madVR - high quality video renderer (GPU assisted)

[yok833]

We have those turned off because we want our picture to match how the director of the movie wanted them to be. Once you start changing settings adding dynamic contrast etc you begin to ruin the experience of getting lost in the film experience. This is because the atmosphere the director is trying to create with lighting is lost.
I totally agree with you about respecting the atmosphere of the director.. but is it incompatible ? Depending of your TV or your source, a noise reduction, a sharpen filter or a dynamic contrast algorithm (specially cooked by madshi!!)... can maybe help in getting closer of the original atmosphere?

[huhn]

I totally agree with you about respecting the atmosphere of the director.. but is it incompatible ? Depending of your TV or your source, a noise reduction, a sharpen filter or a dynamic contrast algorithm (specially cooked by madshi!!)... can maybe help in getting closer of the original atmosphere?

for sure not.
if you go for original you create a 3d lut and get correct colors and disable nearly everything in your tv.

sharpen is total bull if you like original. do you really think they can't sharp there own source bd before encoding it?

[AngelGraves13]

Ok, cool... thought you might have had a version of Smooth Motion that eliminated all blur.

Some movies look great with it, like anything with shaky cam.

The grain filter I can live without. I don't even use Dynamic Contrast, but it would be cool to have the option.

[THX-UltraII]

will madVR benefit a lot when i would switch from an HDD to SSD?

[ryrynz]

Your whole system benefits from upgrading to an SSD, madVR benefits most from GPU performance and drivers but everything else helps.

[James Freeman]

will madVR benefit a lot when i would switch from an HDD to SSD?

The single most important thing I did for my systems speed and responsiveness was SDD.

Everything from boot time, loading software, OS functionality, Internet Browsing, etc...
Everything is just fast, and no HD noises !!

Note that software should be installed on the SDD for best responsiveness,
but media (Movies, Music, Games, Photos, Archives, Projects) should be on a separate big HDD.

Come on.. HDD is like Vinyl, spinning parts, magnetism, noise, etc.
Its 2014, do yourself a favor. :D

[andybkma]

will madVR benefit a lot when i would switch from an HDD to SSD?

I didn't notice any difference with mVR quality-wise when I switched to SSD this year, but of course your media player will be faster (as long as its installed on the SSD), hence faster loading/switching of clips

[LigH]

How much does a renderer access a disk? I would hope "not at all"; maybe except for loading data once in the beginning.

[Razoola]

I guess where an SSD would be good is if there was some king of processing done on a video file before it stated playing. Apart from that the advantage of SSD is limited to noise and access speed only. IE a HTPC would boot faster with an SSD and you would not hear HDD grinding away. I have had my OS on an SSD for a few years now.

[MokrySedeS]

How much does a renderer access a disk? I would hope "not at all"; maybe except for loading data once in the beginning.

Some time ago madVR started to drop frames on my system. It worked fine before so I started to investigate.
I've noticed that madVR is writing a lot to my ssd (I don't remember exactly but it was well over 100 MB/s).
It turned out to be caused by a gpu ram deficiency (forgot to increase it after a bios reset). My guess is that madVR had to use a pagefile.
I was kind of glad that I don't have the fastest and greatest ssd on sata3 because then I probably wouldn't have noticed :devil:


Anyway... for madVR it doesn't matter if you use ssd or hdd. Your player will start faster but that's about it. (Unless you're low on gpu ram :rolleyes: but ssd is definitely not a solution)

[Megalith]

This is probably a silly question, but the scaling options available have no effect on video quality if you're playing the file back in its original dimensions and aspect ratio, correct?

[SamKook]

Right, except for the chroma one on most video.

[magus]

My guess is that madVR had to use a pagefile.

Does madvr 100% require a pagefile? Since I have 16gb ram. I am trying to use my computer without a page file as many people have sucessfully started to do. I just hope it doesn't cause conflicts with some software that requires a page file regardless of the amount of ram you have. I recently had a BSOD but I couldn't figure out if it was related to the page file being off or not because it has not happened again.

