madVR - high quality video renderer (GPU assisted)

[aufkrawall]

Funny: Measured chromatic coordinates of basic colors by Spyder 4 basically match those included in manufacturer's EDID of the device in question. So much for Spyder 4 measuring just garbage, but maybe just a weird coincidence...
Yes, theoretically lots of other colors could be way off. Practically, I'd be surprised if calibration with a different device would look obviously different with SDR content...

[purplebaby]

How can I make video playback take up less GPU %? When playing 1080p with MPC-BE and madVR, I see the GPU being utilized at 30-40% in task manager.

[Schwartz]

- Use less demanding upscalers / image doublers. Jinc and NGU are more expensive.
- Disable artifact removal / enhancement settings.
- Use a dithering method other than Error Diffusion .
- As a last resort, check some "Trade Quality For Performance" options.

30-40% GPU isn't bad if you're getting something out of it.

[purplebaby]

Is there a decoder I can use that utilizes CUDA better or something? I remember there was something called CoreAVC around a decade ago.

[LoRd_MuldeR]

Is there a decoder I can use that utilizes CUDA better or something?

https://github.com/Nevcairiel/LAVFilters/releases

https://i.imgur.com/kRDQRaI.png (https://i.imgur.com/7WfotLf.png)

Though, for decoding video streams, we don't actually use CUDA (or OpenCL), because CUDA (just like OpenCL) is an API for running arbitrary computations on the GPU's "shader" cores.

But, modern GPU's already have dedicated hardware decoders for the most widely-used video formats (AVC, HEVC or AV1), which don't even use/strain the actual "shader" cores and which are accessed via DXVA, D3D11VA or VDPAU.

(Besides that CUDA is Nvidia-only, whereas DXVA/D3D11VA works with AMD or Intel GPUs too!)

I remember there was something called CoreAVC around a decade ago.

CoreAVC Decoder is dead (https://pbs.twimg.com/media/EGcAIDGWwAIXuZt?format=png) for a very long time now!

Also FFmpeg's (libavcodec's) software decoder is now faster than CoreAVC ever was, at least on any somewhat recent CPU. And for hardware-accelerated decoding, you certainly don't need CoreAVC. Just use LAVFilters.

I see the GPU being utilized at 30-40% in task manager.

So? Use Ctrl+J to show the statistics. As long as you don't see "dropped frames" increase continuously (a dropped frame every now and then is kind of normal) everything is probably fine!

...except if your goal is minimizing your electricity bill ;)

[purplebaby]

https://github.com/Nevcairiel/LAVFilters/releases

https://i.imgur.com/kRDQRaI.png (https://i.imgur.com/7WfotLf.png)

Though, for decoding video streams, we don't actually use CUDA (or OpenCL), because CUDA (just like OpenCL) is an API for running arbitrary computations on the GPU's "shader" cores.

But, modern GPU's already have dedicated hardware decoders for the most widely-used video formats (AVC, HEVC or AV1), which don't even use/strain the actual "shader" cores and which are accessed via DXVA, D3D11VA or VDPAU.

(Besides that CUDA is Nvidia-only, whereas DXVA/D3D11VA works with AMD or Intel GPUs too!)



CoreAVC Decoder is dead (https://pbs.twimg.com/media/EGcAIDGWwAIXuZt?format=png) for a very long time now!

Also FFmpeg's (libavcodec's) software decoder is now faster than CoreAVC ever was, at least on any somewhat recent CPU. And for hardware-accelerated decoding, you certainly don't need CoreAVC. Just use LAVFilters.



So? Use Ctrl+J to show the statistics. As long as you don't see "dropped frames" increase continuously (a dropped frame every now and then is kind of normal) everything is probably fine!

...except if your goal is minimizing your electricity bill ;)

Using the LAV filters didn't help at all. Yes, I made them preferred and they were in use.

Minimizing power usage and making sure the fans stay off is a priority.

