what the Queue numbers and Render/Present times mean.

It would help us "madVR-challenged" to understand what is going on when things don't work/look just right.

For example my "Present Queue" is 0-1/8 and "Render Queue" is 1-2/8. And I have a lot of dropped frames.

I believe a low Present Queue means that my computer is too weak for the settings. Is this correct?

Also my Present Time is 4.02ms and the Rendering Time is 86.41ms. What is the acceptable Rendering Time(s)?

BTW - I am so clueless on this stuff, it took me ten tries to answer the spam question in order to post.

Asmodian

24th March 2016, 20:53

I don't think I want to put more of that kind of information in the first post, it is meant to be a description of all the options, not really a guide. I want to keep the beginning of the post limited to only describing options.

I think I do mention everything you need to know about queues and present times already, but in the second post.

CPU queue size: [16] The buffer in system memory for decoded video and xySubFilter subtitles.
GPU queue size: [8] The buffer in the graphics memory for madVR’s rendering.
how many video frames shall be presented in advance or how many backbuffers shall be used: [8 or 3] A final buffer after madVR is done rendering.

At the start of "More Advanced Topics" I have:
"The sum of the average stats "deinterlace" (if present), "rendering", and "present" should be a bit below the frame time in the line "v-sync [X] ms, frame [Y] ms" to avoid dropped frames and presentation glitches."

Is this the information you were looking for or is there something else you want to know? Do I need a "Trouble Shooting" section in "Advanced Topics" so people do not miss it? Or maybe simply a "madVR OSD" section instead? I think I like that idea, I will add that before Advanced Topics.

I did update the description of "how many video frames shall be presented in advance" to:
how many video frames shall be presented in advance or how many backbuffers shall be used: [8 or 3] The present queue, a final buffer in DirectX, after madVR sends completed frames to Windows.

Is this better?

Thanks. :)

CountryBumkin

28th March 2016, 18:18

I think I do mention everything you need to know about queues and present times already, but in the second post.

CPU queue size: [16] The buffer in system memory for decoded video and xySubFilter subtitles.
GPU queue size: [8] The buffer in the graphics memory for madVR’s rendering.
how many video frames shall be presented in advance or how many backbuffers shall be used: [8 or 3] A final buffer after madVR is done rendering.

At the start of "More Advanced Topics" I have:
"The sum of the average stats "deinterlace" (if present), "rendering", and "present" should be a bit below the frame time in the line "v-sync [X] ms, frame [Y] ms" to avoid dropped frames and presentation glitches."

Do I need a "Trouble Shooting" section in "Advanced Topics" so people do not miss it? Or maybe simply a "madVR OSD" section instead? I think I like that idea, I will add that before Advanced Topics.

I did update the description of "how many video frames shall be presented in advance" to:

Is this better?

Thanks. :)

I would love to see a "Troubleshooting Section". I think it would be very helpful to those of us that are technically challenged in this area.

I found this info posted by madshi (about page 10 of madVR thread):

If you want to give madVR even a chance to achieve smooth playback you have to achieve an "average gpu rendering time" that is a notch lower than the "movie frame interval". I'd say it should not be higher than 35ms for 23.976 playback.

Now the optimal configuration for madVR is if you drive the display with the same refresh rate the source is in (in your case 23.976Hz). If you have to use 59.940Hz, things get slightly more difficult for madVR. In that case you may have to bring "average gpu rendering time" down even more.

and

Originally Posted by yesgrey3 View Post
The frame rendering is ~14ms. For playing in my monitor at 72Hz (13.9ms) is not enough, but should be fine to use with my projector at 48Hz (20ms). If I use Lanczos or Spline resampling the rendering time jumps to ~20ms, so I should not be able to use those resamplers with smooth playing...

You would be right if madVR needed to draw a *new* Blu-Ray frame every VSync. But if it did that, you'd get 3x respectively 2x as fast movie playback as you should get. madVR has to skip lots of VSyncs with your refresh rates. Because of that the most important reference time is not the "vsync interval" but the "frame interval".

This is the "kind of info" I am looking for when it comes to troubleshooting (understand the Stats).

Thanks for your help.

Asmodian

29th March 2016, 02:59

From your post I take it that:

The sum of the average stats "deinterlace" (if present), "rendering", and "present" should be a bit below the frame time in the line "v-sync [X] ms, frame [Y] ms" to avoid dropped frames and presentation glitches.

Did not tell you all this already? Everything covered in your post is what I was trying to communicate with that line.

I am having trouble understanding if you want any new information or simply didn't understand what I was trying to say with that line. Do I just need to rephrase it or add an entirely new section? Adding a new section will not help if the problem is with understanding what I wrote instead of actually missing information.

edit: Does this example I included help clarify it?
The sum of the average stats "deinterlace" (if present), "rendering", and "present" should be a bit below the frame time, Y, in the line "v-sync [X] ms, frame [Y] ms" to avoid dropped frames and presentation glitches. For example, progressive 29.97 fps video has a new frame every 33.37 ms so "rendering"+"present" needs to be a few ms below 33.37 ms. 23.976 fps video has a new frame only every 41.71 ms so you can use more demanding settings with lower frame rate video. Exactly how far below the frame time is required for glitch free playback is dependent on the system but a few milliseconds is usually sufficient.

Warner306

30th April 2016, 03:11

Bump.

Asmodian

6th May 2016, 05:39

Bump.

Thanks. I finally updated to 90.17, I used your description for the anti-bloating filter.

This reduces the fatness created by line sharpeners. Not needed for crispen edges, which is already "skinny."

Asmodian

11th May 2016, 09:29

Updated for madVR v0.90.18.

Another artifact removal option, 'reduce ringing artifacts'! :)

Asmodian

23rd May 2016, 23:40

Updated for v0.90.20.

New flow chart.

e-t172

30th May 2016, 17:49

After some experiments I've noticed (http://forum.doom9.org/showthread.php?p=1768980#post1768980) that the way 3DLUTs and GPU gamma curves work can be a bit counter-intuitive at times. I have a suggestion:

Change:

calibrate this display by using external 3DLUT files: Use a calibration package such as Argyllcms, Calman, or LightSpace to generate a 256x256x256 3DLUT to perform a full calibration. For Argyllcms see this AVS Forum thread (http://tinyurl.com/nx87y7c), for Calman (http://tinyurl.com/map94jl), or for LightSpace (http://www.lightillusion.com/madvr_manual.html).
disable GPU gamma ramps: This disables the Windows calibration or monitor profile. It affects everything displayed, not only madVR. madVR will restore the profile when it is closed. This option changes its behavior when using the 'rendering -> general settings' option 'enable windowed overlay' to only apply to madVR because madVR is emulating the gamma ramps in that case.

To:

calibrate this display by using external 3DLUT files: Use a calibration package such as Argyllcms, Calman, or LightSpace to generate a 256x256x256 3DLUT to perform a full calibration. For Argyllcms see this AVS Forum thread (http://tinyurl.com/nx87y7c), for Calman (http://tinyurl.com/map94jl), or for LightSpace (http://www.lightillusion.com/madvr_manual.html). If the provided 3DLUT file contains calibration curves (e.g. in the case of ArgyllCMS, the collink "-a" or "-H" options were used), madVR will load the calibration curves into the system-wide GPU hardware gamma ramps, which affects everything displayed, not just madVR. madVR will restore the original gamma ramps (if any) when it is closed.
disable GPU gamma ramps: Clears system-wide GPU gamma ramps (monitor calibration), if any, replacing them with linear (no-op) curves. This affects everything displayed, not just madVR. madVR will restore the original GPU gamma ramps (if any) when it is closed. This option changes its behavior when using the 'rendering -> general settings' option 'enable windowed overlay' to only apply to madVR because madVR is emulating the gamma ramps in that case. This option has no effect if a 3DLUT file that contains calibration curves is loaded (see above).

(I also took the liberty of avoiding the term "profile", which is confusing because an ICC profile contains more than just calibration curves.)

Asmodian

1st June 2016, 04:37

After some experiments I've noticed (http://forum.doom9.org/showthread.php?p=1768980#post1768980) that the way 3DLUTs and GPU gamma curves work can be a bit counter-intuitive at times. I have a suggestion:
..snip..
(I also took the liberty of avoiding the term "profile", which is confusing because an ICC profile contains more than just calibration curves.)

I like it! Done. :D

More precise and very useful information. I also have enough characters to expand that section a bit.

Thanks. :)

Asmodian

25th June 2016, 00:38

Updated to v0.90.21, no changes. ;)

Asmodian

24th July 2016, 22:59

Updated for madVR v0.90.23, added a section for hdr.

Asmodian

29th July 2016, 10:14

Added a section for the empty file based hidden options for madVR.

leeperry

30th July 2016, 18:30

Added a section for the empty file based hidden options for madVR.

Sweet, thanks!

I believe ShowBlackBars isn't active anymore but DontRenderAfterStop definitely seems to help with hiccuping during transitions between movies with PotPlayer in seamless playback :)

madshi kept telling me that it was PotP's fault and PotP's coder said it was mVR's so I shall run more tests but closing PotP even seems snappier now.

Asmodian

31st July 2016, 18:24

Sweet, thanks!

I believe ShowBlackBars isn't active anymore but DontRenderAfterStop definitely seems to help with hiccuping during transitions between movies with PotPlayer in seamless playback :)

madshi kept telling me that it was PotP's fault and PotP's coder said it was mVR's so I shall run more tests but closing PotP even seems snappier now.

Glad it helped. :)

"ShowBlackBars" is still active, I just tested it with madVR v0.90.23.

Light001

1st September 2016, 05:26

Looking for help Again

i5 3550 4.1ghz, Gtx 980, windows 10 pro 64bit and most recent nvidia driver. BenQXl2420T, i watch mostly pirated 1080p Quality vidoes with some older shows 480p. My monitor only supports 60hz and 100hz and 120hz. I used Tft Centrals color calabration for my monitor. Watching things like game of thrones using madvr and Mpc-hc its abit dark unless i turn up brightness. Do i touch anything in the Nvidia Control Panel? under Video ? (Edge Enhance,Noise Reduc,Deinterlacing-unverse telecine)??
Under Devices I do not know what should be set, i assume the pc levels 0-255 is fine and 8 bit for native display? 3d format is auto. calibration is set to disable display modes is also unchecked color and gamma is left alone and hdr which is new, i selected the top option.

Processing All unchecked, Artifact removal unchecked,Image Enhance unchecked, and zoom, disable scaling if image size changes by only 25 lines or less. Scaling Algorithims are Chroma up- Jinc, Image Down-Jinc,Image doubling-off,image upscale-jinc with active anti ringing on all 3. Nothing in upscale refinement, Rendering has nothing checked off under general and the que for cpu is 16 and gpu is 8, Smooth Motion on, althought i do not know what to be looking for, just that in some movies when the scene is in motion from side to side i see visually its not very smooth, like a little judder or something, Dithering is Ordered and nothing checked in trade quality for preformance.

In Mpc Under lav Decored, top corner is all auto with the box checked 0-255 and ordered dithering, Dxva2 copyback with all boxes checked and software deinterlacing disabled.

Does anything stand out here that i would or should want to change? i would prefer to stay under 70% Gpu usage. Its been along time since i posted and i forgot alot of stuff :(. As i said most the stuff i watch is 1080 or 720 rip with the odd older show @ 480p. or worse. i find it dark watching game of thrones, unless i turn up brightness in mpc, which is kinda annoying. But i gave up trying to find a better set up for the monitors settings. Help me obi won kenobi, your my only hope <3

Sorry to toxic your thread, but if anyone can give me some input base on info provided or can inbox me id greatly appreciate it. if i find movie nirvana or epic results id glady paypal a few bucks for your time. TY!

Asmodian

4th September 2016, 06:56

Updated for madVR v0.90.24

SweetLow

7th September 2016, 12:20

To Asmodian
As minimum one hidden option exists which not mentioned in your reference:
KeepLastFrameOnStop

oddball

11th September 2016, 20:43

I am getting noticeable frame skipping on VFR sources. I can see the numbers of frames repeated/dropped but every now and then I see a visible skip. I tried resetting to defaults and it does not help. I can also not play any video other than in DX11 mode. If I leave it unticked MPC-HC hangs. MadVR settings continue to be accessible. But I have to kill MPC-HC in Win10 Task Manager.

I am using AMD's latest drivers but this has continued through from using a AMD R90 290 to changing the GPU to a RX 480.

I have also tried older versions of MadVR and it still hangs with DX11 unticked. This only occurs over HDMI to a either a Denon AVR or directly to a Panasonic plasma HDTV.

Because MadVR hangs and does not crash I don't know how to get any crash info. Is it MadVR or MPC-HC? If MPC-HC where do I get the crash info? I have also tried older non-nighlty buiilds of MPC-HC.

Also I have had issues with delayed playback enabled with DX11 ticked and switching between fullscreen on the HDTV and windowed on the main monitor.

Please advise on how I can diagnose these issues and post a crash report.

I don't want to have to wipe Win10 and start over but this has been going on for many weeks now and I still don't know how to fix this.

Any help appreciated on what I should do to find crash info and diagnose this pain in the butt problem.

EDIT: I tried it in MPC-BE out of curiosity and with DX11 unticked and sending it stright to fullscreen on the HDTV I get the MPC-BE logo window (no video) and the sound continues on. I also see 'madVR reports -resetting direct3d devices failed (8007000e)' and '-creating direct3d devce failed (8007000e)'. I am at a loss right now.

EDIT2: If I stop it going fullscreen on the HDTV and allow it to play on the monitor first it plays,pauses, ff/rw etc. But if I force it to the HDTV the playback stops and I can do nothing even though the interface still allows me to pause, play etc. But the image and sound remain static on what was previously there.

Asmodian

11th September 2016, 21:00

To Asmodian
As minimum one hidden option exists which not mentioned in your reference:
KeepLastFrameOnStop

Thanks! I will add it when I get back home and off mobile.

I am using AMD's latest drivers but this has continued through from using a AMD R90 290 to changing the GPU to a RX 480.

I believe many users are reporting issues with AMD's newest drivers. I think 13.2 is the still the optimal version according to reports but that version doesn't support the RX 480 so I don't know what to suggest. Maybe try a few of the older versions that do support the RX 480. :(

That said you can also try smaller and larger queues, especially the present queues. Try changing the "use a separate device for presentation" and the "present a frame for every v-sync" options as well. Make one change at a time.

Sorry, I do not have any AMD cards myself so I cannot offer better advice.

oddball

11th September 2016, 21:28

Thanks! I will add it when I get back home and off mobile.

I believe many users are reporting issues with AMD's newest drivers. I think 13.2 is the still the optimal version according to reports but that version doesn't support the RX 480 so I don't know what to suggest. Maybe try a few of the older versions that do support the RX 480. :(

That said you can also try smaller and larger queues, especially the present queues. Try changing the "use a separate device for presentation" and the "present a frame for every v-sync" options as well. Make one change at a time.

Sorry, I do not have any AMD cards myself so I cannot offer better advice.

Yes I can get it to playback using DX11 just not without it being ticked. Anyhow I have discovered something interesting. When setting the HDTV to 60Hz in Windows and setting MadVR to not switch to any display refreshes (1080p23 etc blanked out) and just having smooth motion ticked I get smooth VFR playback. So something is up with MadVR and refresh rate switching I believe. I tend to run the HDTV at 23Hz and MadVR to just auto switch out at that rate. Let me experiment some more.

EDIT: Ooh! MadVR just made a crash dump.

https://drdump.com/UploadedReport.aspx?DumpID=10749439&SecondVisit=1

Asmodian

11th September 2016, 21:36

Yes I can get it to playback using DX11 just not without it being ticked. Anyhow I have discovered something interesting. When setting the HDTV to 60Hz in Windows and setting MadVR to not switch to any display refreshes (1080p23 etc blanked out) and just having smooth motion ticked I get smooth VFR playback. So something is up with MadVR and refresh rate switching I believe. I tend to run the HDTV at 23Hz and MadVR to just auto switch out at that rate. Let me experiment some more.

Really a lot of TVs have bad 24 Hz modes, maybe it is the TV not processing 24Hz correctly with the somewhat odd patterns of VFR video. 60 Hz with smooth motion is a good option IMO, is there a reason you do not like it?

I do not think it is it is likely to be a problem with madVR's refresh rate switching, it works properly for too many people.

Edit: I am highly suspicious that the crashes with anything but DX11 is due to AMD's driver, both AMD and Nvidia have had driver issues exposed by madVR in the past and there have been a lot of reports of issues with the latest Crimson drivers.

Edit2: Are you using Software decoding LAV? If not, try using it.

oddball

11th September 2016, 22:16

Really a lot of TVs have bad 24 Hz modes, maybe it is the TV not processing 24Hz correctly with the somewhat odd patterns of VFR video. 60 Hz with smooth motion is a good option IMO, is there a reason you do not like it?

I do not think it is it is likely to be a problem with madVR's refresh rate switching, it works properly for too many people.

Edit: I am highly suspicious that the crashes with anything but DX11 is due to AMD's driver, both AMD and Nvidia have had driver issues exposed by madVR in the past and there have been a lot of reports of issues with the latest Crimson drivers.

Edit2: Are you using Software decoding LAV? If not, try using it.

I always use LAV as it is inbuilt into the latest MPC-HC.

I had already tested this with the default Win10 provided drivers with the same results. The only way to be sure would be to reinstall Win10 from scratch. Not something I would willingly do lightly.

The only way I have got playback to work is by ticking DX11. Exclusive mode seems to work fine 'most' of the time but I have the odd video that gives a MadVR error in the top left corner. Sometimes they playback fine without exclusive mode selected.

What I have found is that unticking *BOTH* 'delay switch to exclusive mode by 3 seconds' and 'present several frames in adance' has helped prevent dropped frames when playing back VFR sources. Also running at 60Hz and selecting Smooth Motion does the same thing (not dropped frames on VFR).

Unticking 'Delay playback until render queue is full' prevents MPC-HC locking up with a blank screen on the HDTV going to fullscreen. If I open it on the monitor instead it plays back no problemo. If I then force it to fullscreen on the HDTV it just locks up MPC-HC.

BTW when I upgrade AMD's drivers I always DDU in safe mode.

oddball

11th September 2016, 23:14

I always use LAV as it is inbuilt into the latest MPC-HC.

I had already tested this with the default Win10 provided drivers with the same results. The only way to be sure would be to reinstall Win10 from scratch. Not something I would willingly do lightly.

The only way I have got playback to work is by ticking DX11. Exclusive mode seems to work fine 'most' of the time but I have the odd video that gives a MadVR error in the top left corner. Sometimes they playback fine without exclusive mode selected.

What I have found is that unticking *BOTH* 'delay switch to exclusive mode by 3 seconds' and 'present several frames in adance' has helped prevent dropped frames when playing back VFR sources. Also running at 60Hz and selecting Smooth Motion does the same thing (not dropped frames on VFR).

Unticking 'Delay playback until render queue is full' prevents MPC-HC locking up with a blank screen on the HDTV going to fullscreen. If I open it on the monitor instead it plays back no problemo. If I then force it to fullscreen on the HDTV it just locks up MPC-HC.

BTW when I upgrade AMD's drivers I always DDU in safe mode.

EDIT: MPC-BE does not lockup up when going fullscreen to my HDTV when all the above three options I mentioned are ticked. I tried setting MPC-BE exactly the same as MPC-HC is except MPC-BE does not have LAV built in so I used LAV as external filters. I tried uninstalling MPC-HC completely and reinstalled and as soon as I selected MadVR as the renderer it played once and once only with those 3 options ticked. After that it locks up again going fullscreen on the HDTV at launch. MPC-BE with LAV filters and everything set the same as in MPC-HC works perfectly. I don't know what that tells me really though. I still get dropped frames on VFR so it does not fix that issue.

EDIT2: OK simply by switching to MPC-BE has solved a lot of issues. Not all of them. But most. I can now play with DX11 unticked. I can also enable delay playback and I get a lot less dropped frames on VFR sources. I still have to untick 'present several frames in advance' on the exclusive section though.

zmaster

12th September 2016, 00:00

Hello.
Google search not help with questions, so I posted here.
What is the "activate SuperRes filter" in chroma upscaling section? How to use it?
Now I use nnedi32+spline64 (upscaling 720p - 1080p for example), I do not see any difference "activate SuperRes filter" enabled\disabled. This filter works\need only when SuperRes activated in upscaling refinement section? Or what?

Asmodian

12th September 2016, 00:16

It does SuperRes to the chroma when upscaling it. To use it check the box. The upscaling refinement option does SuperRes to both the luma and chroma (actually RGB) after the chroma has been upscaled and the video converted to RGB. When used on the chroma only it is a subtle effect, especially when starting from such a sharp upscaling option.

zmaster

12th September 2016, 00:34

Explain pls: using SuperRes in both sections (SuperRes filter in chroma upscaling section + SuperRes in upscaling refinement) it make sense? thx in advance.

Asmodian

12th September 2016, 00:43

Yes, there are two independent upscaling steps needed, one to get the chroma to match the luma and one to scale the image to the display. SuperRes can be used on both.

Check out the flow chart in the second post, it might help make sense of the steps.

P.J

18th September 2016, 16:37

Can you add some screenshot comparisons too?

Asmodian

18th September 2016, 16:45

No, sorry. This is not intended as a guide or to promote madVR. I simply want to describe what options are available and what they do.

e-t172

18th September 2016, 17:07

Asmodian

18th September 2016, 17:20

Would it be possible to document the smooth motion settings a bit more? I understand what the "always" mode means, but I'm confused about what "only if there would be motion judder without it" means (what's the criteria, exactly? would it match 30p@60Hz? 24p@60Hz? 23.976p@60Hz? I have no idea), and I'm also confused about what "... or if the display refresh rate is an exact multiple of the movie frame rate" means (what's the meaning of the "... or"? And how is that different from "always"?). Can someone please explain?

Sure, but I sadly don't have any data on what the criteria is, exactly.

"only if there would be motion judder without it" will activate for 23.976, 24, or 25 fps on a 60Hz display but not for 29.976 on a 60 Hz display or 23.976 fps on a 143.99 Hz display. So it only activates when there would be large motion judders, something like double digit percentages for frame time variance.

"... or if the display refresh rate is an exact multiple of the movie frame rate", the "..." means it is an extension of the previous phrase. It would make just as much sense, or even more, if the "... or" was removed along with the "..." from the end of "only if there would be motion judder without it...". In practice this appears to be synonymous with "always" but if we ever got a perfect video clock this may change.

Did this tell you anything? I think that is all I know. :o

huhn

18th September 2016, 18:45

it is very obvious to me how it works. if you understand what judder is.

"... or if the display refresh rate is an exact multiple of the movie frame rate", the "..." means it is an extension of the previous phrase. It would make just as much sense, or even more, if the "... or" was removed along with the "..." from the end of "only if there would be motion judder without it...". In practice this appears to be synonymous with "always" but if we ever got a perfect video clock this may change.

it will not trigger with a 23p source at 23 hz always is even used there.

Asmodian

18th September 2016, 18:59

it will not trigger with a 23p source at 23 hz always is even used there.

Ah, you are correct. Even with my fairly inaccurate 23.9701Hz. Does that mean it doesn't trigger only when fps = refresh rate? I know smooth motion will still activate for 23.976 fps video on a well tuned 119.88Hz. :p

huhn

19th September 2016, 07:45

i don't think madVR looks the inaccurate refreshrate to enable SM.

if SM is used this is fixed too.

"... or if the display refresh rate is an exact multiple of the movie frame rate"

SM is not used with this option if the refresh rate is very close to the source.

23p source at 24 HZ does not really have a judder so the always option is needed here to get SM (don't do this...).

I know smooth motion will still activate for 23.976 fps video on a well tuned 119.88Hz.
make sense with:
"... or if the display refresh rate is an exact multiple of the movie frame rate"

Asmodian

19th September 2016, 07:55

make sense with:

Shouldn't SM deactivate with a 23.976 source and a 119.880 Hz display when using "... or if the display refresh rate is an exact multiple of the movie frame rate"? 119.880 is an exact multiple (5) of 23.976.

huhn

20th September 2016, 12:33

Asmodian

20th September 2016, 16:18

no clearly not.

"enable smooth motion frame rate conversation only if there would be motion judder without it or if the display refresh rate is an exact multiple of the movie frame rate."

the only sightly misleading thing is "exact" because it doesn't have to be exact.

that why this option is there so it will trigger with refresh rates that are a multiply of the source or very very close to it. for a 23p source 120 hz is close enough to trigger this option.

my english is pretty bad but this is totally clear to me.

You are right about the english, I think I have always added a "not" in my head when reading that line. :o

So this option is basically "always, unless the refresh rate equals the source frame rate"? I think the "exact" threw me off.

However, why does it work this way? Is there a reason to want smooth motion to trigger when the refresh rate is an exact multiple of the source? Is this for slightly off refresh rates and because smooth motion blurs too much when used at source frame rate?

huhn

20th September 2016, 20:16

So this option is basically "always, unless the refresh rate equals the source frame rate"? I think the "exact" threw me off.

pretty much.

i guess it is there to remove the 1 frame drop/repeat every "what ever".

Shiandow

20th September 2016, 20:22

Would almost make more sense to just define some acceptable frequency of dropped / repeated frames, say once every hour / day / year.

huhn

20th September 2016, 20:31

but when are you going to calculate that?

BTW i'm using this option. at 60 HZ SM is fine for me anyway.

SweetLow

24th September 2016, 13:20

"enable smooth motion frame rate conversation only if there would be motion judder without it or if the display refresh rate is an exact multiple of the movie frame rate."

I understand this two options exactly the same.
But what about "always"?
Now i suppose - in this mode madVR ignore source rate (and simply construct frames with target rate). So we never get dropped/repeated frame.
Is this true?

huhn

24th September 2016, 14:14

that's true.

Asmodian

3rd November 2016, 21:36

Updated formatting and some descriptions and added that black bar detection doesn't work with DXVA Native.

huhn

3rd November 2016, 21:54

shifting the image does works with DXVA so hmm.

Asmodian

4th November 2016, 01:09

I will do some testing and update it to be more specific, thanks.

Edit: Which image shifting did you test? I cannot get anything in "automatically detect hard coded black bars" to function when using DXVA2 Native, the always shift the image option doesn't have an effect. Even the hidden option ShowBlackBars doesn't do anything when using DXVA2 Native, it looks like detect hard coded black bars doesn't run at all when using DXVA2 Native.

huhn

4th November 2016, 08:57

a screen should be better:
https://abload.de/img/imageshiftqcs9f.jpg

and yes hard coded bars are not detected.

"cleanup image boarders by..." and "if there are big black bars:" works with DXVA native too. i even forgot they exist.

maybe something like: most black bar detection features doesn't work with DXVA native(DXVA copy back is recommended) in a proper english is a enough or it is getting complicated in the end.

SweetLow

4th November 2016, 13:37

shifting the image does works with DXVA so hmm.
Of course. Basic logic tell us - this option have nothing related to black bars _detection_ and fully independent of this.
And I don't understand, why madshi put this option in group of "detect hardcoded bars" (and disable image shifting when detect option disabled). May be ask him in main thread?

huhn

4th November 2016, 14:36

they where added at about the same time and he said before that he wants to rework/rearrange all these option/menus at one point and i'm pretty sure he looks over this in thread from time to time anyway.

but the place is not that far of. remove the words "hard coded" and it makes total sense but a lot of thy other option could be confusion. in the end this is about black bars.

madshi had problem with place on settings pages too and the first thing that comes up to my mind was "just add tabs" but in the end i'm pretty sure creating these menus is not fun at all and creating new algorithms feels more rewarding and make more fun.

i will come up on these topic when he is actively working on these options.

SweetLow

4th November 2016, 15:36

i will come up on these topic when he is actively working on these options.
Ok. But i prefer change this in one simple combobox: "Image position on Screen" with 3 values "Top","Center" and "Bottom"

huhn

4th November 2016, 16:15

if hard coded black bars are not the problem them the player should do that in my eyes.

i'm pretty sure mpc-hc can only do this manually maybe another play can do that.

what stops you from using this current option?

off is centered.

Asmodian

4th November 2016, 18:14

Asmodian

20th November 2016, 01:02

Updated for madVR v0.91.0. Next Generation Upscaling (NGU)!

Asmodian

15th December 2016, 05:45

Updated for madVR v0.91.4. Big changes to the image doubling settings, new "let madVR decide" settings which I do recommend and use myself.

edit: I forgot to upload my new settings.bin, now corrected. :o

Hrxn

18th December 2016, 15:25

BTW, what happens if you use such a settings.bin? Usually the preferences are stored in the registry, do they get overwritten?
Or works it like this: If settings.bin present, load settings stored there, if not found, use registry values?

sneaker_ger

18th December 2016, 15:31

IIRC: both have a value to store how recent they are. The most recent one overwrites the other (or create the other if does not exist).

Asmodian

18th December 2016, 20:32

Yes, for this reason resetting madVR to defaults before copying in a settings.bin is recommended.

Kitty

2nd January 2017, 03:36

As explained "Image Upscaling" is how madVR scales to the target resolution if it is larger than the source. So my question is, what's the point of "Chroma Upscaling"? It's explained as, how madVR doubles the chroma resolution in both dimensions. There's no need for it if "Image Upscaling" does the same. So does it serve an alternate purpose, it'd be greatly appreciated if someone would explain.

Asmodian

2nd January 2017, 04:14

Chroma is stored at half the resolution of luma in consumer formats (called 4:2:0 sampling, found on DVD, blurry, streaming, etc.) so you first need to scale the chroma to match the luma before the conversion to RGB. You often want to use a different method for this compared to scaling both the chroma and luma together, as is done by image upscaling.

Referring to the image in the third post might help you understand the steps needed. :)

Kitty

2nd January 2017, 06:49

Asmodian

3rd January 2017, 08:30

Kitty

5th January 2017, 00:36

I seemed to have forgot to ask about this before, but I'd like to know the proper procedure for the "Scaling Algorithms." I use NGU and I was wondering if I should max out the "Chroma Upscaling" all the way before touching the "Image Upscaling" or should I move them together. Basically what I'm asking, is if the "Chroma Upscaling" more important to image quality than "Image Upscaling."

So I have a couple examples and I'd like to know which is optimal:

"Chroma Upscaling" = "NGU - very high"
"Image Upscaling" = "NGU - Algorithm Quality -> low"

"Chroma Upscaling" = "NGU - medium"
"Image Upscaling" = "NGU - Algorithm Quality -> high"

Which option is the better option of the two?

huhn

5th January 2017, 01:19

Kitty

5th January 2017, 03:19

chroma is very unimportant.

you should always judge the picture quality by your self.
increasing the scaling algorithm until the GPU is at his max is not a proper way to get a good result.

NGU for example is not good for all types of sources.

Don't get me wrong, I understand that luma is much more important than chroma, however I just thought that since we had to upscale chroma to RGB first that it might hold some importance. Also, I'm probably not maxing out my GPU in the sense that you are thinking. I press "CTRL + J" to get the OSD and I look at the frame time to make sure I'm a couple of milliseconds lower.

I primarily watch animated content, such as anime, and NGU in my testing is perfectly fine with it. However, the reason I asked about what was better between my options, which you didn't answer btw :P, was because in my testing I noticed little to no difference between "Chroma Upscaling - low" and "Chroma Upscaling - very high" even when looking an inch away from my monitor and was just wondering theoretically if it mattered more.

After some current testing I think you're right in saying that chroma is less important. I kept the "Chroma Upscaling" at "very high" and just modified the "Image Upscaling" algorithms option. I noticed more of a difference between the options than when I was modifying the "Chroma Upscaling."

So in conclusion, I'm pretty sure that the second set of settings that I proposed in my original question would be the better choice. Producing more noticeable practical results. If anyone else wants to add anything, if I might be missing something, then go right ahead. :)

Asmodian

5th January 2017, 05:39

You are correct, go with your second option. Chroma Upscaling is more important than the chroma algorithm quality in image upscaling but it is still not something to spend too much GPU power on.

I like NGU med or high for chroma upscaling but only if the luma quality in image upscaling is already on very high. :)

NGU is great for HD anime but it is too sharp for SD anime, IMO, it shows all the artifacts and oddities too well. NNEDI3 does better with SD anime.