[michkrol]

Does madvr 100% require a pagefile?

99% sure you don't need pagefile.
That said I'd suggest you set a pagefile of like 512MB (minimum size = maximum size) for emergencies, like you need memory dump after BSOD or you have an application that checks for pagefile and not necessarily uses it.
I have it set to 1GB here and I'm not seeing any significant activity on it and I would notice the extra writes on my SSD.
PS. I'm assuming you're using Windows 7 or newer, which have better memory management than XP.

[LigH]

Do we talk about required "work memory" (RAM) or "video memory"?

[michkrol]

Do we talk about required "work memory" (RAM) or "video memory"?

I think mainly system RAM, but actually both, depending which post you're referring to.
Starting with Windows 7 (or Vista?) you get some of your RAM dynamically allocated as video memory (on both integrated and dedicated GPUs alike). When you set madVR's queues too high you can run out of dedicated video memory and Windows decides to use this dynamic memory, taking part of RAM for this. Sometimes things go awry while this happens and Windows uses pagefile while it shouldn't or is forced to use pagefile, if you're out of free RAM and video memory, but need to allocate more dynamic video memory in RAM (probably what happened for MokrySedeS).

Anyway I still say madVR doesn't require pagefile, but having a small pagefile (~512MB), even with loads of RAM doesn't hurt, since it's left unused (especially on Win7+), but can help in borderline cases.

We're getting a little sidetracked with this SSD discussion and all. :rolleyes:

[NicolasRobidoux]

...
Sorry, but I can't take any website serious if they claim that Bicubic looks better than NNEDI3. That's more than ridiculous.
...
They used Y-PSNR. Any "image quality comparison" based solely on PSNR (in one channel no less) is likely to be less than informative unless the "experiment" is very carefully put together. In addition, the results of re-enlargement tests depend more than a bit on the detail of how the test image was downsampled. See, for example, http://web.cs.laurentian.ca/nrobidoux/misc/AdamTurcotteMastersThesis.pdf.

P.S. I see that http://forum.doom9.org/member.php?u=166497 essentially made the same observation.

[LigH]

A computer cannot measure how annoying a difference feels. Neither PSNR nor SSIM metrics are close to your subjective opinion.

[mindbomb]

I have a question, does the bitrate of a video impact rendering performance? And is scaling harder on certain content than others or is it uniform?

[truexfan81]

I have a question, does the bitrate of a video impact rendering performance? And is scaling harder on certain content than others or is it uniform?

if you have a powerful enough gpu it should not matter, but yes 59.94fps requires twice as much power as 29.97, as for scailing well its easier to scale 720p up to 1080p than it is to scale 1080p down to 720p. i hope that helps

[michkrol]

I have a question, does the bitrate of a video impact rendering performance? And is scaling harder on certain content than others or is it uniform?

Bitrate and codec used only impact decoding performance. When using software decoding (CPU) performance impact is quite obvious, but hardware decoding (DXVA/CUDA/QuickSync) is (almost) unaffected by bitrate.
Renderer gets decoded (uncompressed) frames, so what matters for scaling/rendering is the frame resolution, target (screen) resolution and framerate, as mentioned by truexfan81. Another thing that needs more power is interlaced content, but then it's important "who" does the deinterlacing (decoder/renderer) and how (software/hardware), so it's a longer story.

[Soukyuu]

Bitrate and codec used only impact decoding performance. When using software decoding (CPU) performance impact is quite obvious, but hardware decoding (DXVA/CUDA/QuickSync) is (almost) unaffected by bitrate.That heavily depends on the components though. I had a clip that rose to 100MBit on some fine floating particle scene and that killed the CUVID decoding on my 260GTX. Using CPU (Phenom II X4@3.4GHz), it was getting a good load, but nowhere near 80% and of course no stuttering.