I've gone back to MPC-BE's decoders and switched the preferred decoder from D3D11 to NVDEC. This seems to help a bit.

[sexus]

sup , anyone know if theres a way to have BT2020 be forced upon starting media playback, i know theres a hotkey for forcing BT2020 primaries , but id like it to autoforce it, appreciated

[doffen]

There's a checkbox "report BT.2020 to display under calibration in the settings you can activate.
Or are you thinking about profile based automatic bt.2020 selection?

[LoRd_MuldeR]

Using the LAV filters didn't help at all. Yes, I made them preferred and they were in use.

Minimizing power usage and making sure the fans stay off is a priority.

Did you enable "Hardware Acceleration" in LAVFilters settings and ensure that "Active Hardware Accelerator" shows that it is actually in use? May require to re-open the media to take effect.

Also, in what way it "didn't help"? Be aware that a hardware-accelerated video decoder offloads computations from the CPU to the GPU (hardware decoder). So this will probably reduce the CPU usage and it thus may also reduce the overall power consumption of the system, but it certainly does not reduce the GPU usage or the GPU's power consumption. After all, you are offloading more of the work to the GPU and its hardware decoders.

[aufkrawall]

Minimizing power usage and making sure the fans stay off is a priority.

Then CUDA/NVDEC is a no-go for you, as it forces GPU + VRAM into P2 state, which means high minimum clocks. On the other hand, that might not be that kind of a big loss, as dynamic clocking with Ampere GPUs is still garbage.

[purplebaby]

Then CUDA/NVDEC is a no-go for you, as it forces GPU + VRAM into P2 state, which means high minimum clocks. On the other hand, that might not be that kind of a big loss, as dynamic clocking with Ampere GPUs is still garbage.

Hmm. :/ I was not aware of this. Good point.

What's my best option then? LAVFilter video decoder with hardware decoding: DXVA2?

[huhn]

Did you enable "Hardware Acceleration" in LAVFilters settings and ensure that "Active Hardware Accelerator" shows that it is actually in use? May require to re-open the media to take effect.

Also, in what way it "didn't help"? Be aware that a hardware-accelerated video decoder offloads computations from the CPU to the GPU (hardware decoder). So this will probably reduce the CPU usage and it thus may also reduce the overall power consumption of the system, but it certainly does not reduce the GPU usage or the GPU's power consumption. After all, you are offloading more of the work to the GPU and its hardware decoders.

it actually can even lower GPU usage.
the decoder including cuvid are ASIC so as already pointed out they lower CPU usage but can do that for GPU too.
if you don't use them you have to upload giga bytes and giga bytes of decoded data this is work for GPU and may be more work then decoding the file it self here it only has to upload very small encoded files.

these are on by default anyway so there is no work needed that has to be done and are very small work compared to most things madVR does.

so if you want lower power consumption you should lower madVR settings as already pointed out.

[sexus]

There's a checkbox "report BT.2020 to display under calibration in the settings you can activate.
Or are you thinking about profile based automatic bt.2020 selection?

yeah that dont work , im talking about the function ....forcing primaries to BT2020, currently it only works per hotkey , theres no autoforce feature , the report feature dont do that i already have it enabled , BUT what i did was make myself a autodetect .vbs/batch/AHK script combo that autoforces BT2020 primaries as soon as madHcCtrl.exe aka media playback is detected then stops its combo forcing progress and starts from beginning once madHcCtrl.exe is detected again ...meaning once playback ends madHcCtrl.exe is ended as you know, and when you then playback another media madHcCtrl.exe is ran again and thus this works 100% reliable .

[doffen]

Do you have movies with wrong metadata for the primaries? or does that function expand the primaries to bt.2020?
If you have send bt.2020 set I can understand that you would want to expand the primaries but doesn't this introduce banding?

I use this ruleset:
if (HDR) "hdr"
elseif (gamutindex = 3) "dci-p3"
elseif (gamutindex = 4) "bt2020"
else "standard gamut"

Each profile has its own 3d-lut (display profiled with different brightness) and the standard gamut has the send bt.2020 box unticked.