Asmodian

8th January 2017, 01:46

Added some comments on the "trade quality for performance options" (https://forum.doom9.org/showthread.php?p=1709814#post1709814).

Asmodian

9th January 2017, 03:28

Updated for madVR v0.91.5.

NGU can do x4 in one step now (when chroma quality = "normal"). It isn't as sharp and exaggerates artifacts much less than two x2 steps. :)

No other changes.

Warner306

31st January 2017, 01:32

New flowchart, people were often confused by the old one so maybe this one will be better understood. I think I was trying to pack too much into it, now it is really only an example using NGU for image upscaling.

Old:
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20Chart_zpsj6tljmro.png

New:
http://www.mediafire.com/file/kp43u45i8a3ayqt/madVR_Chart_v2.png

According to this, chroma is downscaled after conversion to RGB. If that is true, why doesn't the madVR OSD report this?

http://i1357.photobucket.com/albums/q760/Warner306/madVR%20Guide/NGU-image--doubling_zpsitugr0yk.png

Shouldn't it be Bicubic60 AR < SSIM1D100 AR?

I admit, I don't know this at this stage as a user.

huhn

31st January 2017, 02:14

chroma and luma are totally different things.
and chroma is not downscaled only luma at one point to match the chroma/target resolution.

Asmodian

31st January 2017, 02:19

According to this, chroma is downscaled after conversion to RGB. If that is true, why doesn't the madVR OSD report this?

http://i1357.photobucket.com/albums/q760/Warner306/madVR%20Guide/NGU-image--doubling_zpsitugr0yk.png

Shouldn't it be Bicubic60 AR < SSIM1D100 AR?

I admit, I don't know this at this stage as a user.

Actually it can work both ways, depending on the trade quality for performance option "scale chroma separately, if it saves performance". My chart shows as it would work with this disabled. It does default to enabled.

The OSD doesn't actually ever show downscaling for chroma. Using rendering times and a heavy chroma doubler it is easy to see chroma doubling or not when toggling the trade quality for performance option but the OSD chroma line doesn't change. If both luma and chroma are set to the same doubler the OSD matches my chart and you can see chroma doubling activate and deactivate when toggling "scale chroma separately, if it saves performance"; the OSD line changes from "image" to "luma" when enabled but it still doesn't indicate what happened to the chroma (upscaled, downscaled, or how).

"scale chroma separately, if it saves performance" Disabled:
https://www.mediafire.com/?6rzcyppdks1tk22
"scale chroma separately, if it saves performance" Enabled:
http://www.mediafire.com/?u5hz2usl4hkkx99

I wanted to show chroma doubling enabled to indicate where the performance hit comes from and this is more representative of how it works when doubling without downscaling a lot afterwards. It is hard to come up with a good chart that explains enough without being hard to understand. :o

mboxer511

31st January 2017, 05:16

Great post. This was really helpful as was your guide you posted on wiki.jriver.com which led me to this forum. There were a couple questions/corrections/guide mistakes I wanted to communicate with you:

Smooth Motion -

I believe you might have made a mistake/typo in your description about smooth motion in this line:

"Note that smooth motion has less blurring when displaying 23.976 fps at 60 Hz than 23.976 fps at 23.970 Hz so this option..."

I think you meant 29 fps at 60 Hz (or something to that effect).

A suggestion I had (because this really helped me when I read this in you guide) is related to:

enable automatic full screen exclusive mode -

In your guide you suggest turning this off which is actually very helpful as it allowed me to use the full media bar in full screen when hover over it with a mouse. It also removes any delay between switching between full and windowed. Perhaps you might want to mention that to some effect in your guide under the description of that mode.

Finally I had a question and was hoping you could explain what exactly "activate 200% super sampling" in image upscaling does? I have a decent rig (Nvidea GTX 960 GPU, i5 6600k OC @ 4.6Ghz, 16GB of RAM OC @ 3000MHz on a MSI Z170 M5 Motherboard) and yet saw a severe drop in performance when enabling that feature with no quality improvement. It was actually one of the rare times any setting change effected my video playback performance.

Asmodian

31st January 2017, 05:43

Great post. This was really helpful as was your guide you posted on wiki.jriver.com which led me to this forum. There were a couple questions/corrections/guide mistakes I wanted to communicate with you:

I am glad it helped.. but I don't remember posting anything on that wiki. I am glad whoever did thought it was valuable too. ;)

Smooth Motion -

I believe you might have made a mistake/typo in your description about smooth motion in this line:

I think you meant 29 fps at 60 Hz (or something to that effect).

I didn't make a typo actually, the use of the same source frame rate for both examples was very deliberate. You don't want to use smooth motion when the source frame rate almost equals the display's refresh rate but it is great when the display's refresh rate is much higher. :)

edit: That smooth motion works better for 24p into 60p than 23.976p into 24p is somewhat counter intuitive, and was the point of that example. I will change the rates to these simpler ones make this point better, even if 24p video is quite rare. ;)
edit2: Hmm, I don't like using different source frame rates in that example for some reason, hopefully just using 24 Hz will help.

A suggestion I had (because this really helped me when I read this in you guide) is related to:

enable automatic full screen exclusive mode -

In your guide you suggest turning this off which is actually very helpful as it allowed me to use the full media bar in full screen when hover over it with a mouse. It also removes any delay between switching between full and windowed. Perhaps you might want to mention that to some effect in your guide under the description of that mode.

I do suggest turning off exclusive mode, and I don't use it myself. I updated the description to stress the downsides a bit more. These posts aren't meant to be a guide but I do want people to understand the consequences of each option.

Finally I had a question and was hoping you could explain what exactly "activate 200% super sampling" in image upscaling does? I have a decent rig (Nvidea GTX 960 GPU, i5 6600k OC @ 4.6Ghz, 16GB of RAM OC @ 3000MHz on a MSI Z170 M5 Motherboard) and yet saw a severe drop in performance when enabling that feature with no quality improvement. It was actually one of the rare times any setting change effected my video playback performance.

"activate 200% super sampling" doubles the image all the time, even if it doesn't need upscaling at all. Normally doubling will only trigger if it needs upscaling 1.2x or more.

For 1080p video even on a 1080p display this means that the video will be upscaled to 4K with the doubling algorithm, any image enhancements applied, and then downscaled to 1080. My Titan XP can barely do that with NGU very high.

Personally I don't find 200% supersampling very useful, it certainly isn't useful all the time. The only situation I would think might be useful was blurry but high res video you wanted to sharpen a lot. I tend to prefer sharpening done at higher resolutions and then downscaling but not everyone agrees with that.

edit: Thanks! I missed a description for 200% super sampling. :o

cyber201

19th February 2017, 20:14

More Advanced Topics

My settings.bin (https://www.mediafire.com/?2c118c36s25w554) (v0.91.5), optimized for the Nvidia Titan X (Pascal), Windows 10, and up to a 4K display. Profiles for SD, 720, 1080, UHD, and above. Simply restore madVR to default settings and place it in the madVR program directory. madVR should also have write access to its directory (e.g. not in "program files"). :)

Hi, thanks for the config file.
My setup is Windows 10 + GTX1070 + i5 KabyLake + 16Gb DDR4 + TV UHD Samsung KS8000 + Denon X3300W.

I have to do a modification of the upscaling of 4K and 1080 because the video card reach 70°c and the GPU fans make too much noise.

Can you give me your support to configure at the best the settings for the 1080p and 4k profiles?

Thanks a lot

Asmodian

21st February 2017, 16:49

If temperature is a problem you might want to simply use the defaults, the high quality algorithms in madVR do use a lot of power. My Titan XP also runs hot with madVR, though it is water cooled so never goes above 40C.

That said, I would simply turn down the NGU quality for both chroma and image upscaling to reduce the power requirements. Also change the dithering to ordered for both profiles.

Edit: I should have mentioned that Jinc is a very nice all purpose algorithm. It is a great option if you want to keep temps down with a high end GPU while offering a significant improvement over the defaults.

Asmodian

19th March 2017, 05:58

Finally updated for madVR v0.91.7, sorry for the delay.

NGU Anti-aliasing, NGU Sharp quadrupling for all quality levels, and a new image upscaling settings page. :)

I have also updated my settings.bin file.

cyber201

19th March 2017, 14:56

Thanks a lot....
You are great

Kitty

21st March 2017, 07:12

I'm using MPC-BE, however I was curious what the "Vsync" option does and how it ties in with MadVR? Does it have any affect, since the player is using MadVR instead of it's internal renderer?

Also, if I'm using ReClock, how does it's "Vsync Correction" affect MadVR?

Asmodian

23rd March 2017, 19:43

Sorry, I do not use MPC-BE at all and do not know how it's Vsync options interact with madVR.

I also don't use ReClock anymore, madVR's smooth motion is a better option for handling a mismatched Vsync. madVR does not seem to notice ReClock's Vsync correction so do not use it at the same time as smooth motion.

Kitty

24th March 2017, 00:15

Sorry, I do not use MPC-BE at all and do not know how it's Vsync options interact with madVR.
Well, MPC-BE is just a fork of MPC-HC, but I don't know if you use MPC-HC either.

By the way, what media player are you using, and if applicable, are you using a separate audio renderer or just the internal one?

Asmodian

24th March 2017, 20:48

I use Zoom Player Max and its internal audio renderer.

Asmodian

16th April 2017, 23:17

Updated for madVR v0.91.8. NGU Soft and Standard got implemented. :D
Also improved quality for NGU Sharp 4x quadrupling at very high.

cyber201

19th April 2017, 16:43

Yeahhh
Thanks a lot

cyber201

24th April 2017, 08:22

Hi Asmodian, thanks for your work.
One question. How can I apply the best setup for this two parameters?

CPU queue size: [16] The buffer in system memory for decoded video and xySubFilter subtitles.
GPU queue size: [8] The buffer in the graphics memory for madVR’s rendering.

I have a i5-7600K CPU and a GTX 1070 for GPU

With your setitings, sometimes I have 5/10 dropped or represented frames. Why?
I have setup the medium quality for chroma ad upscaling for 1080 profile and low quality for 4K. But the problem of the dropped frames is the same.

Thanks a lot for your support.

Asmodian

25th April 2017, 00:38

Normally some frames are dropped or have presentation glitches during seaking and startup, is this what you are seeing or do you get them during normal playback?

Queue sizes are a trade-off between responsiveness and stability, I tend to go for smaller queues like 12 CPU, 6 GPU and only 3 frames presented in advance for max responsiveness. My system is still very stable during playback, with no dropped frames or presentation glitches, but I do get a few frame drops and glitches on startup or when seaking.

I have noticed that Windowed Overlay is a little more stable than other modes, remember exclusive overwrites overlay.

cyber201

25th April 2017, 09:11

I have this problem during normal playback.
When i found this option? Windows overlay?
Madvr is setup for exclusive mode

Thanks

Asmodian

25th April 2017, 14:02

Window overlay is on the same settings page as exclusive mode.

Do you have your GPU's power setting on Adaptive? Power optimized often clocks the GPU too low.

If you are using my settings file you probably need to lower some settings, the 1070 is a lot slower then the Titan.

cyber201

25th April 2017, 23:25

Yes, I use your config file with a little downgrade in quality for the profile 1080 and 2160 due to my GTX1070.
My GPU power is set to Adaptive and now I have exclusive mode enable. For you i need to try overlay?

Thanks a lot

Asmodian

25th April 2017, 23:43

Yes, try overlay if you are still having issues. You need to turn off exclusive and DX11 for overlay to work.

cyber201

26th April 2017, 16:01

If i enable only windows overlay, the screen is all green.
Now I have enable exclusive and dx11.
With this settings, the hdr metadata and the other features of madvr are processed correctly?

And if i disable the DX11, into the video, when i press CTRL+J, madvr show 8bit winodowed mode instead 10bit

Asmodian

27th April 2017, 00:40

Ouch! Are you using native dxva2 decoding? Is it an Optimus system?

If you want 10-bit for some reason you have to use DX11 and exclusive mode.

madVR currently does not support sending HDR meta data and it doesn't matter what mode you use, HDR is always handled the same way.

nevcairiel

27th April 2017, 01:25

madVR currently does not support sending HDR meta data and it doesn't matter what mode you use, HDR is always handled the same way.

Actually, that was added in v0.91.9 (https://forum.doom9.org/showthread.php?p=1804847#post1804847). It can now send HDR metadata through two different methods, the NVIDIA API, which works on all OSes and with any rendering mode (but requires an NVIDIA GPU, obviously), and the Windows 10 API, which requires Windows 10 creators update and DX11 presentation mode.

Asmodian

27th April 2017, 01:29

Another new version! I need to stop all this traveling!

Awesome news. :)

cyber201

27th April 2017, 07:16

@Asmodian
Yes, i need 10bit because 4k HDR files are in 10bit.

huhn

27th April 2017, 13:32

that's not how it works.there are nearly no true 10 bit displays out there. and 8 bit is fine.

if you take heavy compression and dithering into account than 10 bit isn't worth much for presentation but a lot for the compression.

XMonarchY

8th May 2017, 17:13

Is it possible to take screenshots in madVR in Exclusive FullScreen Mode?

huhn

8th May 2017, 17:54

there are ways to do that but this can easily cause playback issues.

XMonarchY

9th May 2017, 23:59

there are ways to do that but this can easily cause playback issues.

As long as it does not mess with Stats Info and the issue goes away after you restart the player or even PC - I am all for it, otherwise I will have to take photos...

Asmodian

10th May 2017, 00:18

I don't know how and even if you can causing playback issues would mess with the stats info so photos might be the best option... :p

Mainge

11th May 2017, 16:18

Asmodian

20th May 2017, 22:55

Updated for madVR v0.91.10.

Added a description for a new option for the deep hack in Direct3D to enable true 24 Hz and 60 Hz modes in Windows 8.1 and 10. This option used to be always enabled but it sometimes causes the presentation queues to not fill.

nsnhd

11th June 2017, 09:06

By playing some hdr files and switch on madVR OSD, some files show "matrix BT.709, primaries BT.2020" and others show both "matrix & primaries BT.2020".
What that means and what's the difference ?

Asmodian

11th June 2017, 23:30

Matrix is the set of weights used to covert YCbCr data to RGB and primaries are the color of 100% Red, Green, and Blue. You can think of the matrix as the way to covert to RGB and primaries as the gamut of the RGB space once the YCbCr has been converted.

nsnhd

12th June 2017, 00:29

Asmodian

12th June 2017, 01:10

There are a lot of real HDR videos with BT.709 matrix, it is simply the values used to multiply Y, Cb, and Cr by (while summing the results) in order to calculate R, G, and B. Mastering happens in RGB (YCbCr data is not even viewable as a color image) and as long as you use the correct inverse matrix when converting from YCbCr as you did when converting to it you get the same RGB values back.

The primaries are the important thing, anything with BT.2020 or DCI-P3 primaries could be real HDR, but you cannot tell from just that. How it was mastered and the other metadata is also important, but the matrix used to convert to YCbCr isn't. There will be differences in the rounding errors when converting between RGB and YCbCr with a different matrix; BT.2020 is probably theoretically better for 10+ bit HDR content for some reason(?), but not noticeably so.

Asmodian

3rd July 2017, 07:15

Updated for madVR v0.91.11. I only added some extra description of the current state of HDR support in Windows 10 and a link to madshi's detailed description (https://forum.doom9.org/showthread.php?p=1810891#post1810891). I updated my settings.bin as well.

This version was mostly fixes for HDR support. :)

Asmodian

3rd July 2017, 07:49

Hello Asmodian
I have a question, which I hope is not too stupid.
I just built a new PC with GTX1070 AMP Extrem. With all that new power, I can finally use MadVR has it should and I put the highest possible settings that my GPU can take (NGU sharp very high for Luma and so on). In addition, I also put direct supersampling on always doubling and I even tried direct quadrupling.

For information, I almost only watch 1080p source on 1080p display, which is what leads to my question.

In some of your posts, I could see that you don't recommand to supersample when source matches the display.
Should I really remove supersampling option? What is the downside of Upscaling to 2160p and then downscaling to get back to 1080p? I honestly thought this was the way to go in order to get super picture quality.

I am the type to put the options on the highest level when I can, but I know it can be totally wrong to do that...

Thanks in advance for your feedback.

I am not sure why I missed your post, I am sorry for that. :(

I don't like pure supersampling, it tends to introduce the minor artifacts that are the result of any scaling technique without any benefit. At best even NGU upscaling looks basically like the source after downscaling to the source resolution again. The source was probably downscaled from a higher resolution when it was mastered after all and even NGU upscaling isn't going to improve it beyond that original high resolution master.

The only case where I might use supersampling was if I was using strong upscaling refinements and I liked the look better than simply using image enhancements.

Mainge

5th July 2017, 09:35

I am not sure why I missed your post, I am sorry for that. :(

I don't like pure supersampling, it tends to introduce the minor artifacts that are the result of any scaling technique without any benefit. At best even NGU upscaling looks basically like the source after downscaling to the source resolution again. The source was probably downscaled from a higher resolution when it was mastered after all and even NGU upscaling isn't going to improve it beyond that original high resolution master.

The only case where I might use supersampling was if I was using strong upscaling refinements and I liked the look better than simply using image enhancements.

Thank you for your answer! Everything you said makes sense, I didn't know that the source is usually downscaled from higher resolution, you learn everyday :-)
So I think I'll remove this supersampling and try to play with Image enhancement instead of Upscaling refinements.

Thanks again!

Manni

31st July 2017, 06:50

Hi Asmodian,

Great thread, thanks for this work.

I have a couple of questions:

1) I just got a 1080 TI (EVGA cs2) like yourself so I'm going to take a look at your settings.bin. I'm coming from an AMD HD7870 and all my settings were optimized for a much less powerful AMD GPU. Do you have a profile for 3D in your settings?

2) One thing I'm not sure about, I set LAV HW acceleration to none because my understanding is that in that case MadVR is handling everything, not using the GPU h/w acceleration. For a max quality result, is it what I should use, or is there no drawback using DXVA2 copyback?

3) I'm abroad at the moment and had issues when I first tested with 3D (using mk3D files made with makemkv from my 3D Blurays). AMD handles 3D automatically while nVidia needs to enable 3D in the OS. I don't want 3D on al the time, especially when playing 2D video, so what's the right combination to use in the nVidia control panel settings and in MadVR to get 3D enabled automatically when playing 3D, and disabled afterwards? I tried the option to restore the OS setting but the switch back and forth in and out of 3D wasn't working 100% of the time.

In case it matters I set the resolution to UHD to upscale 2D blurays, but not 3D blurays. My display is a JVC rs500 projector. I play most of my UHD Bluray / HDR content with my Pana UB900, so will play with HDR when the nVidia HDR bug in FSE will be fixed (as we need FSE for 10bits, the JVC supports 12bits from the input to the panels).

Until we can get legt UHD Bluray content onto HTPC, my main use of MadVR is 2D/3D bluray playback (film mostly).

Thanks!

Asmodian

31st July 2017, 07:18

Manni

31st July 2017, 08:24

1) I do not have 3D profiles configured, 3D is actually disabled in my settings.bin. I do have a 3D display but I have not used 3D in a few years and I have not tested it at all. Sorry. I might test it again if people are interested but I thought it wasn't harder to display than 1080p60?

2) DXVA2 copyback offers the same quality video as software decoding but software can sometimes handle oddities better and/or is more responsive. Personally I use DXVA2 copyback only for UHD. Playing UHD HEVC or AVC takes too much cpu power and Pascal GPU's hardware decoding is very capable.

3) Note my answer to 1; sorry but I don't have any recent experiance with 3D playback with madVR. Hopefully someone else will offer some advice. :)

You actually do not need FSE for 10-bit anymore if you enable Windows' HDR mode. That said 10-bit is not important due to madVR's high quality dithering. Have you done any tests of 10-bit v.s. 8-bit output? I have never seen any differences with dithering enabled (it should never be disabled) on any 10-bit displays I have tested it on. 10-bit or higher is very beneficial for compression and processing but not for the final display, assuming good dithering is done. If you don't have to sacrifice anything for it 10-bit is good but once you start making any sacrifices you would probably be better off using 8-bit instead.

Thanks for the quick reply.

Don't worry about 3D settings, I'll sort it out when I'm back. I don't know how it compares to 1080p60, as I almost always use 1080p23, but I do know that I had to create a specific profile to get it to work on my HD7870, as my 1080p23 settings were too taxing.

Re 10bits, I did some testing and the difference is minimal. It's more for a "peace of mind" than a true PQ aspect. I don't want to use the OS HDR setting as it means I would be in HDR all the time, which I don't want. I want to use the nVidia dynamic switching mode as soon as the driver is fixed.

I need to try FSE with the 1080 Ti, I didn't use it (except when testing HDR) with the HD7870 as my JVC needed 15sec or more to sync, especially when using OSD, so I found the improvement in PQ wasn't worth the downside. However, my HD7870 had no frame drop for 2 hours+, while the 1080 Ti has frame drops every 3-5mn (with rendering times way below 30ms in 23p), so I might have to use FSE anyway.

I'll update here when I'm back and have had some time to play with the 1080 Ti more. I only had a couple of hours to test it before going on holiday. :)

Asmodian

13th August 2017, 00:40

After acquiring a new display (LG OLED 55" C7P) I changed my settings a lot. This OLED TV does not like NGU Sharp at all, even bluray 1080p looks artificial when using it, but NGU Anti-Alias is very natural and looks great. I also do not use the ED dithering options with it; I was unable to see any differences with ED dithering, ordered dithering is faster and offers identical quality on this display.

I was surprised by the similarity calibrating this OLED TV to my plasma TV from 2010. The both dither a lot to display shadows as well, they seem to be oddly similar technologies.

settings.bin (http://www.mediafire.com/file/2c118c36s25w554/settings.bin) updated.

MariaX9

13th August 2017, 07:22

Question because I do not understand it. I have at the reduce banding artifacts setting low/medium, should i tick "dont render frames when fade in/out is detected"?

huhn

13th August 2017, 15:24

well the LG OLED screens are more known for not dithering at all than doing heavy dithering.

http://i.rtings.com/images/reviews/c7/c7-banding-large.jpg
http://i.rtings.com/images/reviews/a1e/a1e-banding-large.jpg

Asmodian

13th August 2017, 19:38

Question because I do not understand it. I have at the reduce banding artifacts setting low/medium, should i tick "dont render frames when fade in/out is detected"?

I do tick "dont render frames when fade in/out is detected". If you are using easy settings for your GPU you do get 5 more medium debanded frames during the start of a fade with it not ticked but your GPU has to render 5 extra frames every time it detects a fade. This spike in load can cause a dropped frame or presentation glitch.

well the LG OLED screens are more known for not dithering at all than doing heavy dithering.

http://i.rtings.com/images/reviews/c7/c7-banding-large.jpg
http://i.rtings.com/images/reviews/a1e/a1e-banding-large.jpg

It definitely dithers to display shadows, like a plasma, you can see it quite easily.

Banding was a huge problem when setting it up but the banding does not mean it is not dithering, only that it is bad at it. When I set it to 6 bit in madVR the display's banding goes away pretty much entirely, even with default settings, and when set to 10 bit the banding is more common (I can see more of the steps). Banding test patterns all look very different at 6, 8, or 10 bit so I am sure the display is receiving different bit depths but it seems to have no idea how to go about displaying them without banding. Rounded looks much worse all the time, each step is visible and sharper, so it is not simply rounding anywhere. How do you get visible banding with a 16-bit gradient dithered to 10-bit on a 0-95 cd/m^2 display?!? :mad:

I had to tweak contrast and brightness a lot before finding settings that minimize the banding, which helped but still didn't eliminate it completely. I also have to be careful when watching real content and blaming banding on the display, some blurays I thought were fine turned out to have a lot of banding while others do not. If high debanding in madVR fixes it I blame the source but if it doesn't change it noticeably I blame the display. For my final calibration the contrast ratio was reported as 9528804659:1 by CalMAN, which does expose banding in the source.

All together this display does look very good, the black level and color saturation are both amazing, but banding is its major weakness. :(

Manni

13th August 2017, 23:08

Just a quick update re 3D, nVidia and MadVR.

The key to get this working seems to be to enable steroscopic 3D in the nVdia drivers while in 1080p, run the nVidia 3D Setup, and then set the desktop to whichever resolution/frame rate you actually use. My default is 4K@23p. MPC-BE/LAV/MadVR switch to 1080p3DFP to play the film automatically. They don't go back to 4K23 automatically, I have to leave MyMovies for that to happen. Sometimes they don't go back to 2D either, but in that case I only have to start playing a 2D movie.

The experimental custom rs tool Madshi made available recently has done wonders for my 2D bluray playback: I went from one frame repeat every 3-5 minutes to 1 frame drop every 2 hours or more, so very happy with that.

I still have to play a bit with the settings to find what works best with my new 1080 Ti, I really like the bluray to 4K upscaling using NGU Anti-Alias.

Overall super happy with the upgrade to the 1080 Ti from my HD7870.

Thanks again for all the help.

Asmodian

14th August 2017, 17:39

Thanks for the update on 3D!

I too have been very happy with NGU Anti-Alias. :)

cyber201

23rd August 2017, 07:52

hi Asmodian, I'm waiting your settings for the new version of madvr....
Yeahhhh

Thanks
Bye

Asmodian

27th August 2017, 08:11

Finally updated for 92.2. A lot of fixes along with the amazing new custom resolution tool in display modes. :D

Also DX11 native DXVA2 decoding, no quality loss but still no IVTC or black bar detection.

I also rebuilt my settings, they are still the same settings but I reset to default with the new version and reconfigured from scratch. There were some people reporting crashes with old settings files (and even those looks to have been fixed in 92.2) and while I didn't have any issues I figured a fresh start was the safest option.

aufkrawall

27th August 2017, 15:22

madshi

27th August 2017, 17:58

FWIW, I'm not sure if the custom modes are really limited to 8bit. This will depend on the driver. So it might be totally different with Nvidia, AMD and Intel.

Manni

27th August 2017, 18:01

FWIW, I'm not sure if the custom modes are really limited to 8bit. This will depend on the driver. So it might be totally different with Nvidia, AMD and Intel.

Definitely not limited to 8bits with nVidia. I have a custom mode running for 4K23p @ 12bits 4:4:4

Asmodian

27th August 2017, 18:52

Can you set the custom mode to be 12-bit in the Nvidia control panel? I can also get 12-bit RGB to work but I have to be in a 12-bit mode already before switching to it. I do not have the option to set the custom mode to 12-bit in the control panel.

Native DXVA2 decoding doesn't show the chroma blurring when DXVA2 chroma scaling is used at the same time.
Since DXVA2 scaling seems to equal some kind of bicubic on Nvidia, it's actually not too bad afterall and dimensions better than bilinear.
There don't seem to be any other quality drawbacks with recent madVR version on Windows 10 Creators Update, no banding problems.

New DX11 decoding API btw. isn't called DXVA anymore, afaik, but D3D11VA.

I tried to capture this information, is there anywhere you think I need to add something? Thanks. :)

FWIW, I'm not sure if the custom modes are really limited to 8bit. This will depend on the driver. So it might be totally different with Nvidia, AMD and Intel.

Good point, I will edit my entry to reflect my limited knowledge of the situation. :)

huhn

27th August 2017, 20:48

Native DXVA2 decoding doesn't show the chroma blurring when DXVA2 chroma scaling is used at the same time.
Since DXVA2 scaling seems to equal some kind of bicubic on Nvidia, it's actually not too bad afterall and dimensions better than bilinear.
There don't seem to be any other quality drawbacks with recent madVR version on Windows 10 Creators Update, no banding problems.

New DX11 decoding API btw. isn't called DXVA anymore, afaik, but D3D11VA.

doing a short test:

dxva: https://abload.de/img/dxvaoeob0.png
bilinear: https://abload.de/img/bilineary8of4.png
bicubic: https://abload.de/img/bicubicz8qci.png
they are all the same frame the shift comes from the scaler.
it is as bad as ever and bilinear is doing better.

aufkrawall

27th August 2017, 21:11

wth is this example even supposed to portray? It's far off any reality.
It's completely nonsense to think chroma would be blurrier than with bilinear.

software decoding + bilinear:
https://abload.de/thumb/softwarebilj8sr6.png (http://abload.de/image.php?img=softwarebilj8sr6.png)

DXVA2 native decoding + bilinear:
https://abload.de/thumb/dxva2bil5ms6q.png (http://abload.de/image.php?img=dxva2bil5ms6q.png)

DXVA2 native decoding + DXVA2 scaling:
https://abload.de/thumb/dxva2dxva4asou.png (http://abload.de/image.php?img=dxva2dxva4asou.png)

Oh, and bilinear downscaling also gives you one aliased mess for every target which is not exactly 1/4 or 1/16 etc. of source resolution btw...

huhn

28th August 2017, 01:35

so this BD is not a real source? yes chroma doesn't matter much for this image but that doesn't change what it is doing to the chroma channel.

what this is suppose to show?
obviously that there is something terrible wrong with DXVA chroma scaling.
just to make that clear i was only using DXVA chroma scaling nothing else in that image there is a reason i add the OSD.

and it is nonesense to think it is blurrier than bilinear? it clearly is in this case...

and about banding DXVA is still using low bitdeep and is adding banding you can check this with the BW.avi. where this will be notable is a different story but still it is still terrible at it.

aufkrawall

28th August 2017, 02:22

so this BD is not a real source?
yes chroma doesn't matter much for this image but that doesn't change what it is doing to the chroma channel.

obviously that there is something terrible wrong with DXVA chroma scaling.

and it is nonesense to think it is blurrier than bilinear? it clearly is in this case...

The image is grayscale only (at least to the eye), but actually even between bicubic and bilinear chroma scaling there is a huge difference in sharpness.
So what does it prove? Only that a soft chroma scaler softens the image here. There is already lots of aliasing in the source, and of course a soft scaler has a strong influence on that.

If there was a real problem with chroma blur & DXVA scaling, my cartoon example would look like blurry garbage. But actually, both chroma and luma are way sharper than with bilinear (it's 640x368 -> 2532x1440).

I have the suspicion that DXVA2 scaling uses different sharpness levels for different source resolutions. I tried a 720p non-cartoon sample a few days back and here DXVA2 scaling behaved mostly like madVR's softcubic60.
This is not bad quality, it may be just not to your likings.

and about banding DXVA is still using low bitdeep and is adding banding you can check this with the BW.avi. where this will be notable is a different story but still it is still terrible at it.
I don't see it:
https://abload.de/thumb/dxvaxps26.png (http://abload.de/image.php?img=dxvaxps26.png) https://abload.de/thumb/softwareq4s3o.png (http://abload.de/image.php?img=softwareq4s3o.png)

huhn

28th August 2017, 05:19

my screens are Cr rendered as Y.
it's an madVR feature to look up the chroma channels directly.
so if i want to know it DXVA got fixed i just use that.

but the real issue is that DXVA is known to ruin the chroma channel on nvidia cards and in my test nothing has changed so it is still bad in my book.

and now about your "10" bit image. first of all it is pretty much a solid color so a chroma scale is not adding a lot if at all banding here. second the source has banding you can see it in the 10 bit parts all 1024 steps and this get's removed by madVRs debanding making it a really bad test pattern for banding. that fact that the 8 bit is rounded is the nail in the coffin not a fair comparison at all.

Manni

28th August 2017, 10:36

Can you set the custom mode to be 12-bit in the Nvidia control panel? I can also get 12-bit RGB to work but I have to be in a 12-bit mode already before switching to it. I do not have the option to set the custom mode to 12-bit in the control panel.

I can absolutely select 12bits in the nVidia control panel when my 2160p23 custom mode is selected. It's the one made from the EDID values, as none of the others would work for me here.

Of course I have to select "use nVidia colour settings" not "use default color settings" for this to work, but that's the same with non-custom modes so you should already have this selected.

Manni

28th August 2017, 11:43

Correction: I was still under 385.28. I updated to 385.41 and I now have the same behaviour (can't select 12 bits from custom mode).
Reverting to 385.28

aufkrawall

28th August 2017, 15:03

my screens are Cr rendered as Y.
it's an madVR feature to look up the chroma channels directly.
so if i want to know it DXVA got fixed i just use that.