I'd say, with modern components, you can safely go with CPU decoding, as it's most compatible with various video parameters. I recently had artifacts when seeking a video file in DXVA mode, but they were gone in software decoding mode (CPU)

[G_M_C]

That heavily depends on the components though. I had a clip that rose to 100MBit on some fine floating particle scene and that killed the CUVID decoding on my 260GTX. Using CPU (Phenom II X4@3.4GHz), it was getting a good load, but nowhere near 80% and of course no stuttering.

I'd say, with modern components, you can safely go with CPU decoding, as it's most compatible with various video parameters. I recently had artifacts when seeking a video file in DXVA mode, but they were gone in software decoding mode (CPU)

Plus that software decoding has the advantage of leaving the GPU 100% free for madVR :)

[DragonQ]

Plus that software decoding has the advantage of leaving the GPU 100% free for madVR :)

Makes no difference in my experience.

[Soukyuu]

Again, heavily depends on components. I can't use Jinc3 + CUVID on my 260GTX, but CPU decoding + Jinc3 works fine.

[huhn]

Makes no difference in my experience.

makes a huge one with intel.

intel gpu usage rise highly with dxva and high bitrate content

[6233638]

Plus that software decoding has the advantage of leaving the GPU 100% free for madVR :)I frequently find the opposite to be more useful - hardware accelerated decoding (DXVA2-cb) means that video playback is a lot less likely to be interrupted by high CPU usage in background tasks.

I do prefer the quality of software decoding (it has far better error handling) but the majority of what I watch is just Blu-ray or DVD discs, so that rarely matters.

[THX-UltraII]

i just did a fresh system installation because I bought a SSD 1GB disk. When i start a movie it runs fine but as soon as i right click the screen or go to windowed mode the screen goes black/blank but i can hear that the content is still running 'normal'. when i crtl+alt+del and exit the ctrl+alt+del menu the content is visible again.

any idea if this could have something to do with thw ssd? both mpc-be+madvr and the content are on the ssd. I also noticed that it takes LONGER before the content starts running in mpc-be then it did on my regular hdd.

[michkrol]

any idea if this could have something to do with thw ssd? both mpc-be+madvr and the content are on the ssd. I also noticed that it takes LONGER before the content starts running in mpc-be then it did on my regular hdd.

It has nothing to do with SSD, unless the drive is broken.
If playback starts longer, but everything else works correctly, your system is broken or misconfigured. If whole system is slower, replace the SSD on warranty :(
Is the system you're using the same or did you update it too? Did you install all Windows Updates, newest drivers, DirectX and so on? Also check the settings, especially "use windowed overlay" and "use separate device".

[DragonQ]

makes a huge one with intel.

intel gpu usage rise highly with dxva and high bitrate content
Hmm. Only my laptop using DXVA2 Native or software decoding makes no difference to what settings I can use in MadVR.

[huhn]

play 4k samples with evr and you will see what i mean.

or even better use directx checker and benchmark you will see what i mean in gpu-z. i'm tlking about 80-100% gpu usage with evr how to use that with madvr. i tested this with a hd4000.

[James Freeman]

@madshi, everyone.

Ever thought about 3:2 Pull-Down Removal in MadVR?

I play NTSC DVD's at 29.97Hz on a 60Hz display and it "Judders" because of the 23.976 to 29.97 telecine conversion.
It'll be much nicer experience to undo the 3:2 pull-down and play the resulting original 23.976fps with Smooth motion or 24p/72Hz display.
Unless of course the original content is shot in 29.97 fps.

Any thoughts?

[nevcairiel]

@madshi, everyone.

Ever thought about 3:2 Pull-Down Removal in MadVR?

I play NTSC DVD's at 29.97Hz on a 60Hz display and it "Judders" because of the 23.976 to 29.97 telecine conversion.
It'll be much nicer experience to undo the 3:2 pull-down and play the resulting original 23.976fps with Smooth motion or 24p/72Hz display.
Unless of course the original content is shot in 29.97 fps.

Any thoughts?


madVR already supports inverse telecine for such files, all you need to do is turn it on manually for now.

Press "Ctrl+Alt+Shift+T" to turn it on, iirc.

[James Freeman]

madVR already supports inverse telecine for such files, all you need to do is turn it on manually for now.