If I'm misunderstanding I'd love to be corrected :)

[sexus]

Do you have movies with wrong metadata for the primaries? or does that function expand the primaries to bt.2020?
If you have send bt.2020 set I can understand that you would want to expand the primaries but doesn't this introduce banding?

I use this ruleset:
if (HDR) "hdr"
elseif (gamutindex = 3) "dci-p3"
elseif (gamutindex = 4) "bt2020"
else "standard gamut"

Each profile has its own 3d-lut (display profiled with different brightness) and the standard gamut has the send bt.2020 box unticked.

If I'm misunderstanding I'd love to be corrected :)

well ...nope ...BT2020 on a BT2020 supported Display of course....8bit bitdepth is minimum tho ...no banding and yes that function literally forces the media to be displayed in BT2020 which expands colors aka improves output by BT2020 standards also ...not many older media has BT2020 its usually BT 709, works on HDR as well on SDR content, i dont need millions of different profiles either ...i use 1, never had an issue with that, now for actual HDR 4k 10bit playback .....you of course will need some custom profile separate ...much like you did yes, what i was talking about was just about purely forcing BT2020 primaries

[LordX2]

I am confused - BT.2020 primaries aren't loaded when 4K HDR files are played? Is this something I have to load, or a setting I have to have?

I currently have Display Calibration set to BT.2020 when watching 4K HDR, and I select send BT.2020 to display.

Is there anything else I should be doing to ensure the full BT.2020 is being processed and sent to the display?

[huhn]

nothing needs to be changed.

[brazen1]

Does anyone here use a panel? Also, do the PJ's used here have separate calibrated memories applied based on the various sources used? In other words, do PJ's behave like panels do? When I render SDR for instance, the panel recognizes SDR metadata and uses a unique calibration I have set for SDR. When I render HDR, it recognizes that metadata and from memory uses a completely different unique calibration mode. These HDR vs SDR are separate memories although they share the same HDMI input port (HDMI 1). Same goes with using other HDMI input ports. They too have their own memories. Cable TV (HDMI 2) for example uses yet another unique calibration and Games (HDMI 3) another and OTA TV (HDMI 4) another, etc.

Because every source is calibrated differently at panel level, I use many profiles in madVR fine tuned to each one. IE: DCI-P3 for HDR and BT.709 for SDR BT.1886 and so on. A panel is not limited to one calibration must fit all and unsure if other displays like common mainstream PJ's are? Also, I use madVR profiles for many other things other than HDR vs SDR calibrations:

Color Saturation - I find (some) titles are mastered and/or rip undersaturated. Rather than manually apply it at panel level (as it should be) using the madVR adjustment works very, very well. It's uniquely applied per title using small keywords and a profile for them (Sat, Sat 25, Sat 50, and even Sat 100). A good example of an undersaturated title is Letters from Iwo Jima (2006).

Gamma - Some titles are simply too dark or too light. It might be how they were intended but I intend to view them differently - keywords to identify them as well.

Image enhancements - I apply unique tiny bits to sources less than my native 2160p resolutions. After upscaling 1080p to 2160p and adding a little Crispen, Thin, and engage Anti Ringing, I have a hard time telling if I'm looking at native 2160p SDR or upscaled 1080p SDR... which is gooooooood. Even 2160p vs 1080p vs 720p vs 540p have sub categories further separating and distinguishing amongst them such as titles with keywords.

Scaling - I use a ton of unique profiles mainly based on resolution but keywords too. Basically because I use an older GPU to squeeze everything I can out of it depending on the demand saving me from needing to upgrade to something stronger. Oddly, a 60fps HDR title like Gemini Man can handle more robust settings such as NGU High vs a 1080p title that use High demanding more of the hardware decoding because there is more to process so there's some give and take my little GPU endures.