Nope, I'm sticking to real world tests (except of gradient images).
Especially since we are talking about DXVA scaling for low-end GPUs (it's called compromise).

but the real issue is that DXVA is known to ruin the chroma channel on nvidia cards and in my test nothing has changed so it is still bad in my book.

You haven't told me why the blur would be bad, nor why I do not encounter it with my cartoon example, where the blur by DXVA2 decoding is easily visible. Compare with softcubic?

and now about your "10" bit image. first of all it is pretty much a solid color so a chroma scale is not adding a lot if at all banding here. second the source has banding you can see it in the 10 bit parts all 1024 steps and this get's removed by madVRs debanding making it a really bad test pattern for banding. that fact that the 8 bit is rounded is the nail in the coffin not a fair comparison at all.
Provide a test sample please.

nevcairiel

28th August 2017, 15:08

but the real issue is that DXVA is known to ruin the chroma channel on nvidia cards and in my test nothing has changed so it is still bad in my book.

That isn't an inherent property of DXVA scaling or anything, its just an artifact if you try to access the 4:2:0 "untouched", because there is no 4:2:0 texture format in D3D9, but when you use DXVA to upscale chroma (or the entire image) anyway, then this doesn't really apply. Its not like the chroma is blurred first and then upscaled, thats not how it works. The method madVR uses to try to extract 4:2:0 subsampled chroma just causes blurring.

aufkrawall

28th August 2017, 15:16

Asmodian

28th August 2017, 17:54

Correction: I was still under 385.28. I updated to 385.41 and I now have the same behaviour (can't select 12 bits from custom mode).
Reverting to 385.28

Ack! That doesn't seem like something Nvidia would change on purpose but it would also be a strange bug to randomly appear. :confused:

Thanks for the update.

huhn

28th August 2017, 19:22

That isn't an inherent property of DXVA scaling or anything, its just an artifact if you try to access the 4:2:0 "untouched", because there is no 4:2:0 texture format in D3D9, but when you use DXVA to upscale chroma (or the entire image) anyway, then this doesn't really apply. Its not like the chroma is blurred first and then upscaled, thats not how it works. The method madVR uses to try to extract 4:2:0 subsampled chroma just causes blurring.

well the image you see there is pure DXVA nothing elsed and the OSD isn't indicating that madVR is accessing the 4:2:0 image.
using DXVA native and no DXVa chroma is a whole different story.

Provide a test sample please.

take your screen deband it and the banding is gone.

if you want to see banding from DXVA image scaling take the BW.avi output it as 4:2:0 and see for yourself.

but the most important part i'm not even sure if DXVA scaling is any faster than lanczos 3 the way madVR is handling it.

aufkrawall

28th August 2017, 19:24

take your screen deband it and the banding is gone.

There is no banding for the 10 bit stripes without deband, and neither there is any for them when you convert the video to 8 bit.

huhn

28th August 2017, 19:36

"you should see 1024 vertical bar if your display supports 10 bit"

so there are not 1024 steps of banding in there (i can see them on my TV and they say they are there but...) but that's not banding because what again?

and if you really think 1024 steps dither to 8 bit is enough to hide them well go ahead and set madVR to a 10 bit internal processing.

aufkrawall

28th August 2017, 19:53

"you should see 1024 vertical bar if your display supports 10 bit"

so there are not 1024 steps of banding in there (i can see them on my TV and they say they are there but...) but that's not banding because what again?

Well, you initially mentioned a "BW.avi". Where can I find this sample if it's any better?
You said that DXVA scaling would show banding, so where do you see it?

The 1024 step video is totally enough to make EVR or Win 10 video app show banding, which is not there with madVR + DXVA.

huhn

28th August 2017, 20:18

there is clear 1024 steps of banding on this file. i can see the banding on your screenshoots. 16 bit dither to 8 bit is clearly better than 16 bit rounded to 10 bit.

the bw.avi is a file that allows you to see the used scaler.
EVR normal does a better job than madVR with DXVA scaling so that's important at least on my system: https://abload.de/img/bw04une.png
and it is bilinear right now (385.41) which is odd.

here you go: http://filehorst.de/d/cmdBtJyd
but not sure why you want it because it is not a real video.

aufkrawall

28th August 2017, 21:16

Can you tell me where do you spot banding here (deband off)?

Software decoding + bilinear:
https://abload.de/thumb/softwarebil2csbg.png (http://abload.de/image.php?img=softwarebil2csbg.png)

DXVA2 native + DXVA2 scaling:
https://abload.de/thumb/dxva2dxva2sc6zsb9.png (http://abload.de/image.php?img=dxva2dxva2sc6zsb9.png)

https://abload.de/thumb/softwarebilstatnxsam.png (http://abload.de/image.php?img=softwarebilstatnxsam.png) https://abload.de/thumb/dxva2dxva2scstatm2si7.png (http://abload.de/image.php?img=dxva2dxva2scstatm2si7.png)

The only difference I see is that bilinear is softer here, so the grain in the gradients is harder with dxva2. But I don't notice additional banding.

-> Still not noticing any "real world" problem.

huhn

29th August 2017, 11:50

there is no real banding in this example.
but you are downscaling a 10 bit source it's harder to add banding there.

nvidia 384.41:

bicubic: https://abload.de/img/bicubic6034rgk.png
DXVA: https://abload.de/img/dxva1yu99.png

the last thing i have to say about the DXVA nvidia scaler avoid it if possible...

aufkrawall

29th August 2017, 15:00

Well, it looks like softcubic. At least I don't see quality issues which would prevent me from using it over bilinear when using a lowend/integrated GPU.

huhn

29th August 2017, 18:14

it is doing the chroma scaling wrong and moving the chroma by half a subpixel to the right.

just use EVR CP in this case.

aufkrawall

29th August 2017, 18:29

Now that you say it, it indeed looks shifted and can look stupid with black lines. :(

Asmodian

1st September 2017, 09:07

It definitely dithers to display shadows, like a plasma, you can see it quite easily.

Banding was a huge problem when setting it up but the banding does not mean it is not dithering, only that it is bad at it.

I don't think what I was seeing was actually dithering to display shadows. :o

It looks like it is probably just very uneven brightness between pixels when at low brightness.
flat 5% gray (http://www.mediafire.com/file/88o4ac80fqfx4yn/20170831_232555.jpg)

However, I still don't think the banding is due to a lack of dithering. I think it is due to a much worse problem with this OLED TV not having a smooth response, e.g. there are visible steps between some 10 bit values but not between others or when zoomed in all the way on a luminance graph it would look like a slanted stair step (>90° "stairs"). Also it is not uniform spatially, on a solid 5% gray screen there are some obvious lines, squares, etc.

Sorry, not madVR related. :o

huhn

1st September 2017, 19:58

the sony, Phillips and the panasonic OLED have no real problem with banding so the issue comes down to low bit deep processing with dithering or high bit deep processing with out dithering.

don't forget an OLED needs heavy processing because it is WRGB.

Asmodian

1st September 2017, 20:55

Yes, I am simply not sure it is a problem with dithering (or lack thereof), it may be simply bad processing. My expectation is that it would be better if the banding was only because it didn't dither. The visible banding is not evenly spaced in a gradient, it might have obvious banding at only a few transitions, which are randomly spaced and might involve more than one 10-bit value. The number and placement of the bands does change when changing settings like contrast or black level.

I think each pixel/panel probably responds a little differently and needs to be tuned to the right gamma curve and this tuning isn't complete enough, maybe they only do it at 100%, 70% and 40% and assume the panel is smooth between them when it isn't. You can see it when measuring the gamma curve in very finely spaced steps, slight under/overshoots as you move up the curve (~1.2 dE2000 at most but still visible when looking at what should be a smooth gradient). If 73% is undershooting luminance by 1dE and 77% is overshooting by 1dE you get visible banding even if 75% is nearly perfect. This seems to be worse with some colors too, an orange gradient's bands are more obvious but seem to have the same cause.

I remember reading about the process of evening out pixel variations in the processing/display controller, LG was proud they had figured it out and it had a specific term which I cannot remember. :o

XMonarchY

3rd September 2017, 13:59

What is Compatibility in Display Mode Optimization section? My TV works even with very low compatibility rating after optimization. Does low compatibility mean worse image quality?

Also you stated that custom modes prevent 10bit depth unless 10-12bit is already enabled in NVidia CP, but doesn't madVR use some additional colors when 10bit is selected in madVR's Configurator?

nevcairiel

3rd September 2017, 16:55

Low compatibility means it may not work with many displays.

Asmodian

3rd September 2017, 19:34

Also you stated that custom modes prevent 10bit depth unless 10-12bit is already enabled in NVidia CP, but doesn't madVR use some additional colors when 10bit is selected in madVR's Configurator?

I am not sure what you are asking. 10-bit does allow madVR to send additional colors to the GPU but if the GPU is in an 8-bit mode the GPU drivers will convert madVR's 10-bit output. This is lower quality than simply having madVR output 8-bit in the first place.

Also these additional colors are inside the gamut, 100% RGB is the same color in 8 or 10-bit, 10-bit simply allows finer steps inside the same color gamut.

XMonarchY

4th September 2017, 15:18

I am not sure what you are asking. 10-bit does allow madVR to send additional colors to the GPU but if the GPU is in an 8-bit mode the GPU drivers will convert madVR's 10-bit output. This is lower quality than simply having madVR output 8-bit in the first place.

Also these additional colors are inside the gamut, 100% RGB is the same color in 8 or 10-bit, 10-bit simply allows finer steps inside the same color gamut.

What is likely to provide better image quality on my LCD TV:
A.
- Display set to 12bit (10bit functionality) in NVidia CP
- Default/Non-Custom Display mode is used
- madVR is set to 10bit

OR

B.
- Display set to 12bit (10bit functionality) in NVidia CP
- Custom Display mode is used
- madVR is set to 10bit

?

But I guess in the end it doesn't matter that much since madVR ED is really good.

Asmodian

4th September 2017, 20:18

It is hard to know without knowing which TV you are using. In your two tests all you are changing is whether or not you are using a custom display mode. B is better in this case.

The important point is what the TV does with 10-bit input. If it does internal processing at <10-bit sending 10-bit from madVR is usually worse, if your display does not handle 10-bit input well it might be better to have madVR dither to 8-bit because it is so much better at dithering. This is also an issue if your GPU is sending 8-bit to your display while madVR is sending 10-bit to the GPU, the GPU has to do the final dithering to 8-bit and the GPU isn't as good at it so it would be better to send it 8-bit. So make sure everything is really using 10-bit properly before setting it anywhere.

Edit: I use these gradients to test with, does madVR 10-bit with dithering look better than without? If not I would set madVR to 8-bit. Never use madVR with the dithering off, except when testing. :)
gradient-perceptual-colored-v2.1 24fps.mkv (http://www.mediafire.com/file/4dia3lixs0n9xyo/gradient-perceptual-colored-v2.1_24fps.mkv)
gradient-perceptual-v2.1 24fps.mkv (http://www.mediafire.com/file/z1f4ahg3784lfwz/gradient-perceptual-v2.1_24fps.mkv)

dvd1

29th September 2017, 00:46

activate doubling / quadrupling... good for 576 to 1080p?
how and where to activate or deactivate them in madVR?

Asmodian

29th September 2017, 09:24

Yes, I like doubling for 576 to 1080p.

To activate doubling simply pick a doubling algorithm. These include super-xbr, NGU, and NNEDI3.

The options you quoted allow you to adjust what scaling factors a doubling algorithm will activate for, if the scaling factor is below that ratio madVR will use the option selected in "if any (more) scaling needs to be done:".

Let madVR decide uses 1.2x for the ratio so by default doubling will activate for 576 to 1080p.

dvd1

29th September 2017, 14:44

https://s26.postimg.org/n38rq7qkp/egrhtrynyjmjh.png (https://postimages.org/)

all right?

dvd1

29th September 2017, 15:03

https://s26.postimg.org/x1ybqfpe1/111111.png (https://postimg.org/image/sg27i33ut/)
excuse me, explain me well the values in the red paintings
are too high. how should they be?
thank you so much

huhn

29th September 2017, 15:06

they are fine.

dvd1

29th September 2017, 15:57

sorry, I want to understand why they are good?

Asmodian

29th September 2017, 15:57

all right?

Yes, that is good. That is also the same as “let madVR decide”.

excuse me, explain me well the values in the red paintings
are too high. how should they be?
thank you so much

Your rendering times look good, no issues at all. A few dropped frames are normal during playback start and/or seeking. The reason it looks good is because 27.52 < 40.00. :)

dvd1

29th September 2017, 18:40

https://s26.postimg.org/6bt0bhq2x/Immagine33333333.png (https://postimg.org/image/478naeog5/)
maximum luminance algorithms
with rendering to the limit, but little pictures of how to make a more sharp image and lower the render
grace thousand to all, but in particular to Asmodian

dvd1

30th September 2017, 15:44

nobody knows how to help me?

pirlouy

30th September 2017, 17:44

Nobody. GL !

dvd1

30th September 2017, 19:40

why activating and deactivating upscaling refinement options change scaler algorithms?

huhn

1st October 2017, 04:48

because DXVA deinterlancing is well using DXVA so DXVA scaling is used to in this case.
DXVA processing is pretty much a lost case on nvidia anyway.

dvd1

1st October 2017, 15:35

https://s25.postimg.org/gd9lwj9vz/Isdacvfvccwdfw.png (https://postimg.org/image/ctno6q763/)

it is not DXVA

dvd1

1st October 2017, 15:44

https://s25.postimg.org/w2kqwtrcf/Immagine23345666.png (https://postimg.org/image/t8hljdp63/)

Asmodian

1st October 2017, 16:56

Sorry, I was in the process of switching continents again. :p

If you refine after every 2x the aspect ratio should be correct for the refinement so madVR scales to the correct aspect ratio after doubling instead of after quadrupling.

maximum luminance algorithms
with rendering to the limit, but little pictures of how to make a more sharp image and lower the render

To lower the rendering time I suggest you set "activate doubling / quadrupling..." to 4.0x for quadrupling. Quadrupling and then downscaling does little good and takes a lot of power.

Sharper: I suggest LumaSharpen or AdaptiveSharpen for lower quality sources. I like crispen edges and enhance detail sometimes too, but only with high quality sources. In general I use as few methods and at as low strengths as possible, none by default, I set them up specifically for a source if I think it is too soft. I almost never use "refine the image after every ~2x upscaling step."

Add grain is good for low detail sources. Sharpen edges is good for Anime.

All in my experiance, that page is a matter of taste for every source. :)

huhn

1st October 2017, 17:04

https://s25.postimg.org/gd9lwj9vz/Isdacvfvccwdfw.png

it is not DXVA

well i was reading deinterlancing.

dvd1

1st October 2017, 17:47

@Asmodian

do i have 576 to 1080p of low quality as imadVR imposed for an image with as much detail as possible?
chroma upscaling
image dowscaling
image upscaling
upscaling refiremnt

rendering

thank you so much

Asmodian

1st October 2017, 17:58

Yes, I think you have picked good settings.

Upscaling refinement:
I would never use those settings but if you like how it looks it is good. Enhance detail sharpens artifacts in low quality sources too much for me, especially every 2x and strength 3! That is too strong and too much even for high quality sources.

dvd1

1st October 2017, 18:19

Asmodian

1st October 2017, 21:17

No, sorry. It may be the language barrier but I have told you all those already.

Use my settings.bin (http://www.mediafire.com/file/2c118c36s25w554/settings.bin) if you want to use my recommendations. Turn down NGU quality if you have to and switch to ordered dithering.

70MM

1st October 2017, 21:24

I wanted to see your settings but settings.bin doesnt work for me, is there something wrong?

Asmodian

1st October 2017, 21:28

Did you reset to defaults before copying it into madVR's directory?

dvd1

1st October 2017, 22:31

for low quality source in upscaling refinement
is that okay?

https://s1.postimg.org/3jgjuqu6bz/Catturaasdffgf.png (https://postimages.org/)

huhn

1st October 2017, 22:40

if you like it yes.

Asmodian

1st October 2017, 22:53

Those are reasonable settings for lower quality content, I might use something like that.

But really, as huhn said, if you like how it looks it is correct. I change what I use for all sorts of reasons and there are always sources or even displays that look better with weird options.

Trust your eyes, especially true with that page.

dvd1

1st October 2017, 23:17

I just like that, only rendering is 38ms, is not it high?
how can I lower it without having negative reflections on the obtained images?

Asmodian

1st October 2017, 23:23

As I said:

To lower the rendering time I suggest you set "activate doubling / quadrupling..." to 4.0x for quadrupling.

nussman

1st October 2017, 23:25

i.e. set NGU chroma doubling to medium and test if you see any differences.
This whole madVR thing will not work for you if you keep on asking without testing yourself.

dvd1

1st October 2017, 23:28

As I said:
how should i do it?

Asmodian

1st October 2017, 23:35

http://i1222.photobucket.com/albums/dd496/asmodian3/quadrupling%204.0x_zpsfivf9o9t.png

nsnhd

27th October 2017, 03:09

I'm playing around with NGU options and see that there are NGU Soft/Standards/Sharp, each of them comes with qualities Low/Medium/High/Very High.
So, I wonder what exactly the differences applied for them, eg. between NGU Sharp/Low and NGU Standard/High ? I know experiments count but would like to know the differences but haven't found explanations.

Asmodian

27th October 2017, 05:15

The technical differences in the quality levels are known only to madshi but they are probably something like number of neurons used, a pure speed v.s. quality trade-off. The quality levels don't really change the sharpness, they change the precision of the predictions.

The differences between Soft, Standard, Sharp, and AA are differences in the algorithms themselves, for example Sharp is very sharp at all quality levels and AA is softer and anti-aliases well at all quality levels. Normal and Soft are simply less sharp versions of the Sharp algorithm while AA behaves slightly differently. AA does not have a direct quadrupling mode and shifts the image very slightly up and to the left. Performance is similar between them all, with the exception that direct quadrupling is faster than doubling twice.

nsnhd

27th October 2017, 05:45

Asmodian

27th October 2017, 05:59

Exactly, simply pick an algorithm you like and then tune quality to account for GPU or different resolutions.

dRumMzZ

21st November 2017, 20:04

Hello everyone,
I hope this is the most appropriate thread for my problem, if it isn't please direct me to the right place.
I'm trying to use madVR in order to play a Blu-Ray movie with the "top-and-bottom" option (I was following this tutorial (https://www.tchapin.com/how-to-watch-3d-blu-ray-movies-on-the-htc-vive-or-oculus-rift-without-having-to-re-encode-the-movie/)). I've tried using different versions and settings but nothing changes, it always plays as if madVR's settings are taking no effect.

I am using a laptop and I noticed that in my monitor's device info, the 3d formats is set to none (https://i.imgur.com/4j0pUZQ.png). Is it supposed to be like that?
Also, is the tutorial missing something or am I just not doing something right?

Thank you :)

huhn

22nd November 2017, 01:55

the totorial is for VR headset so are you using a VR headset?
madVr needs 3D BDs as a source to convert them to any other 3D format.
your end device needs to support 3D to get well 3D.

Now, go to the “External Filters” section and click the “Add Filter” button. You will want to add five filters. Make sure that each one’s priority level is set to “Prefer”, so they over-ride the default internal filters:
1. LAV Audio Decoder
2. LAV Splitter
3. LAV Splitter Source
4. LAV Video Decoder
5. madVR
Under the “general settings” section, I also tweaked a couple of things. In particular, I enabled windowed overlay, as well as the checkbox for using Direct3D 11 for presentation.
meanwhile on the internet...

dRumMzZ

22nd November 2017, 09:35

the totorial is for VR headset so are you using a VR headset?
madVr needs 3D BDs as a source to convert them to any other 3D format.
your end device needs to support 3D to get well 3D.

Yes, a rift.
Sorry if this is too stupid but, what do you mean by "support 3D", though? Is that what the "3D formats" option means?
Do I need a 3D-capable monitor? If so, doesn't that defeat the purpose of the tutorial? Also, the monitor identified in the tutorial does not seem to be capable of it, hence my confusion (regarding the "3d formats" option).

Under the “general settings” section, I also tweaked a couple of things. In particular, I enabled windowed overlay, as well as the checkbox for using Direct3D 11 for presentation.

I have also tried using these settings on and off, I can provide a screenshot later if you need.

huhn

22nd November 2017, 13:08

Yes, a rift.
Sorry if this is too stupid but, what do you mean by "support 3D", though? Is that what the "3D formats" option means?
Do I need a 3D-capable monitor? If so, doesn't that defeat the purpose of the tutorial? Also, the monitor identified in the tutorial does not seem to be capable of it, hence my confusion (regarding the "3d formats" option).
ok doesn't look complicated at all. a lot of the setting he is setting up make absolutely no sense... just saying.
your rift support Top bottom 3D and some other formats the 3D support is given kind of at least.
the firs thing you need to do is getting top bottom on the display ignore the rift for now.

deinstall these freaking codec packs and every else related to directshow.
install MPC-hc.
install madVR.
option -> output -> set madVR as the video renderer
go to internal filter now and disable all (right clock disable all filter)
install lavfilter
when you install lavfilter you need to install the software intel 3D decoder. by default this is not done. unlike the tutorial DXVA should not work or well it is simply not used. use the default lavfilter settings and you should be fine.
now you have to play a frame packed 3D file or with other words a BD remux/decrypted disc you can find a sample here: http://kodi.wiki/view/Samples .
SBS encodes will not work.
for the madVR settings try the defaults with top bottom 3D and enable 3D playback.
if you play a 3D file and make me a screen with the madVR OSD (control+j) i can see what is wrong.

after this works you have to get this somehow some way into the rift with an virtual desktop or what ever and what else you find in that guide.
I have also tried using these settings on and off, I can provide a screenshot later if you need.
it's not important it just why...

and think about a new thread in software players.
not sure if a user with a rift and madVR knowledge will look into the madVR option explained thread.

dRumMzZ

23rd November 2017, 22:47

deinstall these freaking codec packs and every else related to directshow.
install MPC-hc.
install madVR.
option -> output -> set madVR as the video renderer
go to internal filter now and disable all (right clock disable all filter)
install lavfilter
when you install lavfilter you need to install the software intel 3D decoder. by default this is not done. unlike the tutorial DXVA should not work or well it is simply not used. use the default lavfilter settings and you should be fine.

Tried following this but I was a bit lost on the Intel decoder part. What is it exactly? I see an "IntelQuickSyncDecoder.dll" inside my "MPC-HC/LAVFilters64" folder.
Now I'm getting "cannot render file", probably messed up somewhere.
Also, I noticed on the madVR's thread people complaining about Nvidia drivers messing up their setups. Could that be my main problem?

huhn

23rd November 2017, 23:21

is about HDR.

this decoder is needed: https://abload.de/img/mvcdecodermdahd.png

dRumMzZ

23rd November 2017, 23:46

is about HDR.

this decoder is needed: https://abload.de/img/mvcdecodermdahd.png

Thank you, it's finally working!
I'll have to waiit till tomorow to test it on the Rift, but MPC is now showing top and bottom.
Once again, thank you dude! :D

Asmodian

10th December 2017, 10:51

Finally updated for madVR v0.92.9, new compression artifact removal and screenshot configuration options.

I also added some lower power level versions of my settings.bin file:
--Links out of date, please use the links in the third post--

Heater is the equivalent of my old settings but now I am using a slightly lower power version I called Max Quality here. Feedback would be greatly appreciated, especially for High and Low power. My GPU downclocks when below Max Quality so I doubt the profiles for different resolutions are all at similar power usage.

Manni

10th December 2017, 11:45

@Madshi

Is MadVR able to passthrough HLG? I'm implementing HLG support on the older JVCs at the moment (testing with my 2015 rs500) and when I try to play an HLG file I get the same magenta bug as when trying to play 4K60p HDR10 files (as you might remember, I found a workaround for that which is to play 4K60p files without HDR Metadata passthrough checked). That doesn't work with HLG files.

Also the file metadata is changed. For example, when I play this LG demo file: http://4kmedia.org/lg-cymatic-jazz-hdr-hlg-uhd-4k-demo/ from the UB900, the Vertex correctly reports:
HDR BT2020 HLG BT2100 (all metadata to zero) and if I apply an HLG calibration, the picture looks fine.

In case it help, here is the captured HDR metadata in that case: 87:01:1a:5b:03:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00

If I play the same file (or the webm version from youtube), the HTPC sends ST2084 instead of HLG and I get the magenta bug. I tried to disable HDR Metadata passthrough manually in MadVR, that doesn't change anything. Of course the metadata is wrong in both case (ST2084 reported).

Any thoughts? Let me know if you'd like any further info.

nevcairiel

10th December 2017, 11:56

HLG is not supported yet. madVR doesn't currently even know if a video is HLG. Also, you're in the wrong thread. :)

Manni

10th December 2017, 12:21

HLG is not supported yet. Also, you're in the wrong thread. :)

Thanks! wrong thread indeed. Does HLG support need to be added to LAV first? If yes let me know if you need any help with testing.

nevcairiel

10th December 2017, 12:30

Once madshi expresses interest on working on HLG, LAV will have to output the information that it is in fact HLG content (unfortunately the value Microsoft defined for that has already been used by madVR for ST2084, so its not as easy as just exporting it already). Fortunately HLG doesn't need any further metadata then that.

Manni

10th December 2017, 13:05

Asmodian

15th December 2017, 04:54

Updated for madVR v0.92.10. Removed NNEDI3 and reduce compression artifacts can be run for free as part of NGU Sharp high (RCA at medium) or very high (RCA at high)!

symus35

19th December 2017, 22:17

Hello

what settings/ profile would you recommand for that :

W10 last update
Nvidia driver last update
gtx 1060 6Go
AMD FX8320
16 go RAM
Last madvr
EPson 1080p projector.

Sources : blu-ray only.

Thank you :)

Asmodian

19th December 2017, 22:24

My low power settings should be a good starting point for a 1060 6GB.

symus35

19th December 2017, 22:38

Thanks for the reply.
Will try it .

What uoscaling and doubling does it use? And refinements?

Asmodian

19th December 2017, 22:48

It uses NGU Sharp for image upscaling and NGU AA for chroma upscaling.

No refinements, I only use refinements for specific sources that need it. Some deblocking for SD profiles.

I just updated them for madVR v0.92.10 so get the new one(s) if you downloaded it a few minutes ago.

symus35

19th December 2017, 23:04

I ll try and let you know how much power it uses

symus35

20th December 2017, 19:04

Hi

just tested low power profile with a standard blu-ray.

GPU usage : 30%.

Asmodian

20th December 2017, 21:05

That seems good for a 1080p display, 1080p is much easier so you might be able to use Max Quality.

symus35

20th December 2017, 21:11

I tried these :

Chroma upscaling :

NGU Sharp High Quality

image downscaling :

SSIM 1D 100 ( scale un LL check AR filter : relaxed)

Image doubling :

NGU Sharp -> luma : high CHroma : high

Doubling : always , supersampling

And thats it ( 60% GPU) 1080p60

Are those correct?

I'm just discovering madvr in depth.i'm not familiar with it

PS : you guys dont use upscale refinements? why not? aren't those any good?

Asmodian

21st December 2017, 08:24

I wouldn't recommend super sampling for no reason. :p

Chroma Upscaling:
I use NGU AA high

Image Doubling:
I use 2x and 4x so it would only run for 540p and below on a 1080p screen. I can also see using down to 1.2x (let madVR decide) but super sampling is rarely useful and actually causes harm.

Upscaling Refinements:
Using super sampling to allow sharpening at higher resolutions before downscaling again is the one time I like super sampling. I only use these for specific sources. Some sources I don't use any, some I sharpen, and some I use soften edges. There are none I would use all the time. Feel free to set one if you like it but do not use too many at once and remember, often "less is more". An good alternative without super sampling is to use image enhancements instead, it offers the same shaders but run at native resolution for a stronger effect.

symus35

21st December 2017, 09:09

Thanks

i was under the impression that it was more efficient tu sharpen an image once upscaled before downscale. In the case of 1080p on a 1080p projector, it means doubling. or supersampling?

What is the difference anyway?

Asmodian

21st December 2017, 17:41

Super sampling (in madVR) simply means the output resolution is the same as the input resolution. Doubling is more general, anytime the resolution is doubled before down or upscaling to the output resolution.

iSeries

24th December 2017, 14:21

In the calibration tab it says one 3dlut to cover all gamuts is fine. I have a bt2020 3dlut so is it ok to use that for all? I notice reds look slightly darker with this vs a rec709 3dlut when playing bluray

huhn

24th December 2017, 14:57

if i'm not mistaken you always need a bt 709 3d lut but a lot has changed.

Asmodian

24th December 2017, 19:56

Yes it is fine but that doesn't mean it is optimal. This depends on how linear your display is within the gamut.

It sounds like your BT.2020 3DLUT has a bit more saturated red than your BT.709 one. Without measurements it is hard to know which one is more accurate but I suspect it is the BT.709 one.

When using a 3DLUT for another gamut madVR first uses pixel shader math to convert the source to the other gamut and then runs that through the 3DLUT.

iSeries

24th December 2017, 20:28

Thanks guys i've loaded 3dluts for all gamuts

Smallville

28th December 2017, 17:41

My low power settings should be a good starting point for a 1060 6GB.

HI, can you pls give-me an help how to setup my madVR v0.92.10 fresh install ? I dont have change anything only setup madVR on MPC-HC.

I have an Gtx 1050 TI 4GB OC gigabyte and connect to a 4k lg, only use to see 1080p and 720p sources.

Thx

Asmodian

29th December 2017, 01:33

My low power settings.bin would be a good place to start. It has profiles you do not need but you can simply ignore them.

Smallville

29th December 2017, 11:32

My low power settings.bin would be a good place to start. It has profiles you do not need but you can simply ignore them.

Thx, i have try your low power and i have an avergae 21ms rendering, so i try your max and reduce the image upscaling to high and chroma to high quality and have an average 32 ms, im using one 1080p file source full screen (tv show). is there anything i can do to improve or is the best my card gives ?

2 more questions pls ;)

1 - On your profiles you have the 3DLUT files setup how can i do this ? or what i setup here ?

2- on the tv side its better i select my hdmi port has pc source or leave on default ? If i choose the pc side i lost the hdr effect from tv, my tv is an LG 40UJ3634V

THx 4 all

Asmodian

31st December 2017, 03:52

Test with 720p content, it is more difficult on a 1080p TV.

1) This is a very large topic. Check out the display calibration thread (https://forum.doom9.org/showthread.php?p=1744236#post1744236). I really like Calman for calibration software but it is expensive and you need a meter. DisplayCal is a great free option but you still need a meter. The i1 Display Pro is a good cheap option, ideally combined with an i1 Pro 2.

2) I would use PC (to get 4:4:4 chroma) but I am not sure what you mean by "HDR effect". If it isn't only for HDR content then I would not want it anyway.

White Eagle

31st December 2017, 23:33

Hi.
how to create a 3DLut profile for madVR with sw CalMAN 5 for Business , Some guides, tutorials, to suggest for beginners on the subject ?

Smallville

2nd January 2018, 02:31

mytbyte

2nd January 2018, 11:00

Thx for your explanation, im thinking in buy a meter for calibration my tv and monitors etc ... Is the spyder 5 good ? I only want good colors for cinema and play games on tv and monitor.... what you think about Spyder 5 meter ?

Thx

I'd get the XRite Colormunki Display, same HW as i1 Display Pro only slower (but not dead slow, actually it's pretty normal)...however, for your PC monitor, your games and apps need to support system color management or LUTs to take advantage of the meter but it's not ideal as you may get color banding if your monitor is very off standard...it's best to just calibrate any monitor controls there are to their best ability and leave it at that...

Smallville

2nd January 2018, 14:17

I'd get the XRite Colormunki Display, same HW as i1 Display Pro only slower (but not dead slow, actually it's pretty normal)...however, for your PC monitor, your games and apps need to support system color management or LUTs to take advantage of the meter but it's not ideal as you may get color banding if your monitor is very off standard...it's best to just calibrate any monitor controls there are to their best ability and leave it at that...

Thx for your answer.

I try calibrate my monitor with controls they stay nice but.... if i can get better why not ? But i really want to calibrate my LG tv 4K to wacth movies from my pc connect do hdmi cable some time i dont have the best black color or the best ambient light ... not the best picture .... what i can use to cablibrate my TV ? The XRite Colormunki Display do the work on the tv ?