Press "Ctrl+Alt+Shift+T" to turn it on, iirc.

It works!!

I turned off the option "Disable automatic source type detection".
so it automatically put the mode on "Video" instead of "Film".

Let me understand this:
Video Mode: Deinterlace + True 29.97fps, (TV content).
Film Mode: Deinterlaced + Inverse Telecine (autodetect cadence) to 23.976 fps.

[DragonQ]

Film mode is reverse 2:3 pulldown ("IVTC") or weave deinterlacing depending on the source (24/25/30 fps). Video is "full" hardware deinterlacing (typically 50/60 fps). If you don't have any sources that need IVTC (e.g. you live in a PAL region) then you can leave it on Video mode permanently if your GPU can detect 2:2 cadence correctly, e.g. modern nVidia cards.

[huhn]

Let me understand this:
Video Mode: Deinterlace + True 29.97fps, (TV content).
Film Mode: Deinterlaced + Inverse Telecine (autodetect cadence) to 23.976 fps.

nope

video mode inputs the telecine stream to the gpu and that deinterlace the stream. the deinterlacer can be bob, motion adaptive and many more

film mode use ivtc and no deinterlacing at all so if it fails (cadence break) it weaves this is called a deinterlacer but the same happens when deinterlacing is disable and no ivtc is used so call it what you like. weave is doing nothing at all just treating a 60i source as 30p.

Film mode is reverse 2:3 pulldown ("IVTC") or weave deinterlacing depending on the source (24/25/30 fps). Video is "full" hardware deinterlacing (typically 50/60 fps). If you don't have any sources that need IVTC (e.g. you live in a PAL region) then you can leave it on Video mode permanently if your GPU can detect 2:2 cadence correctly, e.g. modern nVidia cards.

it works (from time to time) with 4:2:2:2 too and 3:2/2:2 mix content is working too when refresh rate is higher than or 30 fps so it is still field matching and field matching is not weave.

i was able to watch tv streams with 2:2 content and 3:2 with smoothmotion at 60 hz but currently the smoothmotion part is not working for me in this case.
@ madshi:
thnks to smoothmotion i have an idea to improve ivtc a lot.
the real idea of ivtc is to get smaller encodes and judder free/correct motion.
but with adding a bob filter for cadence breaks, smoothmotion and timestamps fixes with ivtc we can theoretical watch all telecine content or all broadcast with one new mode.

a 60 hz smoothmotion, ivtc(~30i look ahead) with bob for cadence breaks and 1:1 content like soccer and co. smoothmotion can handle VFR content perfectly i tested it with a self created VFR video game recording with 16-120 fps.
i'm not a profi with 1:1 content i normally don't watch things like that but bob should be the best way to treat this content or am i wrong about that? at the moment ivtc is run on the cpu (i like that a lot!) and madvr can't handle in stream resolution changes in time but this shouldn't matter if simple bob resize is run on the cpu (not the aim of madvr i know).
in worst case we get 3:2 bobed judder thanks to the bob filter with hard to render 60 fps on 3:2 content with 2:2 or 1:1 moving captions but this should be still better than cadence breaks with visible weave.

sound like hell to program... and it's an feature request X-)

[NicolasRobidoux]

Speaking of cubic interpolation [rule 13] (Eyldebrandt at http://forum.hardware.fr/hfr/VideoSon/Traitement-Video/tutoriel-ffdshow-avisynth-sujet_138425_1.htm#t1815467)
I could not figure out what made the author like this cubic, and a few years ago I would have said that this couldn't be given that Avisynth uses it in a tensor way, instead of EWA, for which Robidoux was designed. I also suspect that the author would like Mitchell-Netravali just as much. Still a :) (for me).

[renq]

MadVR wont go to FSE on a 1080p50 TV. In FS, it stays in windowed mode and says that target rectangle is 1280x720:(

[huhn]

MadVR wont go to FSE on a 1080p50 TV. In FS, it stays in windowed mode and says that target rectangle is 1280x720:(

the target rectangle is player related. check your media player settings

[renq]

MPC 1.7.1.271
FSE works on my 1680x1050p60 lcd monitor

[0rion42]

Hey guys, I have a few questions. Hopefully these haven't been asked over and over again...even with search, this thread is massive.