All of this, and much, much more is automated thanks to my panel, madVR, and other things - I just press play. I'm curious if others use madVR similarly or if LUTs, scripts, and whatnot are workarounds because of limitations and/or preferences? I'm not a PJ owner so know not what they commonly offer or don't? Often I read there is no need for multi profiles be it at display level or madVR processing and I can't help but wonder how all these vastly different sources must look all adjusted the same?

Lastly, I share and inquire because I RARELY read much regarding madVR and panel users here or elsewhere. Surely panel users understand madVR is valuable for them and it's not just a PJ thing... or maybe they don't? ​

One more thing... HAPPY NEW YEAR y'all :)

[LordX2]

Update in terms of Nvidia drivers: 497.29 (the latest release as of this post) FIXES the odd 30 second sync issue. So someone at Nvidia was paying attention, and now I am no longer driver locked at a previous version!

[chros]

All of this, and much, much more is automated thanks to my panel, madVR, and other things - I just press play. I'm curious if others use madVR similarly or if LUTs, scripts, and whatnot are workarounds because of limitations and/or preferences? I'm not a PJ owner so know not what they commonly offer or don't? Often I read there is no need for multi profiles be it at display level or madVR processing and I can't help but wonder how all these vastly different sources must look all adjusted the same?

Lastly, I share and inquire because I RARELY read much regarding madVR and panel users here or elsewhere. Surely panel users understand madVR is valuable for them and it's not just a PJ thing... or maybe they don't?
Of course we do :)
I only use profiles for mostly scaling (see the link in my signature), and use my small remote(s) (https://forum.doom9.org/showthread.php?t=175007) to manually set something if necessary (e.g. black/white levels in madvr, or switching presets/settings in the LG TV).

In addition, I watch SDR like this (https://forum.doom9.org/showthread.php?p=1947198#post1947198) on Oled :)

[el Filou]

My impression is PC games are mostly 2.4, though of course there are examples pointing the other way or having terrible clipping also with 2.2 due to bad tone mapping/lack of indirect light etc.PC games are typically NOT graded to ANY-Standard.Sorry for commenting on old replies, but PC graphics are supposed to use sRGB, so (SDR) PC games are also supposed to use that (and its gamma is different from pure 2.2 at the beginning of the curve)sup , anyone know if theres a way to have BT2020 be forced upon starting media playback, i know theres a hotkey for forcing BT2020 primaries , but id like it to autoforce it, appreciatedRename your video file like this: "<video name> primaries=bt2020 matrix=bt2020.<extension>"I have resorted to setting about -150ms of delay, but if the issue is present only in exclusive mode madVR, kiiiinda would expect to solve it madVR side. I don't want to do audio sync guesswork.I don't think madshi will fix anything for Windows 7 anymore.

[aufkrawall]

Sorry for commenting on old replies, but PC graphics are supposed to use sRGB, so (SDR) PC games are also supposed to use that (and its gamma is different from pure 2.2 at the beginning of the curve)
Really doubt that, that looks just bad in basically every single PC game. There likely also is no difference to console games, and I strongly believe those expect typical TV display gamma (~pure power 2.4).

[chros]

Really doubt that, that looks just bad in basically every single PC game. There likely also is no difference to console games, and I strongly believe those expect typical TV display gamma (~pure power 2.4).
Pure power" only exist if you have 0 blackk, otherwise it's scaled gamma already.

[aufkrawall]

Pure power" only exist if you have 0 blackk, otherwise it's scaled gamma already.
I'd be really surpirsed if most (cheap) TVs would take actual black level for different brightness settings into consideration with BT.1886. It's probably just a rough estimate (or worse) or when it offers to switch between BT.1886 and pure power gamma 2.4 settings, there probably will be no visual difference.

[Siso]

What gamma in a SDR 3dlut should be chosen, if I'm using tone map HDR to SDR via pixel shaders? 2.2 or 2.35 or 2.4 relative?