Thx

huhn

2nd January 2018, 14:22

a meter is used to read the image and without knowable and proper Tv settings you will not get proper results.

mytbyte

2nd January 2018, 15:05

Thx for your answer.

I try calibrate my monitor with controls they stay nice but.... if i can get better why not ? But i really want to calibrate my LG tv 4K to wacth movies from my pc connect do hdmi cable some time i dont have the best black color or the best ambient light ... not the best picture .... what i can use to cablibrate my TV ? The XRite Colormunki Display do the work on the tv ?

Thx

Since I forgot we are on a MadVR-related thread, yes, it is possible to use free DisplayCAL software to create a LUT to use with MadVR, the software will guide you to set basic white balance and brightness on the menus of the TV and, after many automatic measurement, it will produce a MadVR LUT...but even for that, some read of tutorials is required (there are some around)...sadly, you won't get good black from an LG unless it's an OLED, however you calibrate

Audionut

3rd January 2018, 06:47

sadly, you won't get good black from an LG unless it's an OLED, however you calibrate

To nitpick, I like to watch things in a dark room at a 35 degree viewing angle. I don't like peak nit of 120 during these viewing periods as it's simply for to bright for my eyes. Movies aren't so bad I guess, but the TV is linked to a htpc.

I find 52 nits suitable. It's a slight reduction in contrast ratio but nets a black level of 0.045 cd/m2. This black level is probably just inside the 'good' region. I hate grey more then I despise the low peak nit.

mytbyte

3rd January 2018, 09:02

@audionut: you aren't nitpicky enough, you should really drop it to 48 nits to match the cinema calibration standard...:D

Isley

3rd January 2018, 18:21

@Asmodian

Thank for compiling all this useful info. I have a question about your settings.bin file : seems the SD content I have is mostly European PAL DVDs, are the auto-select rules going to work well since PAL and NTSC dvds don't have the same resolution and framerate ?

Asmodian

3rd January 2018, 18:58

Yes, all my profiles use SD for anything equal to or below 1024x576 to account for widescreen PAL DVDs. :)

Asmodian

3rd January 2018, 19:03

The XRite Colormunki Display do the work on the tv ?

Yes, these meters work as well for TVs as for monitors. I also recommend the colormunki over the spyder but the spyder 5 is acceptable. I think the i1d3 is worth it over the colormunki but go with XRite if you can.

I find 52 nits suitable. It's a slight reduction in contrast ratio but nets a black level of 0.045 cd/m2. This black level is probably just inside the 'good' region. I hate grey more then I despise the low peak nit.

I like 100 nits (103 nits at last calibration) for my dark room viewing, 120 is a too bright in a dark room but 80 or below is less vibrant. Of course, with OLED, I get 0.000 cd/m2 blacks. :D

Audionut

4th January 2018, 01:40

@audionut: you aren't nitpicky enough, you should really drop it to 48 nits to match the cinema calibration standard...:D

A recent new GPU resulted in the iDisplay3 being removed from storage, resulted in the partner rolling her eyes with an exasperated "omg this takes ages".

That standard is exactly what I targeted when I realized I needed something to help me with the facts of my shitty display. "It's a cinema display". :rolleyes:

I like 100 nits (103 nits at last calibration) for my dark room viewing, 120 is a too bright in a dark room but 80 or below is less vibrant. Of course, with OLED, I get 0.000 cd/m2 blacks. :D

OLED is the next upgrade. I'm wondering if I'll need sunglasses for the HDR aspect (700+ nits). :cool: I think I'll need a liaison officer first to help me run it past the minister of finance.

Isley

4th January 2018, 12:25

Yes, all my profiles use SD for anything equal to or below 1024x576 to account for widescreen PAL DVDs. :)
Good thinking :)

Another question : I noticed all your profiles had deinterlacing turned off. Does it mean you do the deinterlacing with the decoder? In my case it would be MPC's internal LAV filter.

huhn

4th January 2018, 15:37

keep deinterlancing on.
when lavfilter is deonterlancing madVR will disable it automaticly so you can force film mode in madVR with no harm so if the source is telecine you can disable deinterlancing in lavfilter and you are good to go.
DXVA deinterlancing is currently not that good so using softwere deinterlancing from lavfilter is currently not a bad idea.

mytbyte

4th January 2018, 17:33

OLED is the next upgrade. I'm wondering if I'll need sunglasses for the HDR aspect (700+ nits). :cool:

I doubt it, the devil is in the highlights, not the whole picture :D

Asmodian

4th January 2018, 19:09

Another question : I noticed all your profiles had deinterlacing turned off. Does it mean you do the deinterlacing with the decoder? In my case it would be MPC's internal LAV filter.

Ouch! Are they? I will need to fix that. I don't usually watch interlaced video and I recently had a source that was flagged as interlaced but wasn't so I turned it off for that. I must have forgotten to turn it back on before saving my new settings files. :o

Usually I do want deinterlacing set to the default settings.

Asmodian

13th January 2018, 10:44

I have updated the setting.bin files.

xx2000xx

16th January 2018, 23:03

I have updated the setting.bin files.

I just saw that and a few days ago I was going to blast you with a lot of questions due to the fact we have the same television that has a lot of quirks, hard as hell to calibrate (you took out the 3d/cms I see), and "PC" mode is a complete mess on many levels.

I'm going to find a slide first because I think you might be shocked on how 2.4/BT.1888 crushes blacks even with +1 brit and 20 point IRE 5 luminance set high also, although I'm sure you calibrated and mess with settings non-stop on madvr/mpc and I bet you're coming from a plasma too but OLED is another beast.

With changes the .bin quite a bit it raises questions on top of questions I have with the prior .bin along with OLED/Nvidia/MadVR that I'll think about asking the best ones as I consider you along with Warner306 top of the food chain when it comes to knowledge on setting up something very complex trying to perfect the best possible settings.

Asmodian

17th January 2018, 03:28

xx2000xx

17th January 2018, 07:47

I am happy to answer OLED calibration settings here or even better the Calibration Thread (https://forum.doom9.org/showthread.php?t=172783) in General.

I still use the 3DLUTs myself, I simply removed them from the settings.bin because users couldn't save their settings without deleting or changing the entry for my display due to missing my 3DLUTs. I have found that the newest Calman is much better at creating 3DLUTs for madVR, it is very good now without a lot of tweaking needed. Of course it is pretty expensive, but it has been a relief because I was never quite satisfied with the behavior near black when using ArgyllCMS, Argyll is very good otherwise but it takes a lot more experimentation. I did find a significant improvement near black after switching to a calibrated i1d3 instead of using my i1pro2 (obvious in retrospect), and a calibration is lot faster too.

For my calibration 17 is crushed into 16 but otherwise it is very good, I am quite happy with it.

The biggest requirement I discovered is you have to defeat the forced low power mode, if the brightness doesn't change enough for 60s (or similar) the TV lowers brightness which completely messes up a calibration. I play a video on repeat behind madTPG to keep the TV awake. I also use a smallish pattern and very low brightness (OLED on 18 for ~100 cd/m^2 at 100% white) which keeps the other power limiters from kicking in.

edit: calibrating this OLED has been much easier than plasma was. :p

I am happy to answer OLED calibration settings here or even better the Calibration Thread (https://forum.doom9.org/showthread.php?t=172783) in General.

I still use the 3DLUTs myself, I simply removed them from the settings.bin because users couldn't save their settings without deleting or changing the entry for my display due to missing my 3DLUTs. I have found that the newest Calman is much better at creating 3DLUTs for madVR, it is very good now without a lot of tweaking needed. Of course it is pretty expensive, but it has been a relief because I was never quite satisfied with the behavior near black when using ArgyllCMS, Argyll is very good otherwise but it takes a lot more experimentation. I did find a significant improvement near black after switching to a calibrated i1d3 instead of using my i1pro2 (obvious in retrospect), and a calibration is lot faster too.

For my calibration 17 is crushed into 16 but otherwise it is very good, I am quite happy with it.

The biggest requirement I discovered is you have to defeat the forced low power mode, if the brightness doesn't change enough for 60s (or similar) the TV lowers brightness which completely messes up a calibration. I play a video on repeat behind madTPG to keep the TV awake. I also use a smallish pattern and very low brightness (OLED on 18 for ~100 cd/m^2 at 100% white) which keeps the other power limiters from kicking in.

edit: calibrating this OLED has been much easier than plasma was. :p

I don't have any experience with calibration and always tweaked D-nice's settings in the past with my Panny's as he might be one of the best calibrators in the world but nobody is leaking his OLED settings yet.

I was going to ask what you used but a few days ago I decided that because I'm inexperienced and using Ted's calibration he recommends X-Rite i1Display Pro or the latest Calman that I think can do DV too, but we are looking at $250+ when out of the box it's pretty damn great especially on warm 2 vs the plasmas which I know that is how it's intended but it was just too brown and the blues/whites bugged me to the point where I just settled on Warm 1.

However the OLED, even on 35 OLED light which is about the point where you don't get that ABL at all unless you have a static image for 1 minute and then it'll kick in. You can disable that too if you wish as there's a thread on AVS on how to do it.

I did the test pattern to hit 16 and then checked the boxes pattern but when somebody posted this: http://www.avsforum.com/forum/40-oled-technology-flat-panels-general/2579537-lg-oleds-how-enhance-near-black-detail-3.html#post46798617 - it changed a lot of people's opinions that now 2.2 was the way to go because even with their increased values in bt.1886 it still crushed it and it's a bitch to calibrate correctly on 2.4 with this set. The rtings.com guy whose calibration I didn't like cause it was too dark with settings that nobody uses like 95 contrast actually nailed it on my set.

I don't have any tools and there are only a few true things you can manually do while keeping settings pretty much standard I switched back to 2.2 which I hated before and thought skin color was better at 1886 but switched to 23 OLED, 50 bright, 85 contrast, and not adjusting the 5 point scale and now after a week I'm used to it and it's better for me to have a low OLED because I'm using this as a PC monitor 24/7.

I didn't mean to turn this into a calibration deal but after reading more the past hour I'm a bit lost on the famous TV/PC conversion that may think I was doing it wrong with reading people's comments in the calibration threads which is a lot more technical than your standard threads. Although while I've been typing this short story without asking basic questions on settings I'm now more interested in the calibration side of things and even if it's $250 it's worth it until about 8 months when my tabs from FireFox burn in :-) I keep on forgetting to turn on full screen and it seems you have to baby this more than even the plasmas so I'm using about 20 techniques I've picked up over the years trying to prevent it.

What would you recommend for me to get? Calman's software is confusing and which one to get but on top of that I think you need their $600 tool on top of the software which isn't going to happen. If Cal6 is the only one that supports DV and HLG I think and probably HDR10+ while others don't then forget about it because I'm content with SDR which will be my primary as I'm not that big into Movies unless there was black disc based media that's cheap for ripping 4k, although I should probably go through my 100 blu-ray rips that I have yet to watch which most are probably trash now due to disc rot thanks to companies putting out shitty discs and at the time nobody knew about it and they were well over a dollar back then too. It might not be worth it to get anything else really unless one of the open source people will support it but I really have to do some research after I'm content with this OLED. I'm also a 100% PC guy so recommendations from other people might not apply to me unlike your suggestions.

Only Skylake here but my 970 picks up a few though Amazon, Netflix won't, or I have to d/l some shit like edge or something and Win10 LTSB is bad enough and no linux support. Then with the war coming up they are dropping HDR to HDR10+ so LG won't be able to get it but figured the PC would pick up everything so I can use my 7.1 surround so I guess I'll have to IPT most of this stuff. Anandtech just did an article about what a mess HDR on the PC is: https://www.anandtech.com/show/12171/a-budget-home-theater-pc-setup-4k-hdr-uhd-bluray-and-more

It sucks that my plasma's PSU blew up because next year's LG OLEDS will be a lot easier to calibrate it seems: https://referencehometheater.com/2018/commentary/what-is-a-3d-lut/

Back to my point so I can wrap up this wall of text. This is the correct way right?

(madVR) PC levels (0-255) -> (GPU) Limited Range RGB (TV) Output as RGB 16-235

I use 4:4:4 chroma obviously but when switching to a video go to RGB Limited, make sure LG is set to low black, MadVR to default but also keep it 8 bit?

Rant over. I didn't even cover a lot of settings I'm curious about using the LG with Nvidia regarding settings on both using MadVR and your MadVR settings on top of those including some conflicting things you're using compared to th is guide: https://forum.kodi.tv/showthread.php?tid=259188

I'll get to them later but because my Panny blew suddenly I didn't have time to pour in the massive amount of hours of research and had to pull the trigger on BF and because my old Denon is 1080p I'm reading up on both now. Although HDMI to TV then another HDMI to the receiver works just fine and not being able to see the menu isn't a big deal because I know every single setting by heart now I've been in the market of upgrading anyways and it still seems like Denon's x3300w at minimum is the way to go. The 4000 line, even refurb from A4L is still too expensive but I need to figure out all the scaling stuff and if we're forced to use shitty smart apps then I'll have to get the new EARC to output more than DD+ instead of raw 7.1.

cr0n=0sTT88

28th January 2018, 14:21

Hi,

I have a projector JVC x500. This is a projector 4K with hdmi 1.4. Not compatible with HDR.
I have a graphic card Intel HD 4600 with hdmi 1.4. No compatible with HDR.
I use JRiver MC.

When I use the option of conversion HDR=>SDR i have a problem.

when I play a MKV 4K with Red October HQ my GPU is at 100% of load and the movie looks like a robot.
If I buy a new graphic card with more powerfull GPU is it possible solve this? GTX 1060 3GB is enough?
I have a CPU 4790K but my GPU is only a HD 4600.

thank you!!!

jmonier

28th January 2018, 14:38

Generally, 3gb is not enough for 4k.

cr0n=0sTT88

28th January 2018, 14:45

Generally, 3gb is not enough for 4k.

is better a GTX 1060 with 6GB to play the contents smoothly? :thanks:

mytbyte

28th January 2018, 14:50

cr0n=0sTT88

28th January 2018, 14:55

my experience is that a GT 1030 with 2GB works correctly connected to a UHD TV, even with 60 fps HFR HDR sources converted in real-time to SDR for which some adjustments to the number of buffers and quality options is required

thank you for the information!!!!

do you have a capture of the critical configuration of madvr for solve the problem?? i can copy it and apply for me :D

mytbyte

28th January 2018, 15:11

Can't check it now, but generally (by heart) CPU and GPU queue size: set to 8 or less, trade quality for performance: everything checked, scaling: set to DXVA (not sure if it has any effect at all, perhaps on chroma)...but if you're intending to try that on the integral HD graphics, think it's too weak and too old for anything

Asmodian

28th January 2018, 17:52

trade quality for performance: everything checked

I strongly recommend against using ALL the trade quality for performance options! Use the default ones (everything from the top down to "scale chroma separately, if it saves performance") but the others can cause issues.

cr0n=0sTT88

28th January 2018, 19:37

https://image.ibb.co/dFDZYb/SNAG_0000.jpg

this configuration is no good because the image moves like a robot.
do you have any tutorial for how to configure all the parameters?

If mdvr it works better with a good gpu finally i will buy a new graphic card.

mytbyte

28th January 2018, 20:14

cr0n=0sTT88

28th January 2018, 20:23

when i use red october standard, madvr disappear of the menu of jriver.
when this happens, the reproduction is fluid but the image quality is inferior. I will definitely have to buy a new graphics card. It does not make sense to search among thousands of options.

https://image.ibb.co/nshpzG/SNAG_0001.jpg

mytbyte

28th January 2018, 20:41

Yes, do get a new video card, although I think you'd be best off with a GTX 1050Ti 4GB unless you are playing modern games (somehow I doubt you do with the HD4600)

cr0n=0sTT88

28th January 2018, 20:43

thanks, sir. I am not a gamer. for that reason I only have my integrated card. AMD and nVidia are similar? nVidia is better?

Asmodian

28th January 2018, 21:33

Right now Nvidia is better, NGU runs unusually slowly on Polaris (AMD's 480/580). Also the 480 and 580 are good for mining so they are often hard to find right now. The 1050 Ti is a great card for madVR but there are probably a few options you can change for better performance with madVR. HDR -> SDR in high quality is relatively difficult so you need change to low quality mode with something like an HD4600:
http://i1222.photobucket.com/albums/dd496/asmodian3/madVR_LowQualityHDR_zpsqxi6i1rk.png

cr0n=0sTT88

29th January 2018, 06:32

This forum is a tutorial... You can try one thing: for HDR disable "preserve hue", maybe also "compress highlights"...but i think your GPU is generally the bottleneck, it doesn't have it's own memory, it shares (probably DDR3) memory with rest of the system...your D3D usage is maxing out...how is playback when you disable MadVR? Never mind if the picture is washed out.

I've select Bilinear and DXVA in: chroma upscaling, image downscaling, image upscaling. and despite the fact that speed improves I have the same problem. image moves like a robot.

https://image.ibb.co/nnLobw/SNAG_0005.jpg

but I disable "preserve hue" works very fine!!! thank you master!!

https://image.ibb.co/nEeT9G/SNAG_0006.jpg

thank you!!!!

PS: I'm using a nec 2K monitor for test it. I don't test it in the projector JVC 4K. I do not know if this is important.

cr0n=0sTT88

29th January 2018, 07:03

cr0n=0sTT88

29th January 2018, 08:08

one more thing... this movies have 24fps and I'm playing at 60fps. I think that should overload the gpu.
is it possible play at 24fps and use a better filters?

mytbyte

29th January 2018, 08:41

Asmodian

29th January 2018, 18:14

it works! thank you!!

1050 is enough? 1060 is better for scaling algorithms? memory of 3GB or 6GB are determinants in the results?

A 1080 Ti is better than a 1080 for scaling algorithms. :)

There are still no GPUs that can completely max out madVR, though the 1080 Ti is pretty close. A 1060 6GB is actually faster than the 3GB (it has more cores), in addition to having more memory, both of which are helpful for madVR. 3GB is less than you want for watching 4K content.

A 4GB 1050 Ti is as low as I can recommend, the higher the better.

madshi

29th January 2018, 18:35

though the 1080 Ti is pretty close.
For now... :devil: There's always a fancy new algo around the corner, isn't there? (But no, nothing very soon.)

Asmodian

29th January 2018, 18:47

Your Remove Compression Artifacts algorithm is already a new one that can really put pressure on GPUs. High quality deblocking of 4K content brings my Titan X (Pascal) to its knees. ;)

It is always good to keep software ahead of the hardware. :D

madshi

29th January 2018, 18:53

Haha! :) Can the Titan X do RCA "High" quality for 4Kp60 content?

Asmodian

29th January 2018, 23:17

Not even if I skip processing the chroma channels. :)

I get ~18ms for the RCA medium step on only the luma of 4K video (25ms with the chroma). ~59 ms for RCA high on only the luma (89ms with the chroma).

Edit: in fact I cannot run RCA on 4Kp60 at all. :(

madshi

29th January 2018, 23:27

Ouch. Well, obviously the Titan X is not a very fast card... :p

cr0n=0sTT88

30th January 2018, 07:52

Well, you should switch the GPU output mode to 2160/23p if it's available for HD4600 for smooth pans...if you succeed, you can test it yourself, but I don't think 60 Hz output affects performance unless you have enabled smooth motion (and you have not, as far as i can see)...

and you really must test it on a 4K display/projector it IS very important!

I'm using JRiver and for my monitor Nec PA272W using displayport i only have this options available:

https://image.ibb.co/cCFQuR/SNAG_0000.jpg

why the option of 30hz is not available? i don't understand.

I will test it later with my 4K projector :)

A 1080 Ti is better than a 1080 for scaling algorithms. :)

There are still no GPUs that can completely max out madVR, though the 1080 Ti is pretty close. A 1060 6GB is actually faster than the 3GB (it has more cores), in addition to having more memory, both of which are helpful for madVR. 3GB is less than you want for watching 4K content.

A 4GB 1050 Ti is as low as I can recommend, the higher the better.
thank you sir!

i will buy nvidia 1060 6GB of second hand, new cards are very expensive now! (at least in spain). The last year 250€ and now 350€! :eek:

mytbyte

30th January 2018, 09:50

I don't have a clue about jriver, but it is most likely that your monitor doesn't support 23.976 or 24Hz or 30 Hz input so you probably can try to make a custom resolution/refresh and see if monitor will sync

Sent from my GM 5 Plus d using Tapatalk

el Filou

2nd February 2018, 15:41

Hello, I'm wondering if that part is still up to date:scale in linear light: [Disabled] [...] The chroma must be upscaled to match the luma before downscaling when using this option because linear light is only possible in RGB.I have enabled "scale chroma separately if it saves performance" which means it's not upscaled when I play 2160p on my 1080 display, but the OSD still shows 'LL' next to the downscaler used, and also a 'ConvertToLinearLight' step when I use ShowRenderSteps.
Is it a 'dummy step' and it actually does not use Linear Light in that case, or can you now downscale in linear light only the luma ?

Thanks.

huhn

2nd February 2018, 16:08

linear light scaling clearly works with YCbCr right now.
the OSD doesn't matter for this the image is clearly different.

xx2000xx

7th February 2018, 06:28

I strongly recommend against using ALL the trade quality for performance options! Use the default ones (everything from the top down to "scale chroma separately, if it saves performance") but the others can cause issues.

Also it's supposed to scale from the biggest hit to performance from the top to the bottom.

Sorry for the rambling earlier but I was thinking of getting into calibration and looking at the people with well over 1k worth of equipment, which I take you have too to make a proper 3DLUT, I don't think it's worth it just to make SDR better. If Calman 6 goes on sale or if I can snag it 2nd hand on EBAY then I'd consider but I probably need more than that.

My Ideal-Lume went in the trash after a 10 year run so I talked to the only other true 6500k non-diy bias lighting from the biaslighting.com guys and it's made a massive difference over the crappy $10 popular Asus one on Amazon and even the Lume. Ideal lume: http://www.cinemaquestinc.com/ideal_lume.htm - Has an LED one now too and I can't recommend a bias light enough, just for the eye strain alone. I know the Sony's, from the youtube vids through Japanese guy who might be the best in the world says it's a must for them. My LG has more of a pop and more engulfing even though it's not going to improve the black level or contrast but it's more psychosomatic for me at this point.

The rule of thumb was always 10% of the maximum brightness of your TV but just got changed a few weeks ago to 5 nits and with HDR even lower. They have 11 notches and using their youtube vid: https://www.youtube.com/watch?v=OAwrN6xiqJg - which is about as cheesy as it gets is a decent reference but I should see if I can get some good slides.

Let me rattle off a few questions concerning your LG/Nvidia/MadVR setup.

I take it you use adaptive vsync as Madshi recommended a while back?

Do you use the edge enhancement/noise reduction in Nvidia's image setting on the bottom on the control panel? I know a popular guide says to use it but I'm not sure about anything he posts really besides some decent info on what and how much affects rendering.

In the manage 3D settings I'm not sure what MadVR disregards, which I would suspect most if not all, but besides changing it to optimal power do you change anything else?

The two I'm curious about, because you actually turned down the default CPU settings a few notches which I found odd is the: maximum rendering frames and the brand new option a few patches ago that both limit the number of per-rendered CPU frames ranging from 1-4 before the GPU kicks in. Because your amazing GPU are you putting that low to give all the attention to the Vid Card? I was under the assumption that they both work totally independent.

The frames in advance along with the GPU/CPU settings has always perplexed me and I'm going to spend a few hours the new few days and if the results are quite a bit different then I'll make a big ass spread sheet breaking down all the numbers now that I flashed the bios on my MSI 970GTX (easy stuff) by 20% but still F'd by the 3.5 gigs. along with an i7 Skylake @ 4.8'ish or after I delid it tomorrow.

Last when you calibrated the LG did you find any features besides using true motion which I assume you use useful? I take it their sharpness and other features would totally conflict with MadVR's settings.

I'm 100% on a quest now to dump 2.2 for BT.1886 gama somehow using SDR that works well with MadVR but also a forget it and set while also general use at 4:4:4. Speaking of which, what's up with PC mode when it comes to putting the PC label on it, which I thought was just fine per rtings, but if you change the icon to PC also then it gets all out wack. For example 90% of techicolor's settings are grey so I suspect there's something more to it because I don't remember that on a prior firmware.

xx2000xx

7th February 2018, 07:42

For now... :devil: There's always a fancy new algo around the corner, isn't there? (But no, nothing very soon.)

HTPC would be dead and everybody would be watching videos, after signing their life over, on their fancy google player with their shi**y 5 inch "smart" phones if it wasn't for you. That or 10's of millions of more malware infected computers downloading codec packs using VLC or whatever should be extinct.

Glad to hear that you're in contact with MS/Nvidia too and if Nvidia didn't offer you a job working with the driver team, especially since they're going all in on horrendous idiot boxes that will eventually put an end to PC's as we know it in the next 5-10 years. The writing on the wall has been very clear for years now. Just don't anandtech us and say peace out and get some stuffy office job.

Nvidia merges with MS to exclusively work on the Xbox, AMD folds, game over. Intel knew what was up years ago when they started butchering their CPU's with integrated graphics. Small form factor is here, where the money is, and will only gain steam.

Once they finally got in the golden token market locked up after years and years of struggle in the housewives that has the disposable income, which they know would buy their kids the same products to completely corner the next generation which already started. Then brainwashing the public that you need to upgrade your phone every single year that makes Intel's tick/tock model a joke.

The only hope is the i386/x86 architecture, which should have been dead 10 years ago, something new will rise but I think MS has a little say on how that will go. Enjoy it while it lasts.

Asmodian

8th February 2018, 20:15

I take it you use adaptive vsync as Madshi recommended a while back?

No, I use a simple fixed refresh rate and application controlled v-sync.

Do you use the edge enhancement/noise reduction in Nvidia's image setting on the bottom on the control panel?

Absolutely not! :scared:

In the manage 3D settings I'm not sure what MadVR disregards, which I would suspect most if not all, but besides changing it to optimal power do you change anything else?

Optimal power is a bad option, use adaptive or prefer maximum performance.

The two I'm curious about, because you actually turned down the default CPU settings a few notches which I found odd is the: maximum rendering frames and the brand new option a few patches ago that both limit the number of per-rendered CPU frames ranging from 1-4 before the GPU kicks in. Because your amazing GPU are you putting that low to give all the attention to the Vid Card? I was under the assumption that they both work totally independent.

I think you are confused as to what those options mean. madVR's CPU buffers are simply buffers in system memory for decoded video frames, not pre-rendered frames. I only turn them down a bit because I use a 32-bit video player and don't need more CPU buffers, memory use can go quite high with 4K video and complex subtitles. Leave the Nvidia options on application controlled, you don't want the drivers limiting what madVR wants to do.

Last when you calibrated the LG did you find any features besides using true motion which I assume you use useful? I take it their sharpness and other features would totally conflict with MadVR's settings.

I do not use true motion, I don't like any of the motion interpolation methods. They also increase the input lag, a large negative because I also use my TV for games and desktop use.

Speaking of which, what's up with PC mode when it comes to putting the PC label on it, which I thought was just fine per rtings, but if you change the icon to PC also then it gets all out wack. For example 90% of techicolor's settings are grey so I suspect there's something more to it because I don't remember that on a prior firmware.

I definitely use PC mode, it turns off (almost) all the processing, lowers input lag, and maintains 4:4:4 sampling. Ideal for madVR as well as games/desktop. I can still fix the white point, which is the only processing I want.

cr0n=0sTT88

11th February 2018, 15:07

Hi,

I have a Nec PA272W. This a monitor with 3D-LUT programmable. I don't know how to programming my monitor yet.
Is it possible optimize MadVR for get better quality with this type of the monitor?
I will buy a GTX1060/GTX1050ti the next week, now I'm using HD4600.

Wich are the options of MadVR that it can give a better quality for my 3D-LUT monitor?
HDR only?

I have a super-monitor and a super-software and a good card (GTX1060) but I don't know how to configure it!

I have a friend with a i1Display pro, I can use it :) This is good for get better results?

huhn

11th February 2018, 15:19

cr0n=0sTT88

11th February 2018, 16:49

madVR has nothing todo with 3d lut in your screen (it's just a way to calibrate a screen) . but you can blindly assume that the 3D lut in madVR is of higher quality. but to use either you will need a colorimeter and quite some knowledge.

https://image.ibb.co/hkyWJ7/SNAG_0003.jpg
https://image.ibb.co/ciHfBS/SNAG_0004.jpg

This options are for use the internal colour CPU of 3D-LUT with madVR?

If I use these options, will not I get better quality with my Nec PA272W?

huhn

11th February 2018, 17:32

a 3d lut is a way to calibrate a display. the madVr 3d lut has nothing todo with your display 3d lut.
you can either do it with software or at the hardware in your display.

Asmodian

11th February 2018, 19:08

This options are for use the internal colour CPU of 3D-LUT with madVR?

madVR will apply it with the GPU, not the CPU, but yes. That 3DLUT is all madVR and the fact that your monitor has its own 3DLUT is irrelevant.

If I use these options, will not I get better quality with my Nec PA272W?

You will get better quality if you use some method of calibration. If you already have a very good 3DLUT in your monitor's hardware 3DLUT then madVR might be a little better but it is not going to be like the the difference between calibrated and uncalibrated.

e-t172

11th February 2018, 19:18

As to the question of whether you should calibrate using the hardware 3DLUT in your monitor or the 3DLUT in madVR, it depends.

If you use the monitor for other things besides watching videos, you might want to do the calibration in hardware so that you get a near-perfect sRGB gamut everywhere (games, web browsing, etc.). In that case you should tell madVR that your display is already calibrated to the BT.709 gamut.

If you only use madVR with that monitor, it's probably a better idea to do the calibration in madVR because it can be done in higher precision and madVR has more knowledge about the properties of the source colorspace. (I wouldn't expect that to make much of a difference though, unless you're dealing with tricky cases like a BT.2020 source.)

Whatever you do, it is a bad idea to use a 3DLUT in both madVR and your screen - stacking LUTs on top of each other will result in growing numerical inaccuracies that can lead to aberrant behaviour and especially banding. Choose a single approach and then stick to it.

Isley

18th February 2018, 22:37

Quick question : do I need to restart MPC-HC when I change a madVR setting for it to take effect?

Asmodian

19th February 2018, 06:48

No, except for the rendering method, e.g. D3D9, D3D9 overlay, or D3D11.

takenori

20th February 2018, 15:45

another question here, please:o

if i have this calibration result (factory calibrated),
https://forum.doom9.org/attachment.php?attachmentid=16238&d=1519137895
https://forum.doom9.org/attachment.php?attachmentid=16239&d=1519137911
what should i put on madvrs calibration tab? also, seeing this calibrated on hdmi, is there anychange if i connect to the monitor using displayport?

Asmodian

20th February 2018, 17:49

Attachments take a very long to to be approved, it is recommended to host images somewhere else and link them here.

Sadly your second question has an annoying "sometimes" as an answer. Usually there is no difference between different inputs but I have seen it before. Hopefully it is much less common now. Analog v.s. digital inputs were usually (always) different but often digital inputs are the same.

mytbyte

20th February 2018, 18:41

Can anyone tell me whether it is possible to passthrough BT.2020 color to TV from an SDR video (satellite broadcast files), not only HDR? Search was no help. I think it is as much a part of the standard as is BT.2020 coupled with HDR and not some marginal thing.

Asmodian

20th February 2018, 19:03

Well, you can but the TV will probably assume it is BT.709. HDR added the ability to send metadata to the TV. Before HDR most devices didn't worry about color space very much and just assumed the input was in a particular one (B7.709 for HD). Using HDR mode the data could be fully SDR and as long as the metadata is appropriate it could look correct, but you need the TV to pay attention to metadata, which they usually only do in a HDR mode.