How does the quality of madVR compare to that of a cheap hardware blu ray player? What about compared to an expensive hardware blu ray player? Do these typically use good rendering algorithms, or are they typically made to be cheap and simple?

Also, does anyone have a link to an explanation or a paper on the Jinc algorithm? I'm the most interested in this because it seems to be the highest quality, at least for chroma subsampling and image upscaling, but I can't find any documentation on what the algorithm is. I don't mind math-heavy sources.

EDIT: Also, for chroma subsampling. Is this purely an operation performed RGB -> RGB after the conversion from YUV -> RGB, or is the YUV -> RGB conversion part of chroma subsampling? I've read conflicting things about this.

Thanks.

[baii]

I believe The difference for a typical mastered blu-ray(24p movie) played on a 1080 p display would be minimal(ignoring features like smooth motion,color management) unless there is some thing go really wrong with the player.
The algorithm should be somewhere middle in the thread, I guess if you search the credit, you will get there faster.

[G_M_C]

Hey guys, I have a few questions. Hopefully these haven't been asked over and over again...even with search, this thread is massive.

How does the quality of madVR compare to that of a cheap hardware blu ray player? What about compared to an expensive hardware blu ray player? Do these typically use good rendering algorithms, or are they typically made to be cheap and simple?
[...]


I believe the difference lies mainly in the chipset used in the hardware player. The chipset in an Oppo BDP-93 will produce notability better image-quality and will have more features then the chipset in a sub 150 dollar B or C-brand player. How the different chipsets work internally is difficult to find out. At least; I tried for my oppo bdp-93, but i couldn't find enough info.

I use my Oppo and my PC side-by-side, as the PC plays formats the Oppo doesn't (in my case mostly >8 bit H.264/AVC streams). When I compare image quality between Oppo and PC/madVR, I dont see much difference between them to to be honest. But comparing based on visual appearance is very subjective (it's almost like a 'what's best' kind of question :p )

[James Freeman]

Can anyone explain to me how LAV Video Filter and MadVR Deinterlacing interact?

When I'm watching DVD's,
I turn Off Deinterlacing in MadVR but the deinterlacing is still active until I turn it off in LAV.
Is LAV deinterlacer bypasses (overrides) the MadVR turned off deinterlacer?

Is MadVR deinterlacer is needed only in IVTC when Film Mode is active?

Also,
MadVR fails to automatically detect between Film (29.97 telecined from 23.976) and Video (true 29.97).
It always plays the DVD's as "Video" even if its 23.976p telecined "Film".
When forced to "Film Mode" with IVCT (right for this DVD), Smooth Motion fails to engage on a 60Hz display.
EDIT: Confirmed again and again.

Moreover, when I play a true 29.97 Video DVD (documentary) and switch to Film mode I see combing and
the cadenses go crazy, switching between 2:2, 3:3:2:2, 4:2:2:2 and more...I think it depends on the content.
This crazy cadence switching also happens in menus, warnings, promos, in Film DVD's, which results in minor "freezes" and non smooth playback.
In this case (true 29.97) I have to (force)/(stay at) Video mode.

One more thing,
MadVR should show the actual fps that it delivers after all this processing.
Useful with Reclock also.

[DragonQ]

MadVR doesn't automatically use IVTC yet. You have to force film mode for this.

As I understand it, MadVR will tell the GPU to perform deinterlacing in the highest quality possible if necessary. The only time LAV Filters does the deinterlacing is if you specifically tell it to (e.g. when using CUVID or if YADIF is enabled). In these cases, the deinterlacing is done before MadVR receives the video.

[goribigon]

hi, when using ctrl+j it says display 59.998~ hz but not 60hz as the screen rate. is this normal readings ?

p.s. sometimes it shows 60.000~ hz right when i start the video, and then it goes to 59.99~ hz.