[huhn]

for SDR 3D LUT using HDR -> tone mapping you have to target 2.2 for mathematically correct results.

[Siso]

for SDR 3D LUT using HDR -> tone mapping you have to target 2.2 for mathematically correct results.

I see, thanks huhn.

[el Filou]

Really doubt that, that looks just bad in basically every single PC game. There likely also is no difference to console games, and I strongly believe those expect typical TV display gamma (~pure power 2.4).What would be the reason for that? Do PC developers use TVs to test their games and not PC monitors? Or do they happen to use PC monitors that default to TV gamma (which would be a very strange thing)? Or do they take the time to recalibrate their PC monitor to TV gamma?
Maybe what you're seeing is only applicable to console games ported to PC?

[huhn]

because it is objectively wrong for console games.

we are talking about computer games here they may not calibrated at all.

on the other hand we "know" adobe RGB calibrated devices are gamma 2.2.

i have not seen a single test where a pre calibration monitor has a gamma of 1 at the start of the dark parts. on the other hand the testing isn't done correctly to show it you have to measure the gamma at 0 and a little bit above to show it is a straight line in that part without that it just looks like a bt 1886 curve.

[tp4tissue]

Games usually do not go through a rigorous color grading process. They also have many different technologies to target simultaneously, IPS / VA / OLED, no fixed gamma would be correct across such a wide range.

The assets in the game are made by different people under different conditions, so it's a mixed bag.

You can put gamma to any value where the shadow quality looks good to you. THE IMPORTANT thing is that you're using RESHADE, as it supports 3DLut. Correcting monitor RGB through the gamma table causes tints if the monitor is not well behaved.

But why wouldn't you use reshade, the performance cost is significantly less than 1% for great color.

[aufkrawall]

If you don't care about lowest possible delta values for every single color, this tool also has developed fantastically: https://github.com/ledoge/novideo_srgb
Works extremely well for my wide gamut monitor, including gamma + grayscale calibration (no more Windows GPU gamma ramps banding screw-ups). I suppose LG OLED TVs with internal crap processing would fare worse without full 3D LUT processing though.
ReShade unfortunately doesn't work in all games (anti-cheat) and also causes a bit of banding with 3D LUT, as its internal color precision apparently isn't high enough and/or it doesn't use dithering. This can be mitigated by forcing 10 bit frame buffer, still a bit of a hassle. But yeah. Without the tool linked above, I'd still use 3D LUT via ReShade wherever possible. Wide gamut display oversaturation with sRGB/rec709 content looks so bad, it should be illegal.

[tp4tissue]

I haven't observed any "additional" banding with reshade that isn't already present in the source material or the poor handling of most non-pro display devices.

Oversaturation isn't a huge problem unless it's very uneven or starts really early where it affects skintones too much.

I've watched most movies using the saturation preset in dispcal, it exaggerates the saturated colors while keeping skintone "acceptable" it is a bit too hot for some movies, "most movies" look ok.

White people mostly look fine

Indians and Black people can look a little too red at times.

The East Asians they put in the movie are very light skinned so no huge impact.

South Asian films with darker-Asians can actually look too yellow/saturated.

So it depends on the material.

If you're talking about oversaturation of Objects, I actually prefer Optimus Prime to look Extra red and Extra blue.

NO sense in paying for all this wide-gamut and not using it.

[mpchelp10]

I am encountering a very unique bug with HDR passthrough and I'm wondering if anyone could share some insight.
My laptop has one of those "8+2bit HDR" screens with 2 GPUs (iGPU + dGPU). When I play HDR content with MPC + madVR, HDR colors only show when the player is either windowed or fullscreen but not in focus. If I enter fullscreen, the colors display as very greyed and washed out.