My LG OLED C79 has a DCI-P3 mode (it assumes SD input is DCI-P3) and an "extended" BT.709 mode that I suppose assumes BT.709 input but displays it a bit stretched because who cares about standards? :p

mytbyte

20th February 2018, 19:20

How can I do it? My Tv is 2015 4K Panasonic but not HDR (very angry about it since it only requires update support for PQ transform but nooo...) and it is quite happy to recognize BT.2020 SDR from a chinese TV box...also internal sat tuners and player recognize BT.2020 and are happy to transform the gamut to its native (wider than rec709 but somewhat less than DCI-P3) so it does pay attention to metadata

I don't get the HDR option in Windows 1709, so I cannot enable it to get the metadata passthrough...

extended BT709 may be xvYCC Deep color?

takenori

21st February 2018, 01:51

mytbyte

21st February 2018, 08:29

@takenori: nothing in your pictures suggests what gamut calibration preset to choose in MadVR (if that's what you asked). Gamma is 2.2 at default anyway.

Sent from my GM 5 Plus d using Tapatalk

Asmodian

21st February 2018, 09:02

@takenori: I agree, the default seems the most likely: BT.709, pure power curve, 2.2.
Do you know if you have a "wide gamut" display?

I don't get the HDR option in Windows 1709, so I cannot enable it to get the metadata passthrough...

Ouch, I have no idea. This is why I think HDR still isn't quite ready, let alone the earliest implementations for communicating color information to displays. It is likely none of the current standards will be the HDR standard we actually end up wanting when HDR becomes ubiquitous. :(

extended BT709 may be xvYCC Deep color?

It doesn't seem that much larger, xvYCC should be 1.8 times as large as BT.709. BTW it is only xvYCC for gamut, Deep Color is another thing (10-bit). Complementary but separate technologies. :)

takenori

21st February 2018, 09:12

@takenori: nothing in your pictures suggests what gamut calibration preset to choose in MadVR (if that's what you asked). Gamma is 2.2 at default anyway.

Sent from my GM 5 Plus d using Tapatalk

@takenori: I agree, the default seems the most likely: BT.709, pure power curve, 2.2.
Do you know if you have a "wide gamut" display?

it probably is, here is madvrs identification report:
https://i.imgur.com/tQurNkE.jpg

mytbyte

21st February 2018, 09:16

It doesn't seem that much larger, xvYCC should be 1.8 times as large as BT.709. BTW it is only xvYCC for gamut, Deep Color is another thing (10-bit). Complementary but separate technologies. :)

Yes, I'm usually not that confused but it was so long ago...however, my BD player calls it Deep Color :) wider gamut is still referred to as deep color these day since the two usually do come together BT.2020 and 10-12 bit...

Regarding passthrough, I'll keep trying, perhaps I missed something, perhaps a deus ex machina (madshi) appears and takes the pain away. :D

Asmodian

21st February 2018, 09:16

it probably is, here is madvrs identification report:
Hmm.. then you really need to know what gamut it is closest too. The report shows every gray is a shade of the same color "white" with a 2.2 gamma, but we know nothing about the gamut. It isn't BT.2020.
Based on TFT Central's findings (http://www.tftcentral.co.uk/reviews/samsung_c32hg70.htm#calibration) it is in that difficult area, 127% of sRGB (BT.709) but only 93.6% of DCI-P3. Using DCI-P3 will look a little washed out but using BT.709 will look oversaturated. You need a meter to take measurements and create a 3DLUT. :o

Regarding passthrough, I'll keep trying, perhaps I missed something, perhaps a deus ex machina (madshi) appears and takes the pain away. :D
I assume you have tried madVR's "passthrough HDR content to the display?" That forces HDR mode on my TV, with HDR10 or Dolby Vision content, without using the Windows setting (it uses Nvidia's HDR API).

mytbyte

21st February 2018, 09:23

@takenori: the report only reveals that the TV recognizes BT.2020, but it's native gamut is smaller but unknown...you could manually force the TV into BT.2020 mode and set calibration to BT.2020 or use the passthrough (if it works :D) or you could measure the primaries and use that with "process HDR content by using pixel shader math"

mytbyte

21st February 2018, 09:30

I assume you have tried madVR's "passthrough HDR content to the display?" That forces HDR mode on my TV, with HDR10 or Dolby Vision content, without using the Windows setting (it uses Nvidia's HDR API).

I have tried, but perhaps I missed something, perhaps NVidia drivers need upadating...the only way the TV lets me know that it's switched to BT.2020 mode is when I manually change gamut from Standard to Native, it acts the other way around - the color in Native become less saturated while in 709 mode they become more saturated (actually deeper, pun intended) :D

takenori

24th February 2018, 10:03

Hmm.. then you really need to know what gamut it is closest too. The report shows every gray is a shade of the same color "white" with a 2.2 gamma, but we know nothing about the gamut. It isn't BT.2020.
Based on TFT Central's findings (http://www.tftcentral.co.uk/reviews/samsung_c32hg70.htm#calibration) it is in that difficult area, 127% of sRGB (BT.709) but only 93.6% of DCI-P3. Using DCI-P3 will look a little washed out but using BT.709 will look oversaturated. You need a meter to take measurements and create a 3DLUT. :o

@takenori: the report only reveals that the TV recognizes BT.2020, but it's native gamut is smaller but unknown...you could manually force the TV into BT.2020 mode and set calibration to BT.2020 or use the passthrough (if it works :D) or you could measure the primaries and use that with "process HDR content by using pixel shader math"
:thanks:
i assume that this taking measurement needs some tools?

i tried bt2020 for calibration, and its a way washed out.
for the time being, dci-p3 seems the closer i can get without buying additional tools right?

also, another question (please)
is there anyway to capture hdr screenshot from mpchc-madvr?

i tried direct save image from mpchc, but while its vibrant enough, it doesnt compare to the file being played, while microsoft suggestion to use xbox screenshot results in a very washed out image that doesnt turn to hdr image when i turn hdr on instead it turn to a doubly washed out.

Asmodian

24th February 2018, 10:08

Yes, that display really needs a meter for best results. :(

It is hard to know which would be better, BT.709 or DCI-P3. 93.6% is pretty small but 127% is quite large. Probably DCI-P3 but pick the one you like more.

I don't think there is a way to save the HDR metadata to the screenshot, similarly I don't think most image viewing programs would set the HDR metadata if it were possible to save it.

takenori

2nd March 2018, 13:42

cheers for all the help :D mayble ill look for the cheapest one later.
does calibration needs to be done on a regular basis or its a one time thing only?

another thing, i had sets madvr dislay mode as 2560x1440p120, 2560x1440p100, 2560x1440p144,
but when playing 30 and 60fps video this what i get
https://i.imgur.com/oDBCpUK.jpg
madvr only change composition rate, display still at the native rr, and and to be sure, i run them with reclock and it reported that my display still not changed to 120hz.

huhn

2nd March 2018, 13:48

windows 7?

takenori

2nd March 2018, 13:59

its win10

Asmodian

2nd March 2018, 21:37

does calibration needs to be done on a regular basis or its a one time thing only?

It is a one time thing, displays do change some over time so re-calibrating once a year or so is good, but compared to the un-calibrated state these are very minor changes.

madvr only change composition rate, display still at the native rr, and and to be sure, i run them with reclock and it reported that my display still not changed to 120hz.

Do you have G-sync enabled? G-sync and madVR do not interact well. It would do this if you had the "Preferred refresh rate ([Display Name])" set to "Highest available". Set it to "Application-controlled" if you want madVR to use a set refresh rate with G-sync enabled (but you don't want G-sync enabled...).

With G-sync enabled at all the refresh rate is a little soft and you cannot use custom timings, but you can create a setting for your player in "Program Settings" and set "Monitor Technology" to "Fixed Refresh". This helps a little but it isn't the same as disabling G-sync completely.

huhn

2nd March 2018, 22:11

win 10 composition issues are related to the windows version amd GPU driver.

if you are not at the newest version give that version a try bit may not help.

if you are using 2 screen at the same time try it with one to see if that fix it.

windows 10 composition rate is not as broken as in windows 7 but it is still problematic.

takenori

3rd March 2018, 02:20

im using nvidia and my monitor is a freesync 2. freesync set to off already

cr0n=0sTT88

10th March 2018, 23:04

When does the imageupscalling option work?
We assume that I play the movie in full screen.

If I have a movie of 600x400 and the screen resolution is fullhd, does imageupscalling work here?

If I have a fullhd movie and the screen resolution is also fullhd, does imageupscalling work here?

and the chromaupsamsung function? Does it work the same as imageupscalling?

NGU is always the best algorithm? for image/chroma upscalling I say.
I want to use MadVR with a projector and I need configure it.

My projector is JVC x500 and has the function of scaling the image input to 4K. I do not know if it is better to enter the signal to the projector in fullhd format or 4K format from windows. If I play a fullhd movie and it is upscaled to 4K via madVR, is it better for JVC? or is better play direct in FullHD from madVR and send the fullhd signal direct to JVC? JVC upscalling 4K or MadVR upscalling 4K? this is the question. What is better for the best image quality?

sorry for my poor english.

Thank you!

Asmodian

10th March 2018, 23:49

cr0n=0sTT88

11th March 2018, 15:50

Image upscaling works when the image needs to be upscaled.

Yes, 600x400 to 1920x1080 requires upscaling.

No, 1920x1080 does not require upscaling. However, you can have madVR double the resolution and then downscale it again if you enable super-sampling with one of the doubling algorithms. I don't recommend this.

Chroma upscaling is a more complicated concept. Almost all consumer video has 4:2:0 chroma subsampling which means that FHD video has the luma at 1920x1080 but both chroma layers at 960x540. To convert to RGB the chroma and luma resolutions must match and chroma upscaling is what madVR uses to do it.

NGU is very good but best is not something we believe in here. ;)

I would use madVR to upscale to 4K, assuming your GPU is strong enough, madVR's upscaling is very high quality with a lot of options to adjust it to be ideal for your eyes, projector, and screen.

Thank you very much for the information! I have a GTX 1060 card with 6GB of memory.

I currently use NGU standard with high quality for Image and chroma upscaling.

Apart from these 2 options, what other options can I use to improve the image quality as much as possible? Are these two options the most important in MadVR? Is it worth buying a more powerful graphic card to add more improvement options?

My projector has not HDR option and I use MadVR real-time conversion HDR to SDR option. I use this only in 4K videos. But the most of the videos that I see are fullhd and 4:2:0.

Thank you!

PS if I have a video with 4:4:4 subsampling, ¿chroma upscaling is it necessary too?
In the driver of nvidia I've selected RGB full range and 8 bits (for the projector JVC).

huhn

11th March 2018, 16:23

chroma is not scaled if it doesn't need scaling.

it doesn't matter what nvidia is sending your device madVR will always send RGB to the GPU driver.

takenori

7th April 2018, 13:28

So, about early next month i decided to get a calibration device for madvr, and right now im choosing either
X-Rite ColorMunki Display or X-Rite ColorMunki Smile.
so in terms of novice user, which should i get? im talking about know nothing, had to follow direction step by step novice. also another brands recommendation is welcome if theyre easier to use.

e-t172

7th April 2018, 13:49

I don't know about the ColorMunki Smile, but I can tell you the ColorMunki Display is an excellent sensor. It's the exact same hardware as the more expensive X-Rite i1 Display Pro and has the same accuracy (I have verified that by comparing measurements myself). The only difference between the ColorMunki Display and the i1 Display Pro is that the former is artificially slowed down for marketing reasons, so it takes more time to do the measurements.

Asmodian

7th April 2018, 17:23

Given that the Smile is marketed for "your monitor and iOS devices" I recommend against it but it is supported by DisplayCal (Argyllcms) and would probably offer decent results on a LCD with LED back-light. It does have fewer factory profiles, it only has options for LCD with CCFL back-light or LCD with LED back-light. Given that LCD with LED back-light is very common this may work fine for you but the ColorMunki comes with support for:
A refresh type display - The refresh period is measured, and the integration time adjusted appropriately. [CB2]
CRT (Hitachi CM2112MET, Diamond View 1772ie)
LCD CCFL IPS (CCFL AC EIZO HP with CORRECTION)
LCD CCFL Wide Gamut IPS (WG CCFL NEC241 271)
LCD RGB LED IPS (RGBLED HP SOYO)
LCD White LED IPS (WLED AC LG Samsung)
Projector (Marantz HP Panasonic Projectors Hybrid EDR)
OLED Family

If you need to save the money the Smile will work and probably give decent results but it isn't the most flexible option.

huhn

7th April 2018, 17:31

or you use the correction from the over devices.

Klaus1189

26th April 2018, 18:14

What are the differences of GPU flushed / not flushed, ...
I couldn't a find a explanation of that.

Asmodian

26th April 2018, 20:33

Not much, flushing takes time (not much) bit can help prevent problems is some situations.

Klaus1189

27th April 2018, 10:05

OK, thanks.

What is the difference between these flush types?
flush
flush and wait (sleep)
flush and wait (loop)

Asmodian

27th April 2018, 15:33

In my testing they are basically the same with the only significant difference being that without a "wait" step madVR cannot measure the rendering time.

Sleep v.s. loop seem about the same except sleep should use less power. Really I do not see any difference between them, I recommend you leave those options on default, I have not heard of changing them being helpful for at least several years. They are options to try to workaround troublesome low end hardware, if you do not have issues do not mess with them. :)

leeperry

29th April 2018, 23:20

Can we agree that low/low debanding is actually set-and-forget? Whenever I disable it PQ decreases but after more thorough testing(and better cables), I would say that deringer strength is too high and destroys details making it only useful on hopeless oversharpened SD content and apparently RCA@1 does wonders on anything not native resolution but I would have to run more tests. Long story short decent 720p30@1080p60 looks better with low/low debanding + RCA@1 but not deringer.

Shame I took deringer for granted for way too long but debanding is a strong asset for sure, unlike deposterizing filters in Avisynth that always did as much harm as good giving a clay look to human faces.....but then again it was all processed in 8bit duh ^^

Asmodian

30th April 2018, 00:25

I do agree with that, low/low has never harmed the image in anyway I could prove to myself. Also, sources with truly no banding are few and far between.

I still turn off debanding sometimes, with some UHD content. It is more that I do not see any banding and there is a small performance impact when running debanding on UHD, along with some "untouched" video placebo. :o

(untouched, except for YCbCr->RGB, NGU, 3DLUT, dithering, and whatever my TV does to it :rolleyes:).

leeperry

2nd May 2018, 12:43

I really oughta stop taking RCA & Deringer for granted because even RCA@1 fails on grainy 720p@1080p, I guess grain means that I must A/B RCA beforehand.

I see that there are hotkeys for debander and deringer but not RCA apparently?

And still no hotkey to open the mVR config panel directly huh? Only option is to call madHcCtrl.exe editLocalSettingsDontWait then I guess?

Warner306

2nd May 2018, 16:32

Oguignant

7th May 2018, 04:14

settings.bin
madVR's settings can be backed up or shared via a settings.bin file. My settings are optimized for the Nvidia 1080 Ti / Titan X (Pascal), Windows 10, and on my LG OLED55C7P 4K TV with profiles for SD, 720, 1080, UHD, and above. 3D is disabled. I use Max Quality but even the Heater settings.bin should never have dropped frames due to GPU performance if used with an overclocked 1080 Ti or better. Note that "low" still requires a decent GPU.

To use simply restore madVR to default settings and place the file in the madVR program directory. madVR will automatically create or update the file with new settings madVR if it has write access to its directory (e.g. not in "program files"). madVR will still use the registry to store these settings, along with any changes, without write access. :)

For madVR v0.92.10 or later:
Heater settings.bin (http://www.mediafire.com/file/6wpsd6w45tuotwu/madVR_HeaterSettings.zip)
Max Quality settings.bin (http://www.mediafire.com/file/aq8bevrirokiry2/madVR_MaxQuality.zip)
Low Power settings.bin (http://www.mediafire.com/file/8dhzbbpnkasij9p/madVR_LowPower.zip)

Feedback is appreciated, especially for the low power settings.

Hi, Thanks for sharing!!!

Luckily, I have the same hardware configuration: Nvidia Aorus 1080 Ti Xtreme, Windows 10, and LG OLED55C7P 4K TV. The difference is that I have an Alienware 17r4 i7-7820hk laptop with Graphics Amplifier. (Video Bus: PCIe v3.0 x16 (8.0 GT/s) @ x4 (8.0 GT/s) is my limitation)

I had already set my own settings, but the rules I apply are more complicated because I do it by directory in addition to resulocion and frames. It is more complex, but it works better for me in this way, because it allows me to have, let's say it this way: the three settings: Heater settings, Max Quality settings and Low Power settings in one configuration.

After testing the 3 options (heater, max q and low power) I got excellent performance, modify some things in my scaling settings (like 4k 30fps) based on the Heater settings. (I always use NGU sharp/very high for 4k upscaling but now i replace it with Jinc, as in your settings)

All other processing and rendering settings are the same as yours.

Thks!!!!

Asmodian

10th May 2018, 05:13

I am glad it was helpful. :)

Using folders to automatically set the profile is a good idea but my profiles were already complex, yours must be intense. ;)

NoTechi

11th May 2018, 09:08

Manni

11th May 2018, 09:34

Any hints to a madVR newb wich controls I should play with to improve HD sports. To be more precise I have a JVC7900 and watching soccer I found that on closeups everything is sharp and crisp but once the camera zooms out to show like half of the soccer field I have the impression the single players could be sharper. Could be the JVC pixel shift is just at the edge and it can't be improved but I would like to try :) If someone could share his settings he is using for HD sports would be even better :)

NoTechi

PS Soccer world cup is starting soon I need to be prepared ;)

You might want to try enabling the CMD (frame interpolation) in the JVC, set to low or high (it should be off by default). It doubles the vertical resolution, so it should help with live sports at 50/60p. I assume you're not watching recorded content, so MadVR not relevant. Don't use it with movies unless you like the SOE (soap opera effect).

If recorded (weird way to experience the World Cup!) then try the smooth picture feature in MadVR, but I never use it so I'll let Asmodian give advice.

NoTechi

11th May 2018, 13:22

Manni

11th May 2018, 13:33

Manni many thanks for your Feedback! No I don't like the SOE but you are right I never tried it on sports so far but will do.

I am just setting up madvr with live TV (DVBViewer with madvr getting streamed from a VU+ 4k). I will try smooth motion even though I am not sure if CMD combined with smooth motion is a good idea.

Both might improve the movement but I am more concerned to get the Players more sharp from a bigger distance. In other words even when they are not moving at all but the camera is zoomed out completly to show the field I would like to increase the sharpness of those "tiny players" :)

NoTechi

Don't use both CMD and Smooth Motion, it's one or the other :)

If it's not when the players move, then you might want to use some sharpening, as CMD won't help that much on static shots. I'll let Asmodian advise.

Cool to know that MadVR can work with DVBViewer.

Warner306

11th May 2018, 18:45

Asmodian

11th May 2018, 21:33

I agree with Manni and Warner306.

It is very important to not use smooth motion with any frame interpolation algorithm. Smooth motion generates blended frames when it needs to but these horribly confuse motion interpolation. Smooth motion also does soften the image some so it isn't going to help sharpness at all.

If you are already using NGU Sharp at as high a quality as you can, image enhancement is what you need. Play with one or two sharpeners, don't just click everything, more is not always better. I find good strenth values are usually between 0.5 and 1.5 but this is source and display dependent (I have no experience with projectors) so I wouldn't try to copy others' settings. Trust your eyes but also remember to keep a balance, very sharp might look good at first when studying too closely but look really bad when considering the image as a whole.

As a recommendation (take with some salt, see above):
Crispen Edges is a good option that tends to not make the image look too over sharpened and it can be combined with other sharpeners for good results at fairly low strengths.

NoTechi

12th May 2018, 12:59

Many thanks for all the hints to get me world cup ready! ;)

I am still struggling to get good results on scenes which are zoomed out. When they are not moving it's ok but when they do I get sort of "ghosting". So I might have to balance sharpness and "ghosting" free movements. Well I guess I just have to keep trying and to get a feeling for all the settings.

NoTechi

Asmodian

12th May 2018, 22:53

A sort of ghosting does not sound good!

Are you using smooth motion? Sharpening and ghosting shouldn't be related at all. :confused:

huhn

13th May 2018, 11:02

the cause of ghosting has a lot of reasons.
could be something simply as a PWM.

NoTechi

13th May 2018, 11:05

No smooth motion is off. I tried but it did not improve the "ghosting". Well "ghosting" might be the wrong term it's more like once the camera is zoomed out and the "tiny players" are moving they are not shown sharp anymore.

It might be also related to the refresh rate which I basically did not looked at yet. I get dropped and repeated frames with live tv where I don't get them with movies with same resolution so my guess is thats refresh rate related. I might need more profiles :)

It might be I am just asking too much from a 4k pixel shift projector and HD live TV but I am not done with tweaking the settings yet! Still some time to world cup ;)

NoTechi

NoTechi

13th May 2018, 11:11

the cause of ghosting has a lot of reasons.
could be something simply as a PWM.

I got a X-Rite i1Display Pro incoming next week to create a 3DLUT. Using a 3DLUT should also improve possible problems coming from PWM right?

NoTechi

huhn

13th May 2018, 11:46

no it will not change ghosting.

PWM is kind of an flicker the human eye doesn't see but which can create ghosting.

a screen with 60 hz a source and 60 hz BFI usual flickers but doesn't have ghosting problems while a 120 HZ BFI screen rarely flickers but has ghosting problem.

every type of brightness change on a static image can create this.

QBhd

13th May 2018, 13:16

I have watched these streams in the past. The problem being described is a result of the compression. There really is nothing that can be done. Think in terms of x264, the zoomed out angles of the match are just a huge monolithic green field with very small and fast moving players. No real time x264 settings are going to handle this very well. The players are encoded as blurry smudges, I don't think there is a way to make these blurry smudges any sharper.

QB

NoTechi

13th May 2018, 16:11

I made some close up pictures today from F1 once the camera was zoomed out which basically has the same effect as with soccer with small moving objects (not the best quality but you will get the point).
Link to sample pictures (https://1drv.ms/f/s!AlcgOAEVDZLkmsBoCDIOYI7-aUr7BQ)
I have to add it is not always that bad as in the pictures when zoomed out. It seems to be depending on how the small objects and the camera is moving.

In movies I never get something like this. Also live TV without small moving objects looks great.

@QB Yes the stream compression might explain it. I think it can be somewhat mitigated but it will never look really good.

NoTechi

huhn

13th May 2018, 17:54

this is bad deinterlancing or no deinterlancing.

NoTechi

13th May 2018, 19:30

I had the impression I was able to use higher settings in madvr with dxv11 instead of DXVA2 (copy-back) without dropped/repeated frames with the downside that deinterlacing is not possible in madvr. But if thats required and will improve live sports it would be ok.

Thanks Huhn will try :)

NoTechi

NoTechi

13th May 2018, 21:36

Ok with deinteralcing it is a difference as day and night! Feeling stupid that I did not saw the wood for the trees with all those possible settings. But guess thats part of the learning curve and I am more than thankfull that you all are taking the time to point a newb like me in the right direction!

I am world cup ready now :D

NoTechi

huhn

13th May 2018, 21:39

try yadif in lavfilter nvidia has some major problem with deinterlancing.

nevcairiel

13th May 2018, 21:41

NVIDIA HW deinterlacing is just fine.

huhn

13th May 2018, 21:50

did they fix they chroma issue?
edit: 388.71 still wrong aligned on my PC.

huhn

17th May 2018, 16:00

do you know if you can provide DXVA2 with the chroma sub sample position.

maybe it assumes mpeg1 without additional informations not mpeg2 sub sample position.

that would make the problem a part of EVR and madVR not nvidia even through that'S not a brilliant idea if that is the case.

bluesam20

13th July 2018, 21:07

What's good everyone,
I have been having problems with the OP's config lately. About a year ago I purchased my first 4k monitor, around that time I found this thread. I started to use the settings file posted here as a starting point and decided I really liked the results. I used it with no performance problems at all, fast forward to today and I reinstalled mpc-hc and madvr. Now I cannot play any files without major frame drops. Before I continue my specs:
Core I7 6830k @ 4Ghz
16GB ram @ 3000Mhz
GeForce GTX Titan X Pascal @ 1800Mhz~
120GB SSD for Win 10 64bit/MPC-HC and Madvr 32/64bit
I did find a couple of workarounds:
1. Simply enable FSE, I get zero frame drops except when the player switches to the mode.
2. Setup different display modes within Madvr (ex. 2160p29 etc.). This of course disables Smooth Motion, but the refresh rate now matches the video famerate so it's okay
3. My least favorite and last option if a must, reduce scaling settings.
So I'm hoping someone can give me some advice as to what might be causing the issue, I know it's probably going to be just to reduce the quality settings. However I think that's a little extreme considering it plays fine in full screen, under the same settings with FSE enabled. Thanks for any help or info.

Asmodian

13th July 2018, 22:01

It is interesting that FSE is fine.

Have you tried disabling D3D11 and using Windowed Overlay? How about not presenting a frame every vsync? Usually problems that are solved by FSE involve Windows' desktop composition in some way.

What are your average and max present and rendering times in the different modes? What happens if you disable smooth motion while still using 60 Hz (not recommended for watching, just as a test)?

Are you using my heater or max quality options?

khanmein

14th July 2018, 06:17

The FSE mode is not working again with 398.46.

Asmodian

14th July 2018, 08:18

Hopefully they fix it before the next "official" release, FSE works well in 398.36. :(

2. Setup different display modes within Madvr (ex. 2160p29 etc.). This of course disables Smooth Motion, but the refresh rate now matches the video famerate so it's okay
3. My least favorite and last option if a must, reduce scaling settings.

Thinking about this more, the fact that either no smooth motion OR FSE works you are probably right on the edge of the performance requirements. Do you have the power limit of your titan at 100%?

I think at 100% the power limit does trigger while watching some content at 4K with my "max quality" settings, even at default clocks. Not enough to drop frames when I was testing (I did test my settings with my Titan X (Pascal) at complete defaults) but it was close. I have to admit I always have mine at 120%, even if I am devolting/downclocking. If you are watching 30p especially. :o

bluesam20

14th July 2018, 09:06

Thanks for the fast response, and I'm sorry I left out some information. I use your Max Quality config, the Heater settings are a bit too much. I run with the power target at 120%, a 50Mhz offset on the core clock and a 300Mhz offset on the memory clock. I use the stock TitanX(Pascal) blower fan cooler, so things heat up quick and that 1800Mhz boost drops to around 1772Mhz.
I have tried some of the suggestions you made already, but I am testing them again right now. I disabled Smooth Motion and left the refresh at 60Hz, the results:
2160p - rendering stats
avg 5 ~ 8ms
max 8 ~ 13ms
1080p - avg 16 ~ 19ms
max 20 ~ 23ms
720p - avg 8 ~ 10ms
max 10 ~ 18ms
SD - avg 13 ~ 16ms
max 16 ~ 20ms
with Smooth Motion
2160p - avg 5 ~ 8ms
max 8 ~ 12ms
1080p - avg 18 ~ 22ms
max 20 ~ 28ms
720p - avg 8 ~12ms
max 10 ~ 23ms
SD - avg 14 ~ 17ms
max 17 ~ 22m

Disabling the DirectX 11 present a frame for every vsync, the render times are similar. Enabling it I get a single dropped frame occasionally, but It does not disturb playback.
And just for good measure I also switched to D3D9 windowed overlay mode but I got constant frame drops.

Thanks again for the help and suggestions.

Asmodian

14th July 2018, 10:04

Hmm... I forgot about heat influencing performance. I am actually water cooled so the GPU never goes above ~35°C unless it is really hot inside. My stock settings were running 1860 MHz. Those rendering times look quite good but I still think you might be right on the edge of performance. :(

With madVR v0.92.14 I start seeing dropped frames with 60p content (16.67ms frames) with reported rendering times at or above ~11ms, so you do need a lot of headroom.

Dropping chroma upscaling to NGU medium, or even image upscaling to NGU high, is not a significant loss of quality. :)

bluesam20

14th July 2018, 11:31

Okay thanks again, I did play around with dropping the Chroma upscale to NGU AA (medium). It saved about the same amount of render time as Smooth Motion, about 3~4ms. Dropping the Luma doubling from NGU Sharp (very high) down to (high) nearly halved my render times, I think I went from around 20ms on a 1080p30 source down to 9~10ms. I was hoping maybe I missed something, but I'm glad I asked! Oh and thank you for sharing your settings file, amazing work!

arcspin

22nd July 2018, 21:00

Hi,
I am trying to wrap my head around frame rates in hope of getting a perfect 23.976 frame rate with MadVR.
Right now I am at 23.97743Hz.
HD Fury Vertex says 23.975Hz
On the OSD (Ctrl+J) when I have both my primary PC screen on and my projector on there is information regarding "Composition rate" and that rate is at 23.979Hz.
The composition rate do not show when I turn off the primary PC screen and only use my projector.
There are a lot of different Hz showing on various OSD, is this unavoidable?
https://photos.app.goo.gl/aQwadiGSf69NkD5Z9

So my first question is what is composition rate and why is it at 23.979?

I also have, several times and almost gone mad, tried to optimize pixel clocks in the MadVR custom modes with no success, none (well I tried the first 7 optimized pixel clocks) of the pixel clocks did not take and I am back with the EDID/CTA setting.

Mind you, everything works fine with my setup and I have none or almost no frame drops or repeated frames. (I have MadVR set to PCM and not bitstreaming)
This is not my trying to solve a problem but more to understand how to improve with the help of MadVR.

My second question is if it is common to have to try several of the different optimized pixel clocks before there is a match?
https://photos.app.goo.gl/A4hKbNMccgk5EUGYA

This is my setup:
Hardware: Intel Core i5-8600K, ASUS GeForce GTX 1060 6GB DUAL (NVIDIA FW 385.28), Windows 10 Home 64-bit, version 1803
Software: JRiver 24, MadVr 0.92.14
Projector: JVC DLA-I X5500 (RS420, X570R)
Main PC Screen: AOC 4K

Scaler/Splitter: HD Fury Vertex
Top out -> Sony wireless headphones
Bot out -> JVC PJ

Manni

22nd July 2018, 21:29

Hi,
I am trying to wrap my head around frame rates in hope of getting a perfect 23.976 frame rate with MadVR.
Right now I am at 23.97743Hz.
HD Fury Vertex says 23.975Hz
On the OSD (Ctrl+J) when I have both my primary PC screen on and my projector on there is information regarding "Composition rate" and that rate is at 23.979Hz.
The composition rate do not show when I turn off the primary PC screen and only use my projector.
There are a lot of different Hz showing on various OSD, is this unavoidable?
https://photos.app.goo.gl/aQwadiGSf69NkD5Z9

So my first question is what is composition rate and why is it at 23.979?

I also have, several times and almost gone mad, tried to optimize pixel clocks in the MadVR custom modes with no success, none (well I tried the first 7 optimized pixel clocks) of the pixel clocks did not take and I am back with the EDID/CTA setting.

Mind you, everything works fine with my setup and I have none or almost no frame drops or repeated frames. (I have MadVR set to PCM and not bitstreaming)
This is not my trying to solve a problem but more to understand how to improve with the help of MadVR.

My second question is if it is common to have to try several of the different optimized pixel clocks before there is a match?
https://photos.app.goo.gl/A4hKbNMccgk5EUGYA

This is my setup:
Hardware: Intel Core i5-8600K, ASUS GeForce GTX 1060 6GB DUAL (NVIDIA FW 382.28), Windows 10 Home 64-bit, version 1803
Software: JRiver 24, MadVr 0.92.14
Projector: JVC DLA-I X5500 (RS420, X570R)
Main PC Screen: AOC 4K

Scaler/Splitter: HD Fury Vertex
Top out -> Sony wireless headphones
Bot out -> JVC PJ

In my experience (see config in my sig below) EDID/CTA is the only option that works with the JVCs. It gives me a frame drop/repeat every hour or so in 4K23p RGB 12bits 4:4:4 which is fine. In 1080p, I have a frame drop every 3 or 4 hours.

I use 385.28 which is the last driver working with the JVCs. The more recent drivers introduce a variety of bugs, and only one improvement (for HTPC use): 3D (1080pFP) goes from one frame repeat every 3mn to one frame repeat every 13min.

arcspin

22nd July 2018, 21:47

In my experience (see config in my sig below) EDID/CTA is the only option that works with the JVCs.

I use 385.28 which is the last driver working with the JVCs.

Great and thank you for a swift answer.
Now I know that I do not have to try more optimized settings and can settle with EDID/CTA, that is good news.