[James Freeman]

MadVR doesn't automatically use IVTC yet. You have to force film mode for this.

As I understand it, MadVR will tell the GPU to perform deinterlacing in the highest quality possible if necessary. The only time LAV Filters does the deinterlacing is if you specifically tell it to (e.g. when using CUVID or if YADIF is enabled). In these cases, the deinterlacing is done before MadVR receives the video.

Well that explains a lot, thanks.

I still think Smooth Motion should turn On automatically when in Film Mode and IVTC cadence 3:2 (29.97 -> 23.976).
Also, Film Mode breaks cadence too often, sometimes every scene in some TV documentary (true 29.97) DVDs.
Is there an option to freeze the switching?
Or when 3:2 detected to just "stay at" without switching anymore.

Most true 29.97 DVD's don't have 3:2 cadences unless some trailer or short clip with movies in them.
Will it be safe to "stay at" 3:2 cadence in Film mode once it's detected, and not "go crazy" every non 3:2 Piracy Warning?

For example in a Film DVD:
The first WB (warner bro's) logo is a short (5 seconds) 3:2 video in my DVD.
Warnings and Menu is 2:2.
Main movie is also 3:2.

Why not just stay at 3:2 the first time its detected?

In a Documentary DVD the Cadences are all over the place, because the cuts are shot with different cameras
at different fps, there are still images and many more "non linearities".

So stay at 2:2 would be perfect.

hi, when using ctrl+j it says display 59.998~ hz but not 60hz as the screen rate. is this normal readings ?

Yes.
It's even better then some.
Most people get 0.01 precision, you get 0.001, rejoice.

[goribigon]

thank you :o

[0rion42]

I've been doing some more reading, and from my understanding YUV -> RGB is a lossless conversion, which would mean that chroma scaling is done because of the subsampling after the conversion takes place (ie when it's stored as YUV 4:2:0 or w/e), which makes sense.

So, this means that 4:4:4 YCrCb would require no subsampling and has a lossless conversion back to RGB, right?

Also, does the interpolation occur in the YUV colorspace before conversion back to RGB, or does it occur after conversion to RGB? I'm guessing in YUV since that's what humans are more sensitive to.

I also had a question about the exact function of MadVR. I guess I don't have a clear idea of the pipeline of how media gets decoded and displayed. Does MadVR handle the actual processing of the h264 video, or does it only take over after the h264 has been decoded into the raw YUV values?

I guess my current mental model for the video pipeline is something like this:

h264/AVC Media file -> extract out h264 video -> decode the h264 -> present MadVR with the full YUV values for the entire screen (or whatever was stored given the subsampling) -> MadVR performs chroma subsampling -> MadVR performs scaling up or down -> MadVR converts from YUV to RGB -> present the raw pixel data to the media player.


Is this accurate? Thanks so much for any information you can give me.

[vivan]

So, this means that 4:4:4 YCrCb would require no subsamplingYes, 4:4:4 means that there's no subsampling.

and has a lossless conversion back to RGB, right?No, at least not in 8 bit. If you use YCgCo with at leasts 9 bit color depth (and PC range) only then RGB -> YUV -> RGB conversion would be lossless.

Also, does the interpolation occur in the YUV colorspace before conversion back to RGB, or does it occur after conversion to RGB? I'm guessing in YUV since that's what humans are more sensitive to.I don't even know how it could be possible to upsample chroma in RGB, since there's no chroma in RGB.

h264/AVC Media file -> extract out h264 video -> decode the h264 -> present MadVR with the full YUV values for the entire screen (or whatever was stored given the subsampling) -> MadVR performs chroma subsampling -> MadVR performs scaling up or down -> MadVR converts from YUV to RGB -> present the raw pixel data to the media player.

Is this accurate? Thanks so much for any information you can give me.Yes, that's how any video processing application work, first part at least. Extracting is what spitter for, decoding - decoder. There're several ways to do YUV (with video resolution) -> RGB (with screen resolution) conversion, and you described how madVR does it.