MPC is windowed - I get HDR colors.
MPC is fullscreen - I get washed out colors.
MPC is fullscreen, but I right click to bring up the context menu - I get HDR colors.
MPC is fullscreen, but I mouse over to the bottom of the screen to bring up controls/seek bar - I get HDR colors
MPC is fullscreen, but I alt tab and focus onto a different program - I get HDR colors

[The Luggage]

I’m using a HTPC with Madvr, a JVC projector and a HD Fury Vertex2. The Vertex can show the details of the incoming signal and I use the macro function of the Vertex to switch the projector in the correct user mode (SDR-Rec709 or SDR-BT2020 or HDR). I’m using Kodi DS-player to play content on the HTPC

Lately I’m having an issue with MadVR and seemingly my HTPC in general. It seems like my HTPC always sends a BT2020 signal/flag with SRD content and I do not know how to stop this. I did not have this problem before. I do not know what change of settings caused it.

At first I thought it was limited to SRD-Rec709 content played with the DS-player: the Vertex said it was receiving SDR-BT2020 content (and therefore it kept the projector in SDR2020 mode and did not switch my projector to the SDR-Rec709 mode).

But the problem seems to occur all the time. Not only when using Kodi DS player but in my windows desktop or windows explorer or when playing SDR-709 content with a player that does not use MadVR. All the time the Vertex2 says the incoming SDR signal is BT2020.

When I send a content with a HDR flag, then the Vertex does detect this and my projector switches to HDR mode. For instance when activating the HDR-button in the Windows settings or when selecting pass through HDR in the “hdr” tab of the MadVR-settings. But all content that is not flagged as HDR, is considered BT2020 even though (I think) it is Rec709.

In the calibration tab of MadVR, I made two profiles (Rec709 and BT2020) and a rule to make sure that the “Report BT2020 to Display” check box is only enabled with HDR-material. I do not think there is anything wrong with the rule. And manually switching between the profiles (with a shortcut key) does not solve the issue. Just to make sure, I also unchecked the “Report BT2020 to Display” in my BT2020 profile, but this also does not solve the issue.

I tried updating my Nvidia drivers and have updated MadVR to the latest beta (141) but this also has no effect.

The problem does not seem to be related to the Vertex. If I use my UHD-Blu-ray player to play a regular Blu ray (Rec 709) the Vertex detects Rec709 and sends the macro to the Projector to put in in SDR709 mode.

Does anyone have a suggestion to fix this?

I only have very limited knowledge of MadVR and a HTPC, so I might be missing something obvious.

Am I right that when just looking at the windows desktop or windows explorer this should be Rec709 and not BT2020?

Thank you in advance.

I already asked a similar question in the MadVR support thread on AVS, but the suggested solutions did not solve the issue:

https://www.avsforum.com/threads/madvr-player-support-thread.2215490/post-61310577

[clsid]

I am encountering a very unique bug with HDR passthrough and I'm wondering if anyone could share some insight.
My laptop has one of those "8+2bit HDR" screens with 2 GPUs (iGPU + dGPU). When I play HDR content with MPC + madVR, HDR colors only show when the player is either windowed or fullscreen but not in focus. If I enter fullscreen, the colors display as very greyed and washed out.

MPC is windowed - I get HDR colors.
MPC is fullscreen - I get washed out colors.
MPC is fullscreen, but I right click to bring up the context menu - I get HDR colors.
MPC is fullscreen, but I mouse over to the bottom of the screen to bring up controls/seek bar - I get HDR colors
MPC is fullscreen, but I alt tab and focus onto a different program - I get HDR colors

In the devices section of madvr settings, set "native display bitdepth" for your device to 10bit.

If that does not help, then change HDR mode in madvr from "let madvr decide" to "tone map HDR using pixel shaders" + "output video in HDR format".

[IngramAU]

Have ran into an interesting issue with madvr dropping out bitstream audio. After trying nearly everything I seem to have landed on the Error Diffusion option 1 or 2 causing it for some strange reason. I am using 24Hz 12Bit RBG over HDMI enhanced to an AVR/Sony HDR TV. MPC-HC latest, latest nvidia, D3D11 renderer, wasapi is handling bitstream out.