Sent from my SM-N950F using Tapatalk

arcspin

22nd July 2018, 21:54

I use 385.28 which is the last driver working with the JVCs.

Yes, I made a typo in my original question and I too use 385.28. Many thanks to you for that.

When you deem any other FW suitable for JVC a will try a newer FW but not before you have tried and tested it to be ok.

Sent from my SM-N950F using Tapatalk

arcspin

22nd July 2018, 21:56

It gives me a frame drop/repeat every hour or so in 4K23p RGB 12bits 4:4:4 which is fine. In 1080p, I have a frame drop every 3 or 4 hours.

Question, do you use PCM or Bitstream to your receive?

Sent from my SM-N950F using Tapatalk

Manni

22nd July 2018, 22:40

Question, do you use PCM or Bitstream to your receive?

Sent from my SM-N950F using Tapatalk

Bitstream because I need Atmos, DTS:X and Auro 3D. If you don't use immersive sound, PCM is fine.

jake9wi

5th August 2018, 21:46

convert or process HDR content by using an external 3DLUT: Requires display calibration hardware and software to perform effectively. Possible with Argyllcms.

Would these be the same LUTs i create for color calibration??

Asmodian

5th August 2018, 22:36

No, these are special LUTs created for mapping HDR either to your display's HDR mode or SDR. The normal calibration LUTs are all SDR.

Nastenka

3rd September 2018, 20:51

Hi Asmodian,

First off, thanks for your preset configs which have helped me tremendously in tweaking my settings.

I had a few more questions, though. I have a 1080 Ti as well, but my screen resolution is only 1440p. I've gone through with hardware calibration and all, therefore I am looking to tweak your settings a little more since with less upscaling to be done, shouldn't there be some more hidden potential? Since my sources are quite good, I already changed a couple upscaling NGU AAs to NGU Sharp.

Also, are your presets more of a general use case or specific, like for Anime and such?

I'd appreciate any help.

Thanks.

Asmodian

3rd September 2018, 21:35

Watching 1080p at 1440p is actually harder than watching 1080p at 4K. For 4K you simply double it (with NGU) but with 1440p you have to downscale it afterwards. I do not think there is anything to improve for watching at 1440p.

My presets are intended to be general but I do want Anime to look good as part of that. I might chose different settings if I didn't need Anime to look good as well but it would not be very different (i.e. changing NGU AA to NGU Sharp). :)

UMNiK

24th October 2018, 20:14

Great guide, but I wonder if you could provide a simple table of viable settings for 720p\1080p content on a 1080p monitor, the most common scenario. It can be rather difficult to figure out which settings are completely superfluous in that case, and which algorithms aren't wasted. At the moment I'm just using all Lanczos 3 AR (LL, SL) with smooth motion if judder is detected and no extras. I assume any doubling/quadrupling is pointless, but are refinements/other algorithms? I appreciate that you have provided the settings.bin, but I'd much rather have a nice table to figure it out and have some choice myself.

Olivier C.

22nd November 2018, 22:33

Hi,

this display is already calibrated: [BT.709, pure power curve, 2.20] This sets the current display characteristics. The gamut option affects the conversions used by madVR when playing video with different gamuts but the gamma options only have an effect when using the 'enable gamma processing' option in 'color & gamma'.

This is not totally true if HDR --> SDR conversion is used, at least with "tone map HDR using pixels shaders" option (I didn't tried other HDR options).

huhn

22nd November 2018, 22:35

yes HDR-SDR it is inconsistent and you can get all kinds of gamma if it it used with gamma processing.

Olivier C.

22nd November 2018, 22:50

I would to like to say that the gamma options have an effect even if the 'enable gamma processing' option in 'color & gamma' is not used at all, in case of HDR->SDR conversion

huhn

22nd November 2018, 22:58

kind of but i still hope madshi makes it more transparent first.

Olivier C.

22nd November 2018, 23:09

Thanks huhn

Asmodian

5th December 2018, 02:56

Finally updated for madVR v0.92.17.

This is not totally true if HDR --> SDR conversion is used, at least with "tone map HDR using pixels shaders" option (I didn't tried other HDR options).

Thanks! I updated that description to reflect this. :)

Ava Pug

25th January 2019, 12:48

Hi,

The link for Max Quality settings.bin doesn't work.

K3ka

7th February 2019, 15:50

Finally updated for madVR v0.92.17.

Thanks! I updated that description to reflect this. :)
Hello! Thank you for your guide.

Can you,please, reupload file with max quality settings?

Asmodian

7th February 2019, 15:52

Sorry, I am currently traveling and do not have access to the file where I am. :(

Edit: it looks like the link changed when I updated for 0.92.17, it doesn't usually do that. I have updated the link.

Siso

12th March 2019, 23:19

I just looked through your settings for max quality and low power, I'm wondering why do you use NGU anti-alias for image doubling, instead of NGU sharp?

Asmodian

12th March 2019, 23:30

I kept encountering content that looked weird with NGU sharp, AA seems the most universally good version of NGU.

It is especially true for lower budget anime. I am not sure what they do but it often has some damage that is not related to compression, odd aliasing on edges that NGU sharp preserves too well. I do change to NGU sharp with some content.

NGU sharp can also have a "plastic" look with some content, even live action, that doesn't happen with NGU AA.

I sit pretty close to my TV so that weird look from NGU sharp is too obvious, and NGU AA doesn't do it at all.

70MM

12th March 2019, 23:50

NGU sharp can also have a "plastic" look with some content, even live action, that doesn't happen with NGU AA.

I sit pretty close to my TV so that weird look from NGU sharp is too obvious, and NGU AA doesn't do it at all.

I think most people with projectors and large screens would use NGU sharp wouldnt they?

My scope screen is 145" wide, whenever I test 1080p material upscaled to 4K with madvr and displaying with my JVC NX9 I find NGU AA just too soft. Would this be a general concensus with others using 4K projectors?

Asmodian

13th March 2019, 00:29

It does depend on the display, please pick whatever looks best to you. My settings are simply what I think looks best, with any content, on my display under my viewing conditions. :)

I do like the extra sharpness from NGU sharp when it doesn't look weird. A lot of content looks really good with NGU sharp. I think a big part of why NGU sharp sometimes bothers me is that I am pretty close to the display, I can almost resolve pixels. If I move back several feet NGU sharp doesn't bother me nearly as often.

Also, a projector might soften the image a little, which could mitigate the aspect of NGU sharp that bothers me.

maxkill

13th March 2019, 04:24

Asmodian

13th March 2019, 04:52

Siso

13th March 2019, 09:56

giulianoprs

13th March 2019, 14:20

I also use NGU sharp with my plasma tv and a modest AMD rx 560 (strictly 1080p file) and I must say that I am very well, thanks also to the guides present here in this forum.

maxkill

13th March 2019, 17:39

Your settings look very bad to me, to each their own I suppose. :p

My settings are intended as a place to start, with options that should look good whatever the content and that don't change the image. I have almost all processing off by default and turn it on for specific content if it needs it.

You want to smooth and sharpen the image a lot, it looks artificial to me but if you like it go for it. :)
Yeah I guess we all like different. People like you who like it to look as close to the source as possible even if that means a grainy image, and people like me who enjoy a smooth looking image while sacrificing a little sharpness. My guide achieves the latter beautifully imo.

Klaus1189

13th March 2019, 18:48

Maxkill, please don‘t idolize your settings so much. It‘s already hurting. I‘m sorry but the settings aren‘t as good as you think, so please don‘t tell us what we should do.

Asmodian

13th March 2019, 19:00

I understand, but what about NGU Standart as a medium ground between NGU AA and NGU Sharp? What do you think of it?

I find Standard very similar to Sharp, but not quite as sharp. ;)

Normally I use AA or Sharp. Maybe I should make a set of sharp settings files, if people are interested? I hope people modify my settings some, especially since Max Quality requires a 2080 Ti. :)

maxkill

13th March 2019, 19:07

Maxkill, please don‘t idolize your settings so much. It‘s already hurting. I‘m sorry but the settings aren‘t as good as you think, so please don‘t tell us what we should do.
You forgot so say: In your oppinion. Since I have tried both settings I know I prefer mine BIGTIME and I am correct that it creates a smoother image. If you like grainy image that's totally fine, I sure as hell don't. :cool:
I think people should have the option to use which ever settings they like, if you like grainy use this if you don't use mine, simple as that.
edit: Also I never told anyone what to do, I am all for free choice so stop the lying please that's low.

Klaus1189

13th March 2019, 19:28

I never told anyone what to do, I am all for free choice so stop the lying please that's low.

1. You told us by starting a thread with the title „Optimal 4k settings for smoother image MadVR, enjoy!!“ which suggests to use your holy settings.
2. I am not lying, just look at the reactions in your thread, so many guys who don‘t like it.
3. I did not say that your settings are not good, but I did say that the settings are not as good as you think these are, otherwise you would not always write, when one thinks other than you, you write BUT I DO LIKE IT.
4. Please read your posts and think about it, simply don‘t ideolize it anymore so much. I didn‘t wrote anything else.

That is just hurting so much to read. Please stop it. Thanks.

huhn

13th March 2019, 19:31

so when klaus states his opinion so he is a lying and if you do that you are just stating an opinion?

maxkill

13th March 2019, 20:27

1. You told us by starting a thread with the title „Optimal 4k settings for smoother image MadVR, enjoy!!“ which suggests to use your holy settings.
2. I am not lying, just look at the reactions in your thread, so many guys who don‘t like it.
3. I did not say that your settings are not good, but I did say that the settings are not as good as you think these are, otherwise you would not always write, when one thinks other than you, you write BUT I DO LIKE IT.
4. Please read your posts and think about it, simply don‘t ideolize it anymore so much. I didn‘t wrote anything else.

That is just hurting so much to read. Please stop it. Thanks.
For a smoother image I am stating it is best yes. And I have proof of that aswell.

You are claiming I try to tell people what they should like. This is a false statement and a lie. If people prefer a grainy image there is nothing wrong with that. I am merely trying to give people the option of a smoother image.
If people who love a grainy image attack me in the thread has little to do with the effectiveness of the setting.

DMU

13th March 2019, 21:06

Normally I use AA or Sharp.
Well, why NGU AA does not work in quadrupling (4x) mode like NGU Soft/Standard/Sharp for image upscaling?
Pic (https://drive.google.com/open?id=1QqNN_MxPeQ4RUJ-vDuMsvx5mkvRqvz9V)

huhn

13th March 2019, 21:23

it's just used 2 times which is fine too.

Klaus1189

13th March 2019, 21:45

You are claiming I try to tell people what they should like. This is a false statement and a lie.

I never claimed that, I never did a false statement and I never lied. Please read my post again. The proof is when you get my point. And this is exactly the point. You write so much and say nothing. And if somebody has another opinion you write he/she is a lyer. You really don‘t seem to don‘t get my point, silly me, at least I tried to get you to get down on earth „MrIknowitall“.

To summarize and repeat it, perhaps you get it sometimes when you realize that your holy settings aren‘t as good as you think.

You also have your ridiculous noob thread in your signature. Lol. This is a proove for your problem. The users don‘t need your noob settings and do yourself a favor and don‘t attack users this way.

Please read the reactions in your noob thread. The users don‘t like it.

I have my own settings, but I don‘t create a thread because I think these only define my personal taste and don‘t help anybody, stick with Asmodians settings and tune it if you want to your personal taste as I did.

But please stop acting this way. You don‘t help anybody with your hurtful way. You don‘t get what others write here and seem to need it to hurt these who have another opinion.

Please stop, simply stop. Enough said, but I am sure you will again write hurtful nonsense. If I would be a moderator, you would be blocked for a short period of time to post new stuff, perhaps this will help.

I apologize for this post but it seems you simply don‘t get it with other words.
Sorry ...

maxkill

13th March 2019, 21:50

I can not see any difference from enabling chrome upscaling to max and switching to cubic or dxva on option.

I hear you might need to set the LAV Video up, is this true, do you really need this to run MadVR? I was to check that it the LAV Video was in the D3D11 setting so I tried that but no effect after installing.

I also reinstalled MadVR just to check and tried two different players. I can not tell from on and off of highest settings of chrome on a UHD file.

Must be something wrong right? Interestingly I do get the extra computer load.
I'm thinking the difference between on highest setting and off should be quite big or..?
I'm currently because of this running with DVXA on still looks great, but would be nice to test the other settings if in fact they are not being shown in my setup after enabling them.

Asmodian

13th March 2019, 21:50

Well, why NGU AA does not work in quadrupling (4x) mode like NGU Soft/Standard/Sharp for image upscaling?

madshi never made a NGU AA to do quadrupling, as simple as that. :)

Edit: As huhn said, 2x twice is a good option. It is even slightly higher quality than direct 4x with the other NGUs, though it is also quite a bit slower. It is nice to use very high for the first 2x and high for the second.

maxkill

13th March 2019, 22:16

I never claimed that, I never did a false statement and I never lied. Please read my post again. The proof is when you get my point. And this is exactly the point. You write so much and say nothing. And if somebody has another opinion you write he/she is a lyer. You really don‘t seem to don‘t get my point, silly me, at least I tried to get you to get down on earth „MrIknowitall“.

To summarize and repeat it, perhaps you get it sometimes when you realize that your holy settings aren‘t as good as you think.

You also have your ridiculous noob thread in your signature. Lol. This is a proove for your problem. The users don‘t need your noob settings and do yourself a favor and don‘t attack users this way.

Please read the reactions in your noob thread. The users don‘t like it.

I have my own settings, but I don‘t create a thread because I think these only define my personal taste and don‘t help anybody, stick with Asmodians settings and tune it if you want to your personal taste as I did.

But please stop acting this way. You don‘t help anybody with your hurtful way. You don‘t get what others write here and seem to need it to hurt these who have another opinion.

Please stop, simply stop. Enough said, but I am sure you will again write hurtful nonsense. If I would be a moderator, you would be blocked for a short period of time to post new stuff, perhaps this will help.

I apologize for this post but it seems you simply don‘t get it with other words.
Sorry ...
1. I would suggest you improve your english, it is quite hard to understand what you are driving for here..
2. That's like, just your oppinion man.
3. Tell you what, take your head out of the sand and actually apply my bin file and choose a scene with some grain in it, take a picture of that scene with my settings, then set the player to default and take a picture, then go to my thread and post these there and tell me why I am mistaken and I'll listen to what ya have to say.

Klaus1189

13th March 2019, 22:27

maxkill

13th March 2019, 22:41

I was already in the US for several weeks and the people understood me fine. I suggest to improove your english, perhaps you understand what I am writing about.

In my opinion it is enough, because the users saw your real face. I don‘t want to mess around with your settings anymore. Your behavior is ridiculous. Now forcing me to improve your settings, no thanks. If I am a lyer, why do you need my opinion. Lol.

Your behavior is ridiculous. Bye.
I respect you, alright. But you have 0 actual arguments, just predetermined thoughts and beliefs. If you wanna live your life that way that's fine my man. I think if you are going to accuse me of something you should be able to back that up is all.

Klaus1189

13th March 2019, 22:56

I have my own settings, but I don‘t create a thread because I think these only define my personal taste and don‘t help anybody, stick with Asmodians settings and tune it if you want to your personal taste as I did.

Already said: It is all personal taste, your thread doesn‘t help anybody, it wasn’t me who want to convince anybody with my settings, it is you.

Summary: Do the same, enjoy your personal settings, but don‘t praise your settings with the thread here which is also in your signature. I never said anything else, but please be more modest.

Bye, perhaps you know NOW what „bye“ means.

maxkill

13th March 2019, 23:05

Already said: It is all personal taste, your thread doesn‘t help anybody, it wasn’t me who want to convince anybody with my settings, it is you.

Summary: Do the same, enjoy your personal settings, but don‘t praise your settings with the thread here which is also in your signature. I never said anything else, but please be more modest.

Bye, perhaps you know NOW what „bye“ means.
Oh so it is personal taste now? Didn't sound like that just a moment ago, good!
It doesn't help anybody? Tell me how you came to this conclusion please, would be very interesting to know.

I will praise my settings if I want. I think they are great.
You should try not telling other people what to do and not do, people don't generally like that, just a tip. Practise accepting things as they are.

Asmodian

14th March 2019, 00:49

maxkill

14th March 2019, 04:18

I also wish you would not promote your settings here, they are so much in the wrong direction from what I think is good practice they almost seem like a parody.

They also seem like settings made by someone very new to quality video on PC and madVR. I see these overprocessed videos a lot and find it depressing. :(

If you like it fine, you can use your settings, but expect pushback if you promote them. I am just expressing my opinion too. I don't want doom9 promoting bad newb madVR settings as great after all. :p
Look, I appreciate your settings Asmodian. They are just not for me. I have explained the purpose of my settings many times now.
You liken my settings to a parody? There is a flip side to that coin. And that is that I can say the exact same thing about your settings.

Think about it like this. Do most people who play video games like a pixelated image or a smooth looking image?
This is the same thing if you think about it. I don't think most people would prefer a pixelated image.
The hardcore 'true content looking material lovers' that seem to overrepresent this section of this forum yes, but not most people I don't think so.
The other thing is, what you and some people say about the settings are nothing but loose statements such as 'well that doesn't look good' or 'no your settings are wrong'. No one ever provided a resonable explanation as to WHY smoothing of the picture must be so bad.
To me it's great because it removes the pixelated areas in the image creating a 'looking out window effect'. You can not achieve this with a pixelated image, you will see the pixels. Do you get this argument at all?

huhn

14th March 2019, 05:18

pixelated is an artifact that usualy come from sharpening, low resolution and not using an blurring effect like AA on games yes AA makes games less sharp which is working as intended "nothing" is sharper then aliasing.

and you want a reasonable explanation why your settings are "wrong". well that pretty simple and can be objectively proven.

a movie is created on an calibrated screen and so other user with an calibrated screen will get the same image from the mastering screen. this makes it objectively a correct image and settings that recreate this image are so objectively "correct".

your settings doesn't do that which makes them incorrect.

foxyshadis

14th March 2019, 23:57

maxkill, I think it's obvious that you're in the minority here, and that your opinion on what most people like might not be true. (For an extreme example, I'm not sure anyone wants to go back to the VHS era.) Please drop this pointless argument, no need to be defensive about what looks best to you, no one is telling you that you can't watch the way you want.

v0lt

6th September 2019, 20:18

@Asmodian
I do not see the pictures with the madvr working scheme on the first page of the topic. Can you return it?
Can you add an HDR-> SDR conversion to the scheme? I want to understand if this is done before or after resize.

huhn

6th September 2019, 21:52

you can check the OSD it should be done at the end like all color corrections.

Asmodian

6th September 2019, 22:53

I found a new host that doesn't watermark the images. :p

Like huhn I assume it is done at the end, right before the 3DLUT. I don't feel comfortable placing it on an image because we don't actually know exactly when it happens. If anyone has any evidence for when HDR->SDR happens I would be happy to take a look. :)

Asmodian

2nd October 2019, 09:04

After some experimenting with HDR content and "ShowRenderSteps" I have determined that HDR tone mapping actually happens before scaling. This is nice for performance if you do upscaling but bad if you downscale. Tone mapping before scaling might be important to avoid clipping issues after scaling (edit: though that sounds backwards to me...).

I added a new flow chart for UHD HDR content. :)

chros

2nd October 2019, 10:00

Nice, thanks!

zerowalker

4th October 2019, 04:08

I also wish you would not promote your settings here, they are so much in the wrong direction from what I think is good practice they almost seem like a parody.

They also seem like settings made by someone very new to quality video on PC and madVR. I see these overprocessed videos a lot and find it depressing. :(

If you like it fine, you can use your settings, but expect pushback if you promote them. I am just expressing my opinion too. I don't want doom9 promoting bad newb madVR settings as great after all. :p

I would like to hear your suggestion on settings.
I don't care much for processing as i don't want to alter the video,
but rather the up/downscaling, rendering etc.

I currently have:

Chroma upscaling: NGU - Standard, high quality
Image downscaling: SSIM - 2D - strength: 100% - activate anti-ringing filter [relaxed]
Image upscaling: NGU Standard - very high

Rendering - user D3D11 - use a frame for every VSync
CPU Queue 128
GPU Queue 24
All off on trade quality for performance

I basically want as high quality without messing up the video.
I chose these settings (especially the scaling) as i thought they were the best (meaning not the best for me, but best quality of the options available).
I couldn't use NGU on very high for chroma as it seems to use up my GPU sadly enough (GTX 1070).

The queues i am unsure about, from the explanation i guess CPU queue has a hand in subtitles,
and one kinda frustrating bug/issue i have is that when jumping back/forth in a video the subtitles sometimes don't render,
so i have to do another back/forth to re-trigger it.
It would be awesome if that could be solved,
but i don't think many run into it as probably jump around in videos a lot more than normal people do haha xd

Thanks:)

Asmodian

4th October 2019, 06:07

I think your settings are generally quite good.

Chroma upscaling: I prefer NGU AA, especially for chroma. Even when using NGU sharp I use NGU AA for chroma. I never set it higher than high for chroma, even with a 2080 Ti and in my SD profiles. Very high doesn't seem to improve real world chroma scaling, it may even harm it. So don't feel like you need a better GPU for that. :)
Image downscaling: I actually use SSIM - 1D 100% LL AR relaxed, 2D can be a bit much depending on the content and scale factor. 2D is great sometimes but I like 1D more for most 4K -> 1080p downscaling when I want to watch 4K content in a window. Linear light is important for downscaling.
Image upscaling: I like NGU AA very high, especially for animated or non-real life content but NGU AA v.s. NGU Standard is a fairly subtle difference with trade-offs either way. I don't like upscaling and downscaling for scaling factors below 1.5 for doubling or 3x for quadrupling. I also use Jinc AR for upscaling and Lanczos3 LL AR for downscaling after doubling/quadrupling.

There is no reason to set particularly large buffers, it simply spikes the IO and CPU/GPU for longer after a seek or on startup. As long as they are big enough to never get frame drops due to background tasks I set them as low as possible. The defaults are quite good and I use 8/16 for GPU/CPU buffers. I also use D3D11 with present for every vsync.

One setting you forgot to mention was the frames presented in advance. I found 3 to work best for me on Win10 and current Nvidia drivers (436.48), again as low as possible while never getting presentation glitches or frame drops. 8 seems too high.

Dithering is also very important. On my LG C7 and C9 I like ordered dithering but on my previous LCD screens I preferred ED1 or 2. I also much prefer disabling both colored noise and change dither for every frame. In my "Max Quality" settings.bin, linked at the end of the third post in this thread, I set ED1 for "slow" dithering but I actually use ordered dithering for both fast and slow myself. I just couldn't post a "max quality" settings file and not use one of the error diffusion dithering options. :o

DMU

13th October 2019, 20:36

Why is the rendering time significantly longer in OS HDR than in AMD HDR?
Screenshot. (https://drive.google.com/open?id=1R8GkNgra9ftQxm-9iWcXLhsRdIVdhF1G)

chros

14th October 2019, 12:05

Dithering is also very important. On my LG C7 and C9 I like ordered dithering but on my previous LCD screens I preferred ED1 or 2. I also much prefer disabling both colored noise and change dither for every frame.
That's interesting, can you explain why? Do you have any content to reliably test it with?
I used ED2 (both options were checked) for ~2 months now and sometimes noticed that something is odd, although I couldn't really tell why (maybe it was the source as well).
Now I switched back to Ordered dithering and I'll try to wait for a content to feel the above :)

Cowboydude99

17th October 2019, 01:19

Asmodian

17th October 2019, 01:55

Please attach images somewhere else and post a link, it takes a long time to get attachments approved here. :)

That's interesting, can you explain why? Do you have any content to reliably test it with?

I don't really. Testing dithering options at 8 bit or higher always feels a bit... placebo?

I have noticed colored dithering being on before, and been very happy to be able to turn it off, but I couldn't actually say how I noticed or why I cared. :o

nsnhd

17th October 2019, 02:56

Image upscaling: I like NGU AA very high, especially for animated or non-real life content but NGU AA v.s. NGU Standard is a fairly subtle difference with trade-offs either way. I don't like upscaling and downscaling for scaling factors below 1.5 for doubling or 3x for quadrupling. I also use Jinc AR for upscaling and Lanczos3 LL AR for downscaling after doubling/quadrupling.
With NGU Standard you can do quadrupling, but not with NGU AA, so quadrupling is better for upscaling 720p to 2160p ?
Any reason to choose Lanczos3 LL AR instead of SSIM - 1D 100% LL AR or Bicubic 150 LL AR after doubling/quadrupling since they cost the same performance I think ?
Thank you.

Asmodian

17th October 2019, 19:14

With NGU Standard you can do quadrupling, but not with NGU AA, so quadrupling is better for upscaling 720p to 2160p ?

Not in my opinion. NGU direct quadrupling is lower quality than NGU 2x+2x, but it is also faster. This is a complicated performance/quality tradeoff but if you can do NGU very high 2x followed by NGU high 2x it should be better than NGU very high direct 4x.

Any reason to choose Lanczos3 LL AR instead of SSIM - 1D 100% LL AR or Bicubic 150 LL AR after doubling/quadrupling since they cost the same performance I think ?
Thank you.

Lanczos is much easier than SSIM 1D, but it is actually a little harder than Bicubic 150. I find downscaling after quadrupling does not noticeably benefit from SSIM and lanczos is quite fast and offers mostly artifact free downscaling. Bicubic 150 is great when you need it really fast but it not superior to the spline family (lanczos/spline) in my experiance.

SSIM can even preserve NGU artifacts too well.. at least maybe in ways I would have trouble proving... but it offers no obvious improvements. :o

Klaus1189

17th October 2019, 20:27

Hey there
Can someone help me figure this out!

Why is my video concerting to a different resolution?

I don’t see any setting with that resolution...

The movie won’t really play. A few frames a minute..

Please attach images somewhere else and post a link, it takes a long time to get attachments approved here. :)

As Asmodian already said, upload also here to Google drive or whatever so we can view it.

nsnhd

19th October 2019, 02:20

Not in my opinion. NGU direct quadrupling is lower quality than NGU 2x+2x, but it is also faster. This is a complicated performance/quality tradeoff but if you can do NGU very high 2x followed by NGU high 2x it should be better than NGU very high direct 4x.
Except for NGU AA, I can't find the way to set NGU 2x+2x, can you ?
But in my eyes NGU Sharp very high 2x > Jinc AR looks better than NGU Sharp very high direct 4x <SSIM 1D. You're right, thanks.

huhn

19th October 2019, 02:23

NGU AA is always x2 and you can use it to do x2 two times.

nsnhd

19th October 2019, 03:01

NGU AA is always x2 and you can use it to do x2 two times.
Yeah, but I can't do it for other NGU's

Asmodian

19th October 2019, 09:41

Yeah, but I can't do it for other NGU's

Yes you can, simply set luma quadrupling to 'double again'
https://i1222.photobucket.com/albums/dd496/asmodian3/madVR%20-%20double%20again.png

nsnhd

19th October 2019, 12:57

Yes you can, simply set luma quadrupling to 'double again'

Thanks, upscaling 720p to 2160p works for NGU very high-NGU high with gtx 1060 6gb, though I have to lower the chroma of NGU to Normal (Bicubic60 AR)

cbnbmore

15th November 2019, 23:44

I am trying to settle in on some solid settings now. What Driver do most of you use for the RTX cards now?

Asmodian

24th November 2019, 00:04

I like 441.20 on my 2080 Ti, at least for my uses. The latest at the time of posting.

Bartok

7th December 2019, 18:26

hi Asmodian,have you got some advices to take benfit of my pana FZ-800 for hdr settings: tone map or passtrough ? Calibration settings best choice ? 8,10,12 bits usage ? By the way i'm using you high profile wirh rtx 2080 , works fine :)
thanks a lot !

ng4ever

10th December 2019, 05:35

436.30 imo

I left this forum because I got the best settings for madVR from a certain forum poster for my RTX 2060 Super ! Thanks again.

Can now do 360, 720, and 1080p upscaled easily.

petoulachi

10th December 2019, 09:33

Hello Asmodian,

first of all thanks for sharing your settings, it's a wonderful base to start to do advanced filters with MadVR.
I have one question regarding your devices, in the calibration tab. With your C7, You are using 3DLut, and havce setup BT709, but also SMPTE C,EBU/PAL and BT2020 (with C9 I guess you have set the LUT directly onto the TV). Could I ask you what software did you use to do them, and when are there usefull ?
I'm using DisplayCal/Argyllcms and by default it generate only BT709, and I'm curious to know if there are some points that could be better on my configuration.

Also, why don't you use 3DLut for hdr ?

You also switch to 10bits/8bits depending on the HDR or not, but not with your C9.

Thanks for your precious input :)

zerowalker

9th January 2020, 21:25

About the CPU/GPU queue size, is there some way to determine if there's framedrops (i mean if it occurs always it's easy to spot, but i think i have seen smooth motion not working properly at times, but just jumping forth/back or something to reset it fixes it, and i think it might fix itself, it's not that obvious so not really sure if i am just mistaken though).

And also how does CPU/GPU differ there, as the CPU is the one it's limited to i am guessing they act differently and the CPU is the real buffer and the GPU is a some backup somehow?

Also don't recall if i asked this before, probably did, but is the "use a separate device..." preferred, or is it just overhead for no real reason?

Asmodian

9th January 2020, 22:48

Are you using some form of VRR? G-sync/Freesync? Smooth motion does not work well with VRR, disable VRR for madVR and/or your player.

The CPU buffer is the in system memory buffer. It holds frames before they are uploaded to the GPU, after any CPU processing steps (IVTC, etc.). This is the "before madVR" buffer.

The GPU buffer is for after madVR but before present, this is how many completed frames are stored on the GPU. The "after madVR" buffer. This one has more impact in my experiance, the CPU buffer only benefits from a few extra frames. Too big is as bad as too small in my experiance, the default 8 is good. 6-12, above that has never helped on my systems.

Use a separate device does not seem to help with Windows 10. It was mostly useful in the early days of hardware decoding on the GPU, newer OS and drivers seem to have fixed any issues that using a separate device might help. I leave it disabled today as unnecessary overhead but it did not seem to hurt anything during testing.

maxkill

11th January 2020, 18:14

Getting stutters from 60fps video to 23hz playback. Normally I play 24hz videos etc no problem, but get stutters when playing the 60fps movie. (My screen is set to 23hz with max 30)

I get no performance spikes to 100% and I tried enabling the hack direct3d with no success.

Any ideas what settings to make the 60fps video smooth on under 30hz?

huhn

11th January 2020, 18:42

well you can't show a 60 fps file smooth at 23p that's simple not doable.
at 30 it could work out if the source is true 60 with no 30 fps parts or it will be a disaster.

not sure if madVR has proper decimation it's nothing user usually need.

maxkill

11th January 2020, 20:02

huhn

11th January 2020, 21:07

just use 60 hz.

Asmodian

12th January 2020, 00:32

Yes, madVR's smooth motion is not designed to handle frame rates above the refresh rate. They must be the same at least, but really the refresh rate should be more than double the frame rate before smooth motion works well.

huhn

12th January 2020, 01:00

well i personal think it is not useful for that but it has extra code for this situation and when SM was new i talk about it well i even started a bug report before the talk started no one cared so it was left in like this and it does what you see intentional.

so technically it was mode for this. i simply don't know why someone would want that result.

Asmodian

12th January 2020, 01:17

When first designed it was thought about but I think it has never been optimized for it to actually give a good experiance. Since no one cared it was dropped from the design. ;)

It seems to 'randomly' drop frames instead of maintaining smooth but blurry blending. I think it cannot blend more than two frames so it seems choppy when some frames get dropped.

huhn

12th January 2020, 04:18

AFAIK it is completely finished.
i saw it blending 5 frames.

DMU

12th January 2020, 12:22

Any ideas why the rendering time in the OS HDR mode is much longer (https://drive.google.com/open?id=1R8GkNgra9ftQxm-9iWcXLhsRdIVdhF1G) than in the AMD HDR mode?

chros

12th January 2020, 13:48

Display is at 60Hz, do you use smoth motion? Maybe it switches back to copyback?

DMU

12th January 2020, 17:22

Display is at 60Hz, do you use smoth motion

In this case, no.

Maybe it switches back to copyback?
All settings are the same except for the Windows HDR switch.

maxkill

12th January 2020, 23:31

Well, it seems my receiver is capped at 30hz so I'm shit out of luck then.

huhn

12th January 2020, 23:32

that would break the spec try it at 1080p

DMU

26th January 2020, 19:46

Has anyone dealt in detail with the option "scale chroma separately, if it saves performance"? Can you comment on this (https://forum.doom9.org/showpost.php?p=1896839&postcount=58438) post?

zerowalker

26th January 2020, 22:33

Asmodian

27th January 2020, 01:07

If it seems to work fine most of the time on the lowest settings I would simply use the defaults and be happy.

Don't worry too much about buffer sizes, they are not that critical and the defaults are chosen well.

Asmodian

28th January 2020, 19:24

Has anyone dealt in detail with the option "scale chroma separately, if it saves performance"? Can you comment on this (https://forum.doom9.org/showpost.php?p=1896839&postcount=58438) post?

When we were testing it when it came out it was decided that for real life content it looked better disabled, and in cases where you always had to scale the chroma (e.g. 1080p on a 1440 display). Still, it is more a case of tradeoffs rather than obviously better one way.

Blurring chroma differently than luma can cause odd artifacts, so perhaps disabled is safer. That said avoiding scaling entirely is sharper. For 4K video on a 1080p display I can see leaving it on, but test with something besides text on a solid background too.

I don't think text on a black background is a great chroma test pattern for this feature, it is easy to notice subsampling but it is not very representative of real life content.

DMU

30th January 2020, 09:32

For 4K video on a 1080p display I can see leaving it on, but test with something besides text on a solid background too.

I don't think text on a black background is a great chroma test pattern for this feature, it is easy to notice subsampling but it is not very representative of real life content.
Can you recommend specific real life content for the test?

huhn

30th January 2020, 13:45

mostly video games and mostly if you upscale and downscale at the same time.

they are the reason why i think NGU chroma should always be used with NGU image doubling.

your test should be repeated with a simpler chroma scaler but i'm pretty sure it will still loose with no questions. was this test file correctly mastered with the correct chroma position?

DMU

30th January 2020, 15:14

When we were testing it when it came out it was decided that for real life content it looked better disabled, and in cases where you always had to scale the chroma (e.g. 1080p on a 1440 display).
In my practice, I have not seen the operation of this option when upscaling an image, only when downscaling (for example, 2160p -> 1080p)

was this test file correctly mastered with the correct chroma position?
Not sure. I received the video for the test in 2 different ways, but in both cases the results were the same:
1 way
- from the source .png image was received .jpg image in 4:2:0 format using the paint.net editor;
- using ffmpeg a video was created for the test with the default settings.
2 way
- using ffmpeg a video was created for the test with the yuv420p setting from the source .png image.

ikarad

4th April 2020, 10:12

-target peak nits: [200] Set the target display's peak white brightness. Above this value is used as the highlights. A lower setting increases the brightness of mid range values.
What is the best value to use for my display? Is it 200 as you described or another value?

My display is https://www.tftcentral.co.uk/reviews/acer_xb270hu.htm

Thanks.

Asmodian

8th April 2020, 19:29

200 is simply the default. This is a good starting point but adjust it to your preference. There isn't really a "correct" way to handle HDR -> SDR, you have to decide what looks best to you and different HDR mastering might change your preference for this value.

Audionut

10th April 2020, 13:10

I was under the impression that target peak nits should be set to the measured white level of the display. There are other ways to adjust mid range brightness, right?

There may not be a "correct" way to do HDR -> SDR, but madVR is doing HDR -> SDR. Telling madVR that the target peak nits is going to be 200 when the display is only 100, or 500, or whatever, doesn't seem appropriate.

Asmodian

10th April 2020, 18:45

In my experiance, and depending on the HDR master, I get more pleasing results using values not quite the same as my measured peak brightness. For HDR->HDR target peak nits should be set to the measured white level, HDR->SDR is less exact. Again, mastering has a huge effect as well, some of my HDR movies need very different target peak nits to look decent in SDR. :p

And 200 is about right for the brightness of a XB270HU. I have one myself and 160-220 nits is a good brightness setting for it, without a meter simply adjusting it and evaluating the result is all you can do.

Audionut

11th April 2020, 11:22

Without a meter, everything is guesswork.:D

As I get older I get less anal about these things, hence the good thing about a meter. Measure and forget. :cool:

If doing HDR -> SDR I would assume you would ideally want to be using the latest beta build. There's lots of options in there. Some of those would seem far more preferable then playing with target nits.

But I haven't played with any of them though. The defaults work fine for this old bastard. I'll shut the door on the way out. ;)

takenori

11th April 2020, 15:25

Hi, is htere any reason why the custom mode doesnt switch automatically anymore? is there any new tutorial inline with newer driver/updates of windows and gpu?
my pc used to switch automatically a few years back (2018 i think)
today, i tried the same tutorial and it wont switch automatically.

im on gsync compatible freesync hdr monitor (turned off via monitor OSD), windows 10 home, and nvidia gtx 10
all 3 is in the latest firmware, update, and driver.

thanks for the answers

Asmodian

11th April 2020, 20:48

takenori

12th April 2020, 01:28

Do you have G-sync disable in the drivers as well?

Sorry, I don't know of a guide but automatic switching works for me without doing anything special, I simply need to enable them in display modes.

Edit: Actually this works even with G-sync enabled in the drivers.. so my first question isn't relevant. :o

glad that i can still use freesync on, swithing themregularly is a pain.

so after a few fiddling, the only time im able to auto switch is only from the available standard mode, the missing list is not switchable
also, its a 1440p monitor,
but on the list of mode madvr may switch to, it listed as invalid (1440p23, 1440p59, and so on)
1080p and 2160p listed as valid
so, if i play a 1080p59 video on my 1440p monitor, it would switch to 1080p59 instead of 1440p59
https://i.imgur.com/tg11GkC.jpg

edit: also, instead of apply button after each refresh rate, it shows test mode instead.

Asmodian

14th April 2020, 08:43

Sorry, that all seems to be an issue with that monitor. It seems to be missing several useful modes, I would expect 119 Hz on a high Hz display (useful for 23 and 59). You only need 100, 119 (119.88012), and 120 Hz, everything else fits nicely into just them so do not worry about 50 Hz missing. Maybe try CRU (https://www.monitortests.com/forum/Thread-Custom-Resolution-Utility-CRU) to add 119.88Hz?

Make sure to disable G-sync for your video player, madVR does not get along with G-sync because it relies on the display having a fixed refresh rate. Smooth motion goes really bad but the frame timing with G-sync enabled is not great even with smooth motion disabled.

Edit: A great option, and what I would do, is leave it on 2560x1440@144Hz all the time and use smooth motion. Smooth motion works really well at 144Hz.

takenori

14th April 2020, 13:51

managed to switch to my resolution,
apparently, i need to type the whole resoultion instead of 1440p.
but they result in a bad audio lag.
both with gsync compatible turned on and off, still audio lag.
and thanks for the suggestion, will try smooth motion because lots of my contents is 29 and 25 fps.
but do i need to use them for 23 materials? stat still reports 1 frame drop every 44seconds playing 23 on 144hz

huhn

14th April 2020, 14:02

if you drop a frame every 44 sec and you don't want that the answer is yes.

takenori

15th April 2020, 12:47

sorry, i mean frame repeat.

Asmodian

15th April 2020, 20:40

Smooth motion still offers an improvement for 23.976 fps @ 144 Hz, you will not get full frame drops or repeats without it but you do get judder. Without smooth motion most frames will be displayed for 6 refreshes but every 44 seconds a frame will be displayed for 7 refreshes instead. This is not too bad but with smooth motion you do not get any judder.

pirlouy

16th April 2020, 09:18

Bah, the repeated frame lasts 7 ms, hard to detect for a 24 fps video. Thanks high refresh rates.
But indeed, if you use madVR, Smooth Motion is not hard on CPU/GPU and does not add flickering in that case, so you can use it.

huhn

16th April 2020, 17:09

if you run a frame async every 44 sec ~7 ms correction is not going to fix ~42 ms of sync there are far more repeated frame and they should be to hard to notice.

lazostat

16th April 2020, 23:49

You make all the settings for better image quality. But happen when you duplicate screen and you play the video from pc to tv? Do i have all these image settings adjusted on my Tv also? If yes, don't they conflict with the settings i already have done on my TV in order to have better quality?

Asmodian

17th April 2020, 00:17

lazostat

17th April 2020, 00:32

I am not sure what you mean by 'conflict'. I turn off all image enhancements on my TV (sharpening, denoising, frame interpolation, etc.), but I would without madVR as well.

Most of madVR's improvement is simply due to doing the required math with as high quality as possible, e.g. scaling, color space conversions, dithering, etc.

The only true conflict I can think of is smooth motion, which will interact badly with frame interpolation. You really do not want to use the TV's frame interpolation or those modes that try to pull 24 fps from the 60 Hz signal with smooth motion, it breaks those algorithms. Things like sharpening twice also looks bad but that is more due to over-sharpening. I am still not sure I would call that conflicting with each other but they can interact badly. I consider madVR's filters like sharpening, denoising, artifact removal, etc. as only useful for specific sources, not something to run all time.So what are the best options to use on Madvr if i see movies on my tv? The same settings will apply if i see movies on pc monitor?

Also what about rgb? I have an option on my samsung TV called HDMI black and two options, low and normal. I read that low is full RGB and i see better black levels.

Also, do i have to change some settings on nvidia control panel?

Also, my pc monitor has only 60Hz mode. So i have to enable smooth motion option on Madvr? And how can i understand if all working without problem? What is the ideal sec/min to have dropped/repeated frame? And what about rendering time and glitches?

Klaus1189

17th April 2020, 09:11

Also what about rgb? I have an option on my samsung TV called HDMI black and two options, low and normal. I read that low is full RGB and i see better black levels.
Not "better" blacks, the matching setting will be the right setting. It depends on the output of your GPU:

Limited 16 - 235 --> "normal"
Full 0 - 255 --> "low"

Keep in mind 99,9% of all videos are limited 16-235. There is just a conversion involved, perhaps more, but that doesn't matter now. We discussed it also here, but it all comes down to your display.
https://forum.doom9.org/showthread.php?p=1907286#post1907286 and the next posts

Also, my pc monitor has only 60Hz mode. So i have to enable smooth motion option on Madvr? And how can i understand if all working without problem?

Please download CRU, see the first post here for the download link:
https://forum.doom9.org/showthread.php?t=176013

Then you can add missing refresh rates, but you have to test it, if it works. There is no guarantee, it depends on the monitor / TV / projector, ...

What is the ideal sec/min to have dropped/repeated frame? And what about rendering time and glitches?

the best is no drops or repeats, but in my opinion longer than your longest movie is already enough. My RX 5700 XT does 11,8 hours out of the box, as did my RX480 8GB, thank god it is still here.

The rendering time must be blow the frametime.
23 Hz frametime is longer than 59 Hz frametime, so you can do more heavy upscaling on 23 Hz than on 59 Hz, because the GPU is too slow to deliver it fast enough.

Klaus1189

17th April 2020, 09:12

We can not view your "Capture.JPG", please use google drive or similar.

lazostat

17th April 2020, 22:06

Not "better" blacks, the matching setting will be the right setting. It depends on the output of your GPU:

Limited 16 - 235 --> "normal"
Full 0 - 255 --> "low"

Keep in mind 99,9% of all videos are limited 16-235. There is just a conversion involved, perhaps more, but that doesn't matter now. We discussed it also here, but it all comes down to your display.
https://forum.doom9.org/showthread.php?p=1907286#post1907286 and the next posts

Please download CRU, see the first post here for the download link:
https://forum.doom9.org/showthread.php?t=176013

Then you can add missing refresh rates, but you have to test it, if it works. There is no guarantee, it depends on the monitor / TV / projector, ...

the best is no drops or repeats, but in my opinion longer than your longest movie is already enough. My RX 5700 XT does 11,8 hours out of the box, as did my RX480 8GB, thank god it is still here.

The rendering time must be blow the frametime.
23 Hz frametime is longer than 59 Hz frametime, so you can do more heavy upscaling on 23 Hz than on 59 Hz, because the GPU is too slow to deliver it fast enough.I tried to add some missing Hz modes with MadVr, but then monitor turned off and says that this resolution is not supported. Does CRU work different? It there any change my monitor will get some kind of damage?

Is there some kind of test video, so i can see how 60Hz monitor affects the 23/24 Hz playback? Cause i see many movies and animes on my pc monitor and i don't see something strange..

Should i upload some madvr stats during playback of some movies to tell me where is the problem?

Asmodian

17th April 2020, 22:51

No, CRU will not help if your monitor responds that way. Theoretically there could be damage but it is very unlikely, it just turns off and displays that message.

With smooth motion enabled 24 fps on 60 Hz is great, without it watch medium fast panning scenes for jumpy motion or judder.

lazostat

17th April 2020, 23:08

MadKodi

25th April 2020, 22:59

im trying to make a new profile.
on the scaling algorithms tab... the create new group is grey out. i can not click.
how do i make a new profile ?

Asmodian

26th April 2020, 00:50

You need to have some setting pages that are not already in a profile group.

alisaeed

16th July 2020, 00:16

how can I back up my settings before I use yours ,, in case I want to return to mine?

Asmodian

16th July 2020, 08:00

Simply keep a copy your current settings.bin file. If you do not have one you need madVR to be in a directory where it has write access, you can edit permissions or move it out of program files and run install.bat as administrator.

senzaparole

3rd October 2020, 15:35

Hi Asmodian!
I have been using your madVR HIGH settings for months with great satisfaction.

Maybe I should create a new "Scaling" profile for 1280x720 50FPS videos because madVR considers them "1080p30" and that's not good: high GPU temperatures, 60ms rendering and lots of lost frames.
On the other hand, with real 1080p30 videos everything works correctly. So I think I should create another profile .. what do you say?

If I have to create another profile you would be kind enough to write me the rule to insert here:

If (filePath = "* youtube *") "Streaming"
ElseIf (deintFPS <31) AND (srcWidth <= 1024) AND (srcHeight <= 576) "SDp30"
ElseIf (deintFPS <61) AND (srcWidth <= 1024) AND (srcHeight <= 576) "SDp60"
ElseIf (deintFPS <31) AND (srcWidth <= 1280) AND (srcHeight <= 720) "720p30"
ElseIf ((deintFPS <61) AND (srcWidth <= 1280) AND (srcHeight <= 720)) OR ((deintFPS <31) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) "1080p30"
ElseIf ((deintFPS <61) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) OR ((deintFPS <31) AND (srcWidth <= 4096) AND (srcHeight <= 2160)) "2160p30"
ElseIf ((deintFPS <121) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) OR ((deintFPS <61) AND (srcWidth <= 4096) AND (srcHeight <= 2160)) "2160p60"
Else "Fast"

VBB

3rd October 2020, 19:20

Separate these two

ElseIf ((deintFPS <61) AND (srcWidth <= 1280) AND (srcHeight <= 720)) OR ((deintFPS <31) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) "1080p30"

into

ElseIf ((deintFPS <61) AND (srcWidth <= 1280) AND (srcHeight <= 720)) "720p60"
ElseIf ((deintFPS <31) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) "1080p30"

senzaparole

3rd October 2020, 19:52

thanks VBB.

so, summarizing I create the "720p60" profile and set it up taking a cue from the "720p30" profile

Then I delete the rule: ElseIf ((deintFPS <61) AND (srcWidth <= 1280) AND (srcHeight <= 720)) OR ((deintFPS <31) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) "1080p30"

and I insert these 2 rules:
ElseIf ((deintFPS <61) AND (srcWidth <= 1280) AND (srcHeight <= 720)) "720p60"
ElseIf ((deintFPS <31) AND (srcWidth <= 1920) AND (srcHeight <= 1088)) "1080p30"

and I live in peace :-)

VBB

3rd October 2020, 19:53

That should do the trick. Report back if it doesn't :)

senzaparole

3rd October 2020, 20:42

certainly. I'll give you feedback as soon as I try. :-)

Thank you very much

senzaparole

4th October 2020, 16:41

Hi VBB,

works all perfectly.
With the new rules now madVR classifies all the movies I have in the right profiles, including the new 720p60 profile.
Thanks for your help

lello

12th October 2020, 13:31

A very interesting feature for those who, like me, have a 21: 9 aspect ratio screen, is the automatic detection of black bars. Unfortunately the black bands are not detected when using DXVA2 or D3D11 native decoding.
But using DXVA2 for decoding (copy back), my RX580 can't cope with BD UHD 4k: is there a way to detect black bands even with native DXVA and D3D11 decoding?

Manni

12th October 2020, 15:21

is there a way to detect black bands even with native DXVA and D3D11 decoding?

Not yet. Also BD Menu in jRiver (if you use that) is only available with copyback, not with native.

lello

12th October 2020, 15:24

In fact, that's why I don't use menus. And is there any possibility of a solution soon?

chros

12th October 2020, 16:08

Probably not.

Asmodian

12th October 2020, 23:22

In fact, that's why I don't use menus. And is there any possibility of a solution soon?

The native solutions do not copy the decoded video data to CPU memory so the CPU cannot be used to detect and discard black bars.

I don't know the specific technical reasons but it is harder to do the black bar detection on the GPU, some algorithms only run on the CPU and black bar detection is one of them. I consider it very unlikely that madshi will create a black bar detect and discard that runs on the GPU anytime soon but I do remember long ago mentions of wanting to implement madVR fully on the GPU so it is not out of the question if it is technically feasible.

angmav

13th October 2020, 09:48

I have an unusual question to ask and sorry if this is being asked in the wrong section is it possible to change lav filter settings from copy back to native depending on the video playback in particular I was hoping to use copy back for 23hz playback and native playback for 60hz/50hz

Sent from my iPhone using Tapatalk

lello

29th November 2020, 18:53

I wanted to try SmoothVideo Project to try to improve the smoothness of my projector with 4k movies (frame interpolation only works with HD video) and by pure chance I found that, using SVP, black band detection also works with D3D11 decoding .

I did the double check and disabling SVP the detection no longer occurred: what is it due to?

Too bad that with SVP + madvr I can't convert 4k videos from 24 to 60hz, otherwise I would have bought it.

huhn

29th November 2020, 19:33

SVP is not native so there is your answer it's d4d11 copyback and that can do cropping using blackbar detection.

lello

7th December 2020, 10:44

And is there a way to use SVP only for blackbar detection?

huhn

7th December 2020, 11:02

madVR can do it without SVP just use a copyback decoder.

lello

7th December 2020, 16:21

Yes, of course, but as I said above, with the copyback I have problems with BD UHD: they don't really work, everything crashes :(

huhn

7th December 2020, 16:27

then try DXVA2 copyback or at least post the crash report/OSD from madVR.

Asmodian

29th December 2020, 22:58

Did you need more information than this line?

"To use simply restore madVR to default settings and place the file in the madVR program directory. madVR will automatically create or update the file with new settings if it has write access to its directory (e.g. not in "program files"). madVR will still use the registry to store these settings, along with any changes, without write access."

That should be all you need to know to get the settings.bin working.

lello

18th January 2021, 08:06

then try DXVA2 copyback or at least post the crash report/OSD from madVR.
This is the OSD with 2k files
https://i.ibb.co/NZdvNs2/OSD-HD.png (https://ibb.co/NZdvNs2)
and this one with 4k files.
https://i.ibb.co/rcTtv1W/OSD-UHD.png (https://ibb.co/rcTtv1W)
As you can see with UHD movies the rendering is high by activating DXVA2 copy back so, in my opinion, I can solve the problem only by replacing my RX 580 with a more powerful GPU: could an RX 5700 XT be enough?

huhn

18th January 2021, 08:35

that should be a bug hte UHD file should be "easier".

add a folder named Showrendersteps in the madVR folder and have a look at the more detailed rendersteps.

Asmodian

18th January 2021, 09:34

As you can see with UHD movies the rendering is high by activating DXVA2 copy back so, in my opinion, I can solve the problem only by replacing my RX 580 with a more powerful GPU: could an RX 5700 XT be enough?

Do you have slow HDR processing enabled? DXVA2 copy back is probably allowing more advanced HDR processing which can easily be very slow on a RX 580.

I don't know what GPU you might need for the slower HDR processing. :o

lello

18th January 2021, 10:47

https://i.ibb.co/tpLvSM6/OSD-UHD-2.png (https://ibb.co/tpLvSM6)

I added the Showrendersteps folder and this is the result with D3D11 selecting the device and NOT native.
By slow HDR we mean D3D11 with device selection while advanced DXVA2 copy back?
I have the problem in both cases.

I thought I could solve with a more powerful GPU because by installing madvr 0.92.17 smooth, without making any settings, the detection of the black bands works

huhn

18th January 2021, 14:01

the screen seems to be from paused mode so not that useful.

this could be an issue with copyback check your PCI lanes.

Asmodian

18th January 2021, 15:02

By slow HDR we mean D3D11 with device selection while advanced DXVA2 copy back?

No, you have madVR's more advanced HDR processing enabled which can run very slowly depending on the options chosen. It has nothing to do with DXVA/D3D11 except that some of the HDR processing does not work when using native decoding.

I thought I could solve with a more powerful GPU because by installing madvr 0.92.17 smooth, without making any settings, the detection of the black bands works

Yes, you very likely could run it with a more powerful GPU, but I don't know the minimum GPU you would need.

We might get new HDR options at some point too, and they could be faster or slower, so GPU advice for madVR is somewhat tricky. Ideally get the fastest GPU you feel comfortable with and then tune your madVR settings for your GPU. That said, you should be able to get something running smoothly on your 580.

What are your HDR options now?

lello

18th January 2021, 16:31

this could be an issue with copyback check your PCI lanes.

I have absolutely no idea what it is :o:

These are the HDR settings:

https://i.ibb.co/mhHsNt4/HDR-settings.png (https://ibb.co/mhHsNt4)

https://i.ibb.co/chkZvB8/Trade-quality.png (https://ibb.co/chkZvB8)

If I select "compromise on hdr tone & gamut mapping accuracy", the rendering goes down and everything works, but I lose several things of the tone mapping

quietvoid

18th January 2021, 17:04

36-39ms per frame should be fine for 24p content. What are you using for dithering?
You could try cropping as well, that might help.

@huhn: UHD is supposed to be slower because tonemapping is done on more pixels than lower res.

Klaus1189

18th January 2021, 17:15

PCI lanes -> Check the connection of your GPU with your Mainboard. A mechanical PCIe x16 slot may only have 8 or maybe even only 4 lanes electrically. It is rare, but sometimes users do strange things when building computers.

What mainboard and GPU do you have?

lello

18th January 2021, 17:38

Asrock Z77 Extreme 6

Klaus1189

18th January 2021, 18:11

https://download.asrock.com/Manual/Z77%20Extreme6.pdf
On page 14 you can see the PCI and PCIe slots
1.) Which one do you use for your GPU?
2.) Do you have any other PCIe cards installed which can cause the x16 lanes to be split up to x8 and x8 to the GPU and the other PCIe-card? Can be a PCIe-SSD, ...
3.) What GPU do you use?

lello

18th January 2021, 22:06

1) PCIE2
2) No, I only have one Essence ST sound card installed on the PCI2 slot
3) AMD RX 580 nitro+ 4GB

Klaus1189

18th January 2021, 22:21

Since you have an AMD GPU, open AMD Radeon driver and look for info, there is a line which tells you how the card is connected, that means how many lanes are used, it should be „3.0 x16“

Asmodian

18th January 2021, 22:56

If I select "compromise on hdr tone & gamut mapping accuracy", the rendering goes down and everything works, but I lose several things of the tone mapping

This seems pretty conclusive, your GPU is too slow for your HDR settings. A faster GPU is the only fix.

I would not worry about the PCIe link; if it works with those trade quality for performance options enabled the problem is not the PCIe link.

huhn

18th January 2021, 23:41

there are big where PCIe issues cost unregulary performance you can see that when the first step and the last step have unusual high rendertimes.

which is kinda present here.

@huhn: UHD is supposed to be slower because tonemapping is done on more pixels than lower res.
with tone mapping yes.

Asmodian

19th January 2021, 00:09

there are big where PCIe issues cost unregulary performance you can see that when the first step and the last step have unusual high rendertimes.

Hmm... it would be good to check it but given how slow HDR could be on my 2080 Ti I still think it is simply that the GPU is too slow.

That motherboard and slot should be fine for 16x PCIe 3.0.

huhn

19th January 2021, 07:50

i'm not even question that it is much more likely that the GPU is just not fast enough.

the reason i would look into it is that blur dif is the first step and blur dif usually takes 0.3 -1.50 ms on this screen it's 4.85 ms.

on the other hand i don't even know how he got it to be the first step.

edit: and the 580 is a pretty fast card when NGU isn't used which is the case here.

lello

19th January 2021, 09:25

Since you have an AMD GPU, open AMD Radeon driver and look for info, there is a line which tells you how the card is connected, that means how many lanes are used, it should be „3.0 x16“

I checked and that's it

Klaus1189

19th January 2021, 18:23

Then, you won't have a PCIe bottleneck. I assume anything else. Sorry, but I can't diagnose anymore. Do some more tests.

Asmodian

19th January 2021, 22:15

Sorry, but I can't diagnose anymore. Do some more tests.

Does the GPU simply being too slow seem unlikely?

huhn

19th January 2021, 22:27

the old old screens showed something like 40 ms so a little bit to much and spikes in the 71 ms.

lello

28th February 2021, 10:40

I have a doubt.

Since black band detection is done by the CPU and not the GPU, could it be that my current Core i5 3300 is insufficient? So is it better to change the processor instead of the video card?
Could an i5 10400F fit?

I also tried to see the percentages of CPU utilization with DXVA2 copy back active, and I saw that with an HD movie the usage was 20% while with a BD UHD it went up to 70/80%.

Thanks for your help

Asmodian

28th February 2021, 11:08

Since black band detection is done by the CPU and not the GPU, could it be that my current Core i5 3300 is insufficient?

I do not think so. Black bar detection is very easy for the CPU.

Have you tried with it disabled?

lello

28th February 2021, 14:50

Certainly, these are the CPU usage data:

d3d11 native & black bar detection: CPU 20% rendering 38
Copy Back automatic & black bar detection: CPU 70% rendering 68
Copy Back automatic NO black bar detecion: CPU 38% rendering 68

manni500

18th March 2021, 18:06

Asmodian,

I have been using your config suggestions for some time and to very good effect; many thanks.

Meanwhile I have an issue where I wonder if it is related to madVR or to the videos shown.

Whenever I run a lower-resoultion video on this screen at its native resolution (3840x2160) it comes to judder, with or without smooth motion enabled.

However, the judder goes away when I reduce the resolution to 1920x1080.

Is there anything to be done ad madshi level ?

Thanks in advance

Asmodian

18th March 2021, 20:04

It is very likely that your GPU is not fast enough to run my settings at full resolution.

What GPU are you using?

What do you mean by lower resolution, anything less than 1080p?

manni500

19th March 2021, 00:07

Thanks for the prompt response.

By lower resolution I mean 720p or less; but even a 720p is not always giving trouble.

Just for clarification: I am not using either of the two settings.bin that you keep available for download (I tested with them, but couldn't get them to work) but rather the basic settings that you mention in the text on https://forum.doom9.org/showthread.php?t=171787

My graphics cards is not great; it's a Zotac GT 1030 2GB. But no problem with films of 1080p and up and sometimes even 720p.

Asmodian

19th March 2021, 01:05

The first thing I would do is make sure quadrupling is disabled.

If you use ctrl-J, do you see the average rendering times above the frame time?

manni500

19th March 2021, 13:58

Quadrupling is not enabled; on the scaling algorithms I use
- Bicubic60
- Bicubic150
- Lanczos 3 taps

On ctrl-J, rendering is shown as 27.01 and 34.75ms resp. (screenshot attached); this on a video where judder is indeed prevalent

manni500

19th March 2021, 17:23

ADDENDUM

An interesting development: the judder goes away when I take out HW Acceleration in the LAV Filter

This probably will have other ramifications but I haven't found those yet

Asmodian

20th March 2021, 00:29

Interesting, that is weird.

Do you have the frames presented in advance set to 3 or 1? (recommended)

Sticking to software decoding is a great option though, if you aren't trying to watch 4K HEVC on a lower end or older CPU.

manni500

20th March 2021, 14:16

Frames presented are set to 3, I will now test things with 1 and report back.

My CPU is indeed a bit of a grandfather, I will likely have to get a new motherboard sometime soon.

Still, that will not explain the judder question: When the low-resolution films run on the fairly new GPU, I have judder. But when they run on the old CPU, there is none

manni500

20th March 2021, 16:33

UPDATE

I fiddled around with the frames presented - 1, 3 or 8: no difference.

Meanwhile, I realized that there were a lot of frames dropped when showing in top resolution (3940x2160). This caused me to play with some of the madVR settings, and low & behold - no more or only very few frames dropped and no more judder!

Unfortunately though, I cannot put my finger on what exactly caused this improvement.

But if you have an interest, I will be happy to send you the new settings.bin file

doffen

4th November 2021, 01:04

I've tried to set up separate profiles for 23,976 and 24 fps content, but my rules seem to round up and treat both framerates as 24 fps.
Does anyone have rules to differ between these two or any idea about how I can make them?

QBhd

4th November 2021, 15:04

Not sure why you would need to differentiate between the two, the processing power used for both should be nearly the same. But if I was going to make rules <24 and <25 should do the trick

QB

doffen

4th November 2021, 18:51

it's about reducing framedrops.
with 23.976fps content and 59.94Hz I get 2:3 pulldown, but with 24fps content and 59.94Hz I get lots of dropped frames. If I increase the framerate to 60Hz I don't get as many dropped frames anymore.

Ive tried these three rules:
if (deintFps < 24) "23fps"
else "24fps"

if (fps < 24) "23fps"
else "24fps"

if (fps >= 24) "24fps"
else "23fps"

The result is always the same, and the 23fps profile (first profile) is chosen with 24fp content.

It's almost like MadVR rounds 23.976 up to 24 and can't differ between the two.

.m4jX

4th November 2021, 22:17

Try

if (fps<23.999) "23fps"
else "24fps"

QBhd

4th November 2021, 22:38

But for what you want, it is better to setup a proper 23.976 and 24.000 refresh rate

QB

huhn

6th November 2021, 12:54

it's about reducing framedrops.
with 23.976fps content and 59.94Hz I get 2:3 pulldown, but with 24fps content and 59.94Hz I get lots of dropped frames. If I increase the framerate to 60Hz I don't get as many dropped frames anymore.

no you will get 1 extra repeated frame every 1000 frames which should create a 2:2 cadance for that rare case and short case.

doffen

7th November 2021, 00:10

Try

if (fps<23.999) "23fps"
else "24fps"

Thank you soo much!
This ruleset works perfectly :D

no you will get 1 extra repeated frame every 1000 frames which should create a 2:2 cadance for that rare case and short case.

I'm not sure we agree (or maybe we do) but you made me read a bit more about cadence. If I understand it correctly I will get something like a 4:10 cadence with 23.976 content and 59.94Hz screen.

MadVRs dropped/repeated frames stay up in the hours range when I match it the way I want to.
I did a short test having it the "wrong way" and I got numbers wich add up to what you said: 1 frame repeat/drop every 1000 frames.
24.000fps 59.94Hz = 1 frame drop every ≈41s
23.976fps 60Hz = 1 frame repeat every ≈41s

@QBhd
Yes normally it would be better to setup a proper 23.976 and 24.000 refresh rate.
Because my TV doesn't accept 4:4:4 chroma at 23.976 or 24.000hz refreshrate I use 59.94 and 60Hz instead. I never imagined the TV would have a limitation like this, but the only thing I can do is to try and work my way around it ;)

huhn

7th November 2021, 01:41

the numbers for 3:2 judder are not calculated you have to add them which is 4.8 a sec so do you really think 1 more or less every 41 sec makes a real day night difference?

doffen

7th November 2021, 19:01

No, I don’t think it will make a night and day difference.
It’s still an improvement within the compromises I’ve chosen.
Do you think I shouldn’t make this (now automated) improvement because it’s not a night and day difference?
I’m just curious, I’m doing it like this until either my equipment or opinion changes ;)

huhn

7th November 2021, 19:06

you are just waisting your time instead of usinf one of the other ways to fix the issue compeltely.

doffen

7th November 2021, 22:18

Which is?
I want 4:4:4 chroma resolution to my TV, which doesn’t work on 23/24hz.
If there is a better way I’d love to learn it!

huhn

8th November 2021, 00:06

smooth motion (literally made for this exact case).
you can try your luck with a custom refreshrate.
reclock.

doffen

8th November 2021, 03:07

Thank you for your input!
I'm already using smooth motion (with the setting: only if there would be motion judder without it) Using the Ctrl-J stats I see it as enabled.
So even with smooth motion I see an improvement in dropped/repeated frames with changing the refresrate.