I'm wondering if with error diffusion 1/2 I'm maxxing out the HDMI cable and it's dropping audio as a result. It seems DTSHD/DTS doesn't have this issue but happens frequently with DD+ 5.1 audio.

I can use PCM 5.1 and encounter no issues as a work around.

I'll keep testing to make sure this specific option is what is causing the bitstream drop outs.

I was convinced it was a Windows issue and was going to format and start from scratch!:confused:

EDIT:

Well plot thickens. It seems the issue happens with Jinc for chroma and Error Diffusion 1 or 2 on my 4k HDR files which don't need upscaling besides the chroma. I've changed to NGU Sharp Med/High and Error Diffusion 2 and so far seems to be resolved. Glad to have gotten to the bottom of this one.

[aufkrawall]

I haven't observed any "additional" banding with reshade that isn't already present in the source material or the poor handling of most non-pro display devices.

I can easily observe it in old Mirror's Edge, it gets noticeably worse with ReShade 3D LUT inject with 8 bit frame buffer.
Also, most LCD displays shouldn't introduce additional banding anymore these days.


Oversaturation isn't a huge problem unless it's very uneven or starts really early where it affects skintones too much.

It affects everything and even slight tints by e.g. color grading get annoyingly obtrusive.


NO sense in paying for all this wide-gamut and not using it.
Geez. You can't make proper use of wide gamut when the source isn't explicitly using it (which is ~never the case for SDR content). You just get shifted chromatic coordinates: Saturated red will turn pinkish everywhere and blue turquoise etc.. This is just garbage...

[mpchelp10]

In the devices section of madvr settings, set "native display bitdepth" for your device to 10bit.

If that does not help, then change HDR mode in madvr from "let madvr decide" to "tone map HDR using pixel shaders" + "output video in HDR format".
Native display bitdepth is already set to 10bit
HDR mode was set to "passthrough HDR" with "send HDR metadata" checked.
Changing it to "tone map HDR using pixel shaders" + "output video in HDR format" results in the same issues.

[DMU]

my laptop has one of those "8+2bit hdr" screens with 2 gpus (igpu + dgpu)
Specify, please, which GPUs are used in your laptop?

[ashlar42]

This might also happen if you use an old version of LAV Filters (pre-0.74) and madVR has dropped its previously accepted values for PQ HDR in recent development versions.
Some time ago, madVR introduced a custom value for PQ HDR signaling, but that value then collided with the official value for HLG when it was introduced, so you need to make sure all parts are up to date.
Found this as I was looking for why a file with this mediainfo was not activating HDR on my LG OLED (which supports HLG AFAIK):

Video
ID : 1
Format : HEVC
Format/Info : High Efficiency Video Coding
Format profile : Main 10@L5.1@Main
Codec ID : V_MPEGH/ISO/HEVC
Duration : 58 min 0 s
Bit rate : 18.6 Mb/s
Width : 3 840 pixels
Height : 2 160 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 25.000 FPS
Standard : Component
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 10 bits
Bits/(Pixel*Frame) : 0.090
Stream size : 7.52 GiB (99%)
Default : Yes
Forced : No
Color range : Limited
Color primaries : BT.2020
Transfer characteristics : BT.2020 (10-bit)
transfer_characteristics_Original : HLG / BT.2020 (10-bit)
Matrix coefficients : BT.2020 non-constant

I use LAV 0.75.1.17 and a still active madVR beta (I think the last one available).

[mpchelp10]

Specify, please, which GPUs are used in your laptop?
integrated graphics is Intel UHD 630
discrete graphics is GTX 1050 Ti MaxQ

[huhn]

Found this as I was looking for why a file with this mediainfo was not activating HDR on my LG OLED (which supports HLG AFAIK):

it does not matter that your TV support it the file it is HLG and madVR and windows, nvidia/amd or both does not support it yet.

mpcVR can convert it to HDR10 which madVR should do in my opinion too but doesn't.