I guess I could try a custom refreshrate with reclock, but I can already send 23.976 and 24.000Hz to my tv, which makes me lose 4:4:4 chroma.
Using ≈ 60Hz (59.94 and 60.00Hz) I keep 4:4:4 chroma. Since this is a limitation in the display (as far as I can tell) I really doubt re-clock will do any difference.
What's your thougts on it?

huhn

8th November 2021, 05:40

with SM active the drop/repeat number is still calculated but will never ever happen this is not a joke you will not drop or repeat a frame for sync reasons. it should be remove so user don't get confused it's clearly miss leading.

for a custom resolution i would try 48 from the usual also PC mode 50. custom resolution that change the refreshrate are unlikely to work on modern TVs.

doffen

8th November 2021, 18:54

Well I've clearly been misled.
I tried 48Hz wich my TV didn't accept.
50Hz works. I'll try switching back and forth and see what I prefer.
I guess it might make it easier for smooth motion as well, since it's closer to double frame rate than 60Hz.

Asmodian

8th November 2021, 19:52

Well I've clearly been misled.
I guess it might make it easier for smooth motion as well, since it's closer to double frame rate than 60Hz.

For smooth motion the faster the better. You always want the fastest refresh rate you can get and do not care what the ratio between the frame rate and refresh rate is.

flossy_cake

27th March 2022, 03:58

[deleted - posted in wrong thread]

fotb

21st July 2022, 18:18

it works! thank you!!

1050 is enough? 1060 is better for scaling algorithms? memory of 3GB or 6GB are determinants in the results?

Hello all,

Is the HDR to SDR conversion setting only available on version 0.92.11? I am currently using test build 132 and I dont see that option.

Can HDR to SDR conversion be done with a string of rules? If so, can someone provide me with one?

Thanks in advance!

crossclass

19th August 2022, 19:21

Mr. Asmodian very thanks for your explanation, but if I may ask but one question:
Is it possible to make profile in madvr that only triggers when an online video is played through it? Say I have profile "1080p60" for offline hdd videos but I want a profile for youtube "1080p60YT" with different settings that only triggers for youtube 1080p 60fps videos?
esp. this video: https://www.youtube.com/watch?v=Bw7qa94P0ls
Thank you.

Asmodian

23rd August 2022, 01:26

You can use the rule:
if (filePath = "*youtube*") "online"

to activate a profile for anything from YouTube.

crossclass

23rd August 2022, 23:35

I am sorry but I could not get it to work, my script:
,, ,, ,,
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 61) && (!HDR) && ((filePath = "*youtube*") "online") "1080p60YT"
nor this:
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (src=online) && (deintFps < 61) && (!HDR) && (filePath = "*youtube*") "1080p60YT"
Thanks in advance.

crossclass

24th August 2022, 00:10

doffen

24th August 2022, 00:53

Have you tested with a line just like this:
else if (filePath = "*youtube*") "1080p60YT"

Just to make sure your other requirements aren't messing with you ;)

Asmodian

24th August 2022, 01:52

Sorry the "online" in "if (filePath = "*youtube*") "online" is the profile name right? My 1080p scripts are as follows:
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 31) && (!HDR) "1080p"
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 61) && (!HDR) "1080p60"
else if (filePath = "*youtube*") && (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 61) && (!HDR) "1080p60YT"
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 31) && (HDR) "1080pHDR"
What am I doing wrong?

You need to put the more restrictive conditions first. With your ordering anything 1080p and 30 fps or less will use "1080p" (and 60fps or less "1080p60") because as soon as a condition is valid it uses that one, without checking the rest. Only if the previous condition was false does the next condition get evaluated.

I suggest new ordering. You also don't need to check for not HDR in the last two conditions, because if it was HDR it would already have matched for 1080pHDR. I would also add a new profile for everything that doesn't match any of these because if no conditions match madVR will just use the first profile in the list, but you can point at a specific profile with an else condition if you want things that are 4K or over 61 fps to use a different profile.

if (filePath = "*youtube*") && (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 61) && (!HDR) "1080p60YT"
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 31) && (HDR) "1080pHDR"
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 31) "1080p"
else if (srcWidth <= 1925) && (srcHeight <= 1085) && (deintFps < 61) "1080p60"
else "fast"

QBhd

24th August 2022, 15:00

I'll say it again.... why include srcWidth... totally not needed.

QB

Grimsdyke

6th October 2022, 22:45

Hi Asmodian, do you maybe know if MadVR is aware of "DiscID" and could use it for profiles ?

Asmodian

7th October 2022, 03:01

No, madVR does not get the DiscID from the player so it has no way to use it in profiles.

You might be able to use filePath or fileName instead.

StrifeLeonhart

15th November 2022, 06:57

Hi Asmodian,

I'm using your Max Quality settings.bin for madVR in MPC-HC, so for that I thank you very much.

I would like to know if its possible to use your Max Quality settings.bin but only use CPU and no GPU at all or lowest possible, my PC is 14core and I tend to think its more than enough for a video player even on Max Quality.

Would you be able to make something like that work?

huhn

15th November 2022, 10:11

no that's impossible and even if the code would exist to run GPU code on a CPU a thread ripper would struggle to produce 1 frame a sec.

StrifeLeonhart

15th November 2022, 21:38

no that's impossible and even if the code would exist to run GPU code on a CPU a thread ripper would struggle to produce 1 frame a sec.

Sorry, I think didn't ask the question right.

What I'm trying to say is make the Load more on CPU instead of GPU.

Maybe that makes more sense I hope?

Asmodian

15th November 2022, 22:33

madVR's features either run only on the GPU (mostly) or only on the CPU (IVTC, black bar detection). There is nothing you can move from CPU to GPU or GPU to CPU.

StrifeLeonhart

16th November 2022, 17:28

madVR's features either run only on the GPU (mostly) or only on the CPU (IVTC, black bar detection). There is nothing you can move from CPU to GPU or GPU to CPU.

I understand, thank you very much

JNW

25th November 2022, 01:06

Not exactly what you wanted but best you could do would be to use software decoding. That will use your CPU instead of GPU.

Asmodian

25th November 2022, 19:11

But that really just adds work for your CPU without reducing the workload on the GPU. The GPU even has to do a little more work due to copying the decoded data from the CPU's memory to the GPU.

StrifeLeonhart

26th November 2022, 20:10

Not exactly what you wanted but best you could do would be to use software decoding. That will use your CPU instead of GPU.

My friend that is exactly what I was trying to explain to Asmodian, that is exactly what I wanted to do, thank you so much for saying what I wanted to say but didn't know how.

I don't mind about adding more work for the CPU, I'm just trying to make the CPU take the work load instead of the GPU.

Now I would like to ask Asmodian, which options would I need to go through the "MaxQuality" settings to use Software decoding.

huhn

27th November 2022, 05:23

you should stop ignoring him and understand that this will cost you GPU power...

Asmodian

27th November 2022, 23:18

I don't mind about adding more work for the CPU, I'm just trying to make the CPU take the work load instead of the GPU.

Using software decoding does not make the CPU take the workload instead of the GPU. It makes the CPU do a lot more work, while also making the GPU do more work.

I know it is counterintuitive, it seems like the CPU doing a lot more work would mean the GPU has less to do, but the way hardware decoding works on modern GPUs means that madVR has the most GPU power available when using native hardware decoding. The GPU has extra fixed function hardware that does the decoding directly into GPU memory which takes less GPU power than copying the data from system memory.

There are reasons to use software decoding but moving workload from the GPU to the CPU is not one of them.

Now I would like to ask Asmodian, which options would I need to go through the "MaxQuality" settings to use Software decoding.

You do not change anything in madVR, hardware v.s. software decoding is an option in LAV Video.

alexnt

9th December 2022, 15:32

Is it correct to use
1080p23, 1080p24, 1080p60, 1080p120, 2560x1440p23, 2560x1440p24, 2560x1440p60, 2560x1440p120
under "list all display modes madvr may switch to" ?

My monitor is 1440p@144 but I use it @120Hz
The content I watch on this monitor is mostly 23.976 & 24 fps.

huhn

9th December 2022, 22:18

alexnt

9th December 2022, 22:43

you sadly have to check the refreshrates manual 144 and 240 are usually 143 and 239 so a muliply of 23 refreshrate.

looking at all the modes provided maybe enough.
2560x1440p120 and 2560x1440p144

you can skip 144.

why would you ever want to switch to 1080?

Thanks for the reply!
Because movies are 1080 but as I understand from what you said this not a correct assumption.

I'll stick with 2560x1440p120

JNW

24th December 2022, 08:01

Using software decoding does not make the CPU take the workload instead of the GPU. It makes the CPU do a lot more work, while also making the GPU do more work. Verified with task manager.

I know it is counterintuitive, it seems like the CPU doing a lot more work would mean the GPU has less to do, but the way hardware decoding works on modern GPUs means that madVR has the most GPU power available when using native hardware decoding. The GPU has extra fixed function hardware that does the decoding directly into GPU memory which takes less GPU power than copying the data from system memory.

There are reasons to use software decoding but moving workload from the GPU to the CPU is not one of them.

You do not change anything in madVR, hardware v.s. software decoding is an option in LAV Video.

I'd never argue with you but not how it works on an Intel 11th Gen XE. Get better performance dropping to software. Depends on GPU most likely. Verified task manager. Same madVR, mpv.

EDIT: Merry Christmas Doom9 :)

Klaus1189

24th December 2022, 08:19

So what GPU do you use?

Asmodian

24th December 2022, 22:27

Get better performance dropping to software.

I am curious how you measured "better performance" along with which GPU you are using. Are you comparing to DXVA2 or D3D11, copyback or native?

Do you get lower GPU power draw when running software decoding?

The point I was arguing against with was the idea that you can free up GPU resources by using software decoding. Unless your GPU is pretty old, e.g. using a hybrid decoder, software decoding does not free up GPU resources. It actually uses more GPU resources due to the increased data transfer over the PCIe bus. This is usually a very minor effect with higher end GPUs; however, it is counter intuitive if you expect the CPU to have taken a lot of work from the GPU.

Sometimes you do get better performance overall with software decoding. This is because the GPU switches to a higher power (higher clock speed) mode due to the increased workload, which actually runs madVR better. The drivers don't seem to be that good at picking power levels for madVR's workload. I suggest using a mode other than "Optimal power" for Nvidia GPUs, unless power usage is very important to you.

zamadatix

4th September 2023, 02:49

Thanks for this thread, it helped me get IVTC working because I did not realize it would be broken with DXVA or D3D11 decoding (which is default in a lot of cases).

For those on MPC-BE you can configure this under View -> Options -> Internal Filters -> Video Decoders -> Video decoder configuration. In that window you'll find a "Hardware Acceleration" area. In my case, simply switching it to NVDEC works since I have an Nvidia GPU.

TheGameMaster

22nd October 2023, 04:14

Maybe someone here can help. My MadVr is stuck on super-xbr for doubling. I want to continue to use NGU-Anti Alias. But no mater what I set MadVr to, it still runs super-xbr. I am using the 169 beta release on an HDR system. I did a complete re-install with no change. Any ideas?

huhn

22nd October 2023, 12:06

you are mixing up chroma with luma?

TheGameMaster

22nd October 2023, 23:45

I set it to NGU Anti-Alias and it's wont' switch to it. Chroma upscaling is set to NGU Anti-Alias Low. I have had it set to Very High luma doubling, and medium luma quadrupling, and high chroma for years. But after this update, it switched, on it's own, to super-xbr when you do a ctrl-J to view it. Unless you do not double, regardless of setting in MadVR settings, it only uses super -xbr for doubling. Very annoying. I want it to work like it did. I did a complete re-install and went back to an older version of MadVR and no change, and that included deleting the setting.bin and the madvr reg settings in windows 11. It's really strange.

Asmodian

24th October 2023, 18:30

Can you post a screenshot of your settings pages?

TheGameMaster

4th November 2023, 06:13

Here are the two screen shots. The control-j and the upscale page. As you can see, it is set to NGU-Antialias in settings. But in control-J it shows super-xbr. I have uninstalled, cleared the registery, and reinstalled. Same issue. I even tried going back to an earlier version of MPC-BE and MadVR to see if that would fix it. But no luck.

TheGameMaster

9th November 2023, 02:53

The OS I was on was Windows 11 Pro 22H2 Beta channel. I upgraded to 23H2 Canary channel (in order to force the upgrade I had to switch channels). Post-upgrade, MadVR is now behaving normally again. It is no longer stuck on super-xbr. That said, forcing me to use super-xbr did open my eyes to how useful that code set can be. It does run a lot cooler with less dropped frames than NGU-Anti-Alias with my set up. I'm running two GPUs - Radeon RX 5700XT and a still supported Nivida GTX 750ti - on a Aurus 570 Master board and Ryzen 3900 12 core CPU with 64 gigs of RAM. I use the 750ti as my audio card and hardware video decoder (CUVID - in LAV Video with hardware deinterlaceing). It frees up the 5700XT to do the rendering and presentation. Anyway, it works now.

huhn

9th November 2023, 14:14

hardware decoding is actually "free".
the real cost is the upload of the decode data which is gigantic with a copyback decoder like cuvid a native decoder could be cheaper.

what so ever a 5700 xt doesn't really care i speak out of experience with it.

Charky

9th November 2023, 14:19

I'm running two GPUs - Radeon RX 5700XT and a still supported Nivida GTX 750ti - on a Aurus 570 Master board and Ryzen 3900 12 core CPU with 64 gigs of RAM. I use the 750ti as my audio card and hardware video decoder (CUVID - in LAV Video with hardware deinterlaceing). It frees up the 5700XT to do the rendering and presentation. Anyway, it works now.
Seems like an unnecessarily complicated setup.

EDIT : well, huhn beat me to it :D

mark0077

30th January 2024, 22:07

Just regarding this option

"calibration [disable calibration controls for this display]"

the guide mentions that if its set to disable calibration, that madVR defaults to assuming the TV is using REC 709. Is this still true though if the content is HDR and madVR is outputting in one of its HDR modes (like passthrough or pixel shader mode)j, or does madVR instead assume BT. 2020 in such cases?

huhn

30th January 2024, 22:16

in HDR output it will assume bt 2020 and PQ this also counts when output in HDR format is check in pixel shader else with pixel shader it will assume bt 709 and gamma 2.2.

mark0077

30th January 2024, 22:42

Ah right ok great, thanks. I wonder is it something that's worth updating in the main page of this thread, to explain the behavior? I saw some forum members on avsforum creating profiles for this, eg if (HDR) BT2020 else REC709, but I guess thats not needed in many cases.

huhn

31st January 2024, 02:18

the profile is usually to use the nvidia only feature to report bt 2020.

but if that option is used it will also assume bt 2020.
this is only an issue for PJ.

Asmodian

29th February 2024, 19:27

I wonder is it something that's worth updating in the main page of this thread, to explain the behavior?

I updated the first post. Thank you!

webbo

20th March 2024, 09:54

Definitely need some advice here, please

Regarding setting up madVR correctly in the following environment :
-ISF calibrated LG G3 OLED (calibrated both for SDR / HDR)

Should madVR run like this =>
tone map HDR using pixel shaders + output in HDR - so TV switches to its HDR mode?

Or just use tone map HDR - so the TV remains in SDR mode, yet madVR does all the tonemapping

Posting two screenshots

https://images2.imgbox.com/b0/28/s72wEbwv_o.jpg (https://imgbox.com/s72wEbwv) https://images2.imgbox.com/88/ec/sOhELYfP_o.jpg (https://imgbox.com/sOhELYfP)

Sorry if this is confusing, hope someone gets my point
Many thanks

JaylumX

21st March 2024, 14:29

Definitely need some advice here, please

Regarding setting up madVR correctly in the following environment :
-ISF calibrated LG G3 OLED (calibrated both for SDR / HDR)

Should madVR run like this =>
tone map HDR using pixel shaders + output in HDR - so TV switches to its HDR mode?

Or just use tone map HDR - so the TV remains in SDR mode, yet madVR does all the tonemapping

Posting two screenshots

Sorry if this is confusing, hope someone gets my point
Many thanks

You may want the latter if you like HDR.

Asmodian

21st March 2024, 19:11

HDR really needs that peak brightness. I would put the TV into HDR mode for HDR content.

jmonier

22nd March 2024, 12:45

HDR really needs that peak brightness. I would put the TV into HDR mode for HDR content.

It's my understanding that earlier LG OLED's DID provide extra brightness for HDR. I don't know for sure about other brands, but I've always thought that they didn't. I seem to recall that madshi stated that this was the reason for that checkbox, although he seemed to be relying on reports from others.

I DO know that current LG OLED's (like my C3) no longer do this. I also know that that it has no effect on my JVC HDR projector (RS3000), although it doesn't seem to hurt. I leave it off.

I would be interested in any comments from others who have direct experience.

huhn

22nd March 2024, 13:41

on a PJ outputting HDR as HDR is a waste of time.
just tone map it directly to SDR.

with a TV everything changes.
yes some TV can do high brightness in SDR or are even brighter in SDR then HDR in rare cases.
but that means you watch SDR at like 1500 nits that's gonna hurt.

according to rtings so not that trustworthy anymore... your C3 is dimmer in SDR then HDR by like 200 nits:
https://www.rtings.com/tv/reviews/lg/c3-oled

BTW. did they fix the PC mode chroma and it is back to 4:4:4 with newer OLED TVs? pretty sure Asmodian talk about this before could be compete BS i'm saying here.

Asmodian

22nd March 2024, 16:18

did they fix the PC mode chroma and it is back to 4:4:4 with newer OLED TVs?

SDR is 4:4:4 in PC mode on my C1. I don't have any LGs newer than that.

HDR tone mapping is noticeably worse in PC mode though, so for HDR I am not sure there is a good solution to preserve 4:4:4 (at least on my C1 I decided 4:2:2 was the less-bad tradeoff for HDR content).

huhn

22nd March 2024, 17:36

so there is nothing like reconstruction going on in PC mode and 4:4:4.

my CX didn't care about the PC mode HDR issues when it was new.
i never use HDR anyway.

acgm

16th May 2025, 15:27

Hello, I'd like to use the latest madVR test build. Could you provide the download link? Thank you!

huhn

17th May 2025, 00:28

https://www.videohelp.com/software/madVR

Bily

3rd June 2025, 20:29

Hello

Is there a configuration file that leaves all MadVR parameters for MPC-HC at maximum?

In which you just have to put the file in a specific folder?

I use an RTX 2060 and a 4K TV.
I have here: madVR v0.92.17

Thanks

huhn

4th June 2025, 01:42

your GPU can't do that.

Bily

4th June 2025, 01:52

your GPU can't do that.

But is there a basic configuration? That works for all types of VGA?

And how do I use this file?

Another question:

Is there a way to reset the settings?
Return the installation settings?

Thanks

huhn

4th June 2025, 03:20

But is there a basic configuration? That works for all types of VGA?
no
And how do I use this file?

you overwrie the settings.bin in the madVR installation folder
there are 2 here:
https://forum.doom9.org/showthread.php?t=171787

Is there a way to reset the settings?
Return the installation settings?
the "restore default settings.bat"
will do that

Bily

4th June 2025, 03:36

no

you overwrie the settings.bin in the madVR installation folder
there are 2 here:
https://forum.doom9.org/showthread.php?t=171787

the "restore default settings.bat"
will do that

you said:
you overwrie the settings.bin in the madVR installation folder
there are 2 here:

do both of them work for my video card?

or will the more powerful configuration crash my mpc-hc?
Is there a way to use the maximum configuration file by changing just a few parameters to make it compatible with the VGA RTX 2060?

thanks a lot

huhn

4th June 2025, 12:22

i don't have your GPU i can not say but it was created with a 2080 ti in mind so pretty sure that's a no.

is it possible to change a couple parameter and make it work yes.

Bily

5th June 2025, 00:22

Hi

I made a very good configuration.

But a problem arose:

The image in most of the movies is "dirty", I think it is grainy.

Is there any way to fix this? To have a very clean image?

Thank you very much

huhn

5th June 2025, 00:34

denoise can be done by using reduce random noise.
it will destroy the image and is very expensive to do.

a grainy image with no applied sharpening filter should be just fine.

Bily

5th June 2025, 17:01

Is there a way to have crystal clear/very clean images when upscaling 1080p to 4k?

It could be through other filters or shaders

Thanks

Sunspark

5th June 2025, 17:48

Does your TV have a decent scaler built-in?

Bily

5th June 2025, 17:53

I don't know.
What is this?

Bily

5th June 2025, 18:10

Is there any configuration for MADvr that is essential to have better upscaling quality?
a simple option . as which is the best : image upscaling and its adjusters

My VGA is RTX 2060 6Gb

Thanks

huhn

5th June 2025, 19:37

i would say it is NGU AA but there is a reason this forum has rule 12.

there are way to many variables to make a general statement.

Bily

5th June 2025, 22:33

In movies with 480p to 720p, the upscaling of both MADvr and FSRCNNX is very bad.

Is there any way to improve it?

I have a TCL 4k TV that is doing better upscaling than them.

Is this normal?

Is there any specific upscaling program for 480p and 720p videos? preferably for mpc-hc?

Another question:
Is there any filter, shader, etc... that improves the quality of files with deteriorated video?

Thanks

Thanks

huhn

6th June 2025, 03:00

Bily

6th June 2025, 16:08

make an image example so i have at least an idea of what you are talking about.

NGU sharp on it own and FSRCNNX are horrible with images with compression artifacts. NGU sharp can be used with compression artifact reduction in madVR.

There was a mistake here: the 720p source was in very poor condition.

One question: when Windows is configured in 4K and connected to the TV via HDMI, is it possible to play a 1080p video on the MPC-HC and have the HDMI output send it to the TV in 1080p? So that the TV can do upscaling?
Or do I have to configure Windows to 1080p for that?

Thanks

Sunspark

6th June 2025, 17:07

Yes. What you want is for the player to change resolution. I do this. So what you do is select the resolution and refresh rate in the player options to take effect when you go into fullscreen mode. This way the 1080p signal is sent to the display and then the display scales it to whatever the panel resolution is.

The reason this works is because if the source video is 1080p and a PC display resolution change is to 1080 to match it, then there is no need for image scaling up or down from the perspective of the PC. It only scales the chroma to 4:4:4.

Bily

6th June 2025, 17:24

Is it possible to configure it so that whenever the video has: 480p or 720p resolution, to send via HDMI, it changes to 1080p?
How?

Is this activated when we put it in full screen?

Thanks

Bily

7th June 2025, 03:28

i would say it is NGU AA but there is a reason this forum has rule 12.

there are way to many variables to make a general statement.

In MadVR, is there any automatic feature that reduces the upscaling quality if the VGA cannot process the image? Changing to a lower quality to suit the VGA's processing power?

Thanks

QBhd

7th June 2025, 11:19

In MadVR, is there any automatic feature that reduces the upscaling quality if the VGA cannot process the image? Changing to a lower quality to suit the VGA's processing power?

Thanks

You make profiles... for example here is my Upscaling profiles:

if (srcHeight < 721) and (deintFps < 25) "720p23"
elseif (srcHeight < 1081) and (HDR) "1080p23 HDR"
elseif (srcHeight < 801) and (deintFps < 25) "1080p23 35mm Film"
elseif (srcHeight < 1081) and (deintFps < 25) "1080p23"
elseif (srcHeight < 2161) and (deintFps < 25) "2160p23"
elseif (srcHeight < 721) and (deintFps < 26) "720p25"
elseif (srcHeight < 1081) and (deintFps < 26) "1080p25"
elseif (srcHeight < 2161) and (deintFps < 26) "2160p25"
elseif (srcHeight < 721) and (deintFps < 31) "720p29"
elseif (srcHeight < 1081) and (deintFps < 31) "1080p29"
elseif (srcHeight < 2161) and (deintFps < 31) "2160p29"
elseif (srcHeight < 721) and (deintFps < 51) "720p50"
elseif (srcHeight < 1081) and (deintFps < 51) "1080p50"
elseif (srcHeight < 2161) and (deintFps < 51) "2160p50"
elseif (srcHeight < 721) and (deintFps < 61) "720p59"
elseif (srcHeight < 1081) and (deintFps < 61) "1080p59"
elseif (srcHeight < 2161) and (deintFps < 61) "2160p59"

QB

Asmodian

7th June 2025, 17:10

Nothing truly automatic, you need to figure out what is too much and set the profiles up accordingly.

Bily

10th June 2025, 02:36

Is there a way to divide the processing work of MadVR between: VGA, processor and RAM?
Or is it only possible to use VGA?

Is there any way to know how much processing power each MadVR resource uses? Any graphics?

Thanks

huhn

10th June 2025, 03:47

you add an empty folder in the madVR installation folder and name it showrendersteps

now madVR control+j will show what takes what.

no you can not split stuff everything that runs on the GPU run on the GPU and everything that needs the CPU will use the CPU. CPU can not do these types of calculation efficiency anyway and even if it would be possible to split moving data from and to the CPU cost more then you would get out of it.

Bily

10th June 2025, 17:39

Sunspark

10th June 2025, 18:36

You would benefit from reading various guides or general material to get a better understanding of things.

Then learn by doing.

huhn

10th June 2025, 19:16

AS FOR: DISPLAY MODES:

Is it appropriate to customize? Or is the native configuration that comes with it sufficient?

How do I find out which ones my TV supports?

And where do I have to configure them: in the tab: display modes or custom modes?

Thanks

by default it does nothing.
the TV supported more are hard because nearly ever TV report the wrong refreshrate which is not native but support 4096x2160.

the custom tab is for something different.

Bily

10th June 2025, 21:16

Is there any way to display these numbers in graphs?

or some program that makes a graph of the VGA operation and saves this graph for us to view after we finish watching a video?

Thanks

huhn

11th June 2025, 01:40

it is not needed madVR load is linear 5 sec tells you everything you need to know from just one frame from of the OSD.

Bily

11th June 2025, 16:17

Why does the screen flicker when we open a 4K video in mpc-hc madvr? When we switch to full screen and lower resolutions, this doesn't happen?

Thanks

huhn

11th June 2025, 19:01

the switch to HDR trigger a resync.

real FSE will also do that so it changing the resolution.

Bily

11th June 2025, 19:25

I installed here: msi afterburner
and noticed that with my configuration the GPU usage is at 80%

What is the maximum percentage of GPU usage you recommend for use with madvr?

Thanks

huhn

11th June 2025, 20:38

there is no general answer to that. as long as madVR isn't dropping frames actively it is fine.

Bily

12th June 2025, 19:43

How to remove compression noise (macroblocks) in madvr? Or by another resource of mpc-hc?

And other types of noise, what is the most appropriate way to remove them?

Thanks

huhn

12th June 2025, 22:11

remove compression artifacts and remove random noise should be the stuff you are looking for.

they are very expensive to use.

Bily

12th June 2025, 22:57

Is there any shader or filter for mpc-hc that does this well and is lightweight?

Another question :
Is there any feature, in VGA or MADVR, that reduces the image quality of the video over time? Even if the options in the folder: trade quality for performance are turned off?

Thanks

huhn

13th June 2025, 06:02

over time no.

Bily

15th June 2025, 20:27

Here the quality drops after 30 minutes of filming.
The GPU usage does not exceed 88%.
Is there any configuration I can do so that the image quality does not drop over time?
Thanks

huhn

16th June 2025, 03:31

filming?
there is nothing in madVR, mpv or any other renderer i know that can drop quality over time.

Gladzilla

7th July 2025, 23:03

I installed here: msi afterburner
and noticed that with my configuration the GPU usage is at 80%

What is the maximum percentage of GPU usage you recommend for use with madvr?

Thanks

Not sure but the 2070 super I am using hits 90+ % usage. I cannot tell a real difference when using madvr.

Sunspark

16th March 2026, 17:14

I was taking screenshots last night in D3D9 Overlay using the player's screenshot key shortcut.

When it is written that it is not possible to take screenshots in overlay, do you mean taken from within madvr or taken from the player and/or desktop?

nevcairiel

16th March 2026, 19:07

From the desktop. The renderer can clearly still take screenshots, and the player may leverage the renderers function - depending on if thats hooked up.

Although its unclear of Overlay is even still quite as overlay as it used to be, as these legacy rendering modes get deprecated over time.

huhn

16th March 2026, 19:56

madVR did not implement screenshots for overlay

Sunspark

16th March 2026, 20:32

It must have been a silent update.. in .92.17 this is working. I am able to change screenshot settings inside madvr and it affects the output. But it's weird right? I was under the impression an overlay bypasses a lot of stuff.

Unless, it switches out of overlay, takes the screenshot, and then switches back into overlay automatically? I do have Fullscreen optimizations disabled. Win 10.

I will try taking one with the HUD showing just to see if it's switching out of the mode, however, it's possible it's not quite what it used to be as Nev said, because looking at the refresh rate and the composition rate, it's different than what I'd see if it was in exclusive mode, so that implies it is going through a compositor. Maybe.

nevcairiel

16th March 2026, 23:36

Screenshots from the renderer should really just be independent of the output, anything else would be weird. The renderer has the RGB image already, why would it not just write that into the screenshot output rather then somehow actually screenshot the rendering window?

huhn

17th March 2026, 01:32

he said back in the day that's like overlay rendering original from xp and the original xp version couldn't do screenshoot too or soemthing.
this overlay actually renders a blackish image.

and yes it should do it.it just as it is.

the screenshoot function in madVR works different not all settings are honoured that may work.

nvidia should also not just kill it entirely and instead should have marked it as deprecated. it is as it is.

x7007

4th August 2026, 02:21

Hey please help someone with your smart thinking because this is very hard for me to even understand, I have no idea what could activate it, but only when test, but I don't know what the disadvantage of doing that could be.

I am using Potplayer + LavFilters + Madvr, I have Samsung S90C connected to 5090, The S90C has Always ON GSYNC FullScreenOnly selected. I need it to use 144Hz and I don't want to switch between them because I have games in windowed mode while watching movies.

It seems everything fine when I launch the movie and don't touch anything, frames would be 144hz or 138hz as NVCP global limiter and won't move
The problem is when using Madvr when the settings I am using when moving foward/backfoward the Gsync start to act up and would flicker none stop while fps goes 130-45 between all those ranges and it won't stop until I relaunch the movie and it would be fine again.

Using any other renderer, built in or such won't cause that, it works 144hz all the time.

The options I had it used are
Hey please help someone with your smart thinking because this is very hard for me to even understand, I have no idea what could activate it, but only when test, but I don't know what the disadvantage of doing that could be.

I am using Potplayer + LavFilters + Madvr, I have Samsung S90C connected to 5090, The S90C has Always ON GSYNC FullScreenOnly selected. I need it to use 144Hz and I don't want to switch between them because I have games in windowed mode while watching movies.

It seems everything fine when I launch the movie and don't touch anything, frames would be 144hz or 138hz as NVCP global limiter and won't move
The problem is when using Madvr when the settings I am using when moving foward/backfoward the Gsync start to act up and would flicker none stop while fps goes 130-45 between all those ranges and it won't stop until I relaunch the movie and it would be fine again.

Using any other renderer, built in or such won't cause that, it works 144hz all the time.
Is there anything else I should look for to make it 138hz or 144hz ? will i lose anything not selecting those options?
it happens 100% when I have game in windowed mode and I just put open it and minimize it, then the movie just lose it, even those settings won't help

The options I had it used are
https://i.imgur.com/9MBZLVk.png

while I did a quick test it seems it works fine and doesn't happen
https://i.imgur.com/uNpmWW6.png