[janos666]

I started to have some stuttering issues lately. I am currently on Win11 (22000.434) and nVidia 511.17. I tried to go back to an older nVidia driver (497.29) but it didn't make a difference.
The video playback is smooth while the media player is in windowed mode. Stuttering starts when it enters borderless windowed fullscreen mode and the surface switches from 8 to 10 bit. Bringing up the GUI (the seek bar) drops it back to regular 8 bit windowed mode and playback is smooth again. Once the GUI disappears and 10 bit is engaged again, the stuttering comes back.
I think I had this issue a while back (maybe 1 year ago) but it was so long ago that I can't remember what solved it. But back then it was enough to pause and resume the playback a few times after the transition to 10 bit mode and then playback remained smooth (until another transition or pause/play cycling, at somewhat random). This time pause/play doesn't help.
What could be the problem? (The hardware and the MB UFEI Setup settings didn't change for months.) Should I go back even further with the nVidia driver? Or is it a Win11 bug? (This was already present with the previous Win11 CU, .376.)

[clsid]

To fix stuttering:

* madvr settings > general settings > windowed mode / exclusive mode > frames presented in advance = 1

* NVIDIA settings > Power Management Mode = Adaptive / Normal
* NVIDIA settings > Vertical Sync = Adaptive
* NVIDIA settings > GSync = Disabled globally (use profiles for games)
* NVIDIA settings > disable Multiplane Overlay (https://nvidia.custhelp.com/app/answers/detail/a_id/5157)

[ashlar42]

it does not matter that your TV support it the file it is HLG and madVR and windows, nvidia/amd or both does not support it yet.

mpcVR can convert it to HDR10 which madVR should do in my opinion too but doesn't.
Oh, ok. So that's madVR's limitation. Well, at least I know what's not working. Thanks for your answer. :-)

[Manni]

To fix stuttering:

* madvr settings > general settings > windowed mode / exclusive mode > frames presented in advance = 1

* NVIDIA settings > Power Management Mode = Adaptive / Normal
* NVIDIA settings > Vertical Sync = Adaptive
* NVIDIA settings > GSync = Disabled globally (use profiles for games)
* NVIDIA settings > disable Multiplane Overlay (https://nvidia.custhelp.com/app/answers/detail/a_id/5157)

Thanks. Not everyone has all these options though. For example, in NCP 8.1.961.0 that comes with the latest 511.09, I can't find the multiplane overlay option in my nVidia settings and I can't set Gsync to disabled globally (I can only set monitor technology to fixed refresh instead of Gsync compatible).

[IngramAU]

Hmm I'm still experiencing bitstream audio drop outs. Have just changed HDMI cables hopefully that's the cause. Weird I've never had this sort of problem before.

[janos666]

To fix stuttering:

* madvr settings > general settings > windowed mode / exclusive mode > frames presented in advance = 1

* NVIDIA settings > Power Management Mode = Adaptive / Normal
* NVIDIA settings > Vertical Sync = Adaptive
* NVIDIA settings > GSync = Disabled globally (use profiles for games)
* NVIDIA settings > disable Multiplane Overlay (https://nvidia.custhelp.com/app/answers/detail/a_id/5157)

Well, thank you for the tips but installing 511.23 somehow fixed the stuttering issue (without applying any of the changes you listed).

Hmm I'm still experiencing bitstream audio drop outs. Have just changed HDMI cables hopefully that's the cause. Weird I've never had this sort of problem before.

Try to update the nVidia driver (unless you have an AMD card). There is an audio issue mentioned in the latest changelog.

[70MM]

Sorry but I havent checked this thread for a very long time....
Im on version madVR v0.92.17 which was the last one, will this still be safe to use, I heard v0.92.13 was the safest since there are no more updates?

[ryrynz]

There aren't any deal breaking bugs in the latest version that would make you want to use an earlier version. Use .17