madVR - high quality video renderer (GPU assisted)

[ajp2k11]

On boards like to use 'up to' which means shared system ram. 6650M doesn't have much info out there, I have seen up to 2GB listed on a few sites. My old desktop onboard HD4200 had 128MB of dedicated vram, but could use up to 2GB. 1080p24 uses about 220MB of vram with windows 7 with 4 gpus queue, with 8 gpu queue (default) it's using 260MB. Anyhow you can check it out with process explorer to see if it's ram or gpu load. Load issue has always been dropped frames for me, ram being no video.

Really appreciate your feedback, thanks! I think it used to work quite some time ago but I can't remember when or what changed since I usually only play 'lighter' 720p files. I guess the major factors are Win8 and new AMD drivers... and/or new MadVR/LAV.

The 6650M doesn't have shared memory according to this...
http://www.notebookcheck.net/AMD-Radeon-HD-6650M.43962.0.html

Could this have something to do with it?
http://forum.doom9.org/showthread.php?p=1625470#post1625470

I'll try your suggestions tonight, thanks.

[ajp2k11]

I if calibrate my screen using my Spyder4 and it's software, will MadVR benefit? Even the gamut on a wide gamut display? Or would I have to use ArgyllCMS and/or create a 3dlut?

I know it's been discussed before but I can't find a definite answer so I would be grateful if someone could shed some light on this..

Thanks!

[Soukyuu]

Thank you for the answers Asmodian, Niyawa. This thread is so fast moving that finding anything in it is getting rather difficult unless you follow it daily.

[Asmodian]

I if calibrate my screen using my Spyder4 and it's software, will MadVR benefit? Even the gamut on a wide gamut display? Or would I have to use ArgyllCMS and/or create a 3dlut?

MadVR will benefit if you just calibrate but only as much as everything else and if you do not use the Windows 7 or newer windowed overlay and don't bypass the GPU gamma ramps. This will not correct the gamut.

To correct the gamut you can make some measurements using HCFR after your calibration and enter the values into MadVR's yCMS calibration fields. You can also make these measurements without your calibration and use the "disable GPU gamma ramps" option to bypass the GPUs CLUT and let MadVR do all of the calibration when playing video.

You can also create your own 3DLUT and use that to do the calibration in MadVR and again you have the option to either use the GPU gamma ramps or not. I currently use a method similar to the one described in "How To Get Perfect Video Color With JRiverMC (http://yabb.jriver.com/interact/index.php?topic=78302.0)" (it doesn't need JRiverMC). I leave the "Auto-calibrate" option checked in TI3 Parser and bypass the GPU gamma ramps (ICM profile) as I want to do the calibration in one step using one LUT.

You can skip past the first half to the final steps:

"Download and install the following software:
dispcalGUI: DispcalGUI
Argyll Color Management System: Argyll Color Management
TI3 Parser toolset: TI3 Parser

Set dispcalGUI to video, profile type XYZ LUT + matrix. Click the advanced button to set gamut mapping options: check both boxes; select an Rec. 709 profile; set the intents to absolute colorimetric. Click on the Calibrate & Profile button. When finished install the ICM profile. Check that it is loaded as the default profile for your display in the color management control panel.

Now go to the bin folder of the Ti3 Parser toolset and run LinkICCGUI.exe (important: Uncheck Auto-calibrate, this will then make a 3D LUT that will work with the installed ICM profile). The application will browse to the default locations to load the files created by dispcalGUI. Click on generate to make the 3D LUT for madVR. Go back to madVR settings and under calibration pick the 3D LUT option. Make the link to the .3dlut file you just created. Click apply and OK. Your done!"

The first half does give some useful information but you don't need to follow the steps to get a good calibration in MadVR. I think the Spyder 4's software does walk you through tuning the manual settings on your display.

Thank you for the answers Asmodian, Niyawa. This thread is so fast moving that finding anything in it is getting rather difficult unless you follow it daily.
Very true. :)

[leeperry]

How often? Once every couple of seconds? Or all the time (micro-judder)?
Oh well, apparently Sammy TV's plain suck at 24p: http://www.avforums.com/reviews/Samsung-UE-40-46-55-F7000-46-Inch-3D-LED-LCD-Smart-TV-Review_506/Review.html
Our only other minor complaint was that on occasion we could see [..] judder, with 24p content
I guess I didn't get the memo :scared: and eventually the TV has gone back to where it came from(as they all do :sly:).

BTW, is the judder test of Reclock meant to be really really dead smooth with mVR? Because Reclock seems to be hijacking D3D and draw it on top anyway so I'm not sure whether mVR really likes that :o

All I know is that if I enable a mirroring PS script then Reclock's judder test hiccups like hell.

Also, is this test video (http://forum.doom9.org/attachment.php?attachmentid=13406&d=1365967904) any good for testing 24p judder in your opinion?

[JarrettH]

While we're on it, what is "disable GPU gamma ramps for"?

I calibrate myself like ajp2k11

[6233638]

Also, is this test video (http://forum.doom9.org/attachment.php?attachmentid=13406&d=1365967904) any good for testing 24p judder in your opinion?That is a terrible video for testing judder. The white bar is far too thin, and not only is there no overlap between frames, but there are actually gaps between them.

That is going to look like it's juddering no matter what you play it back on.

Unfortunately, I don't have an example of a good 24p judder test video. I've yet to come across one. (I did like the built-in test that my old DVDO scaler had though)

EDIT: This is an example of what a good judder test pattern should look like, though they are usually white-on-black, rather than black-on-white: http://www1.datafilehost.com/d/3246045d

The only real issue with this is that there's no audio track, so neither ReClock or JRiver's VideoClock are engaged.
It gives you an idea of roughly what size the bar should be, and the speed it should move at though.

[ajp2k11]

@Asmodian,

thanks for the explanation on calibration! I will install Argyll/DispcalGUI and have a go. I need the regular ICC in Windows though since I edit lots of photos so I'll have to keep the gamma ramps.

[MasterMeNL]

For those that like this smooth motion feature but want more and have enough power you could look into Smooth Video Project (http://www.svp-team.com/), I've been using it for a long time and I definitely like it better than MadVR's feature. You do need a lot of CPU power though, especially for 1080p content.

I can't wait for xy-vsfilter to render at the monitor's native resolution using MadVR, but I heard that's being worked on.
I'm still using MPC-HC's internet subtitle renderer with a 60 frame buffer but it shits the bed when some complex stuff comes in.

[DragonQ]

For those that like this smooth motion feature but want more and have enough power you could look into Smooth Video Project (http://www.svp-team.com/), I've been using it for a long time and I definitely like it better than MadVR's feature. You do need a lot of CPU power though, especially for 1080p content.
They really aren't the same thing. MadVR doesn't interpolate any new information, so the "film-like" effect of 24/25p content isn't changed when played back at 60 Hz, unlike when using SVP.

[karamancho]

what are presentation glitches and how can they be avoided?

[n3w813]

WARNING: Using a 3dlut generated from ArcgyllCMS/TI3parser will not produce a 100% corrected color gamut. Although the saturation/hue of targets inside the color gamut will be moved closer to the target, the luminance(Y) levels will result in huge deltaEs.

To test, you can take measurements using 20%/25% saturation sweeps of the gamut after the 3DLUT has been implemented in MadVR. The xy may be better but be sure to check DeltaL of the points read.


MadVR will benefit if you just calibrate but only as much as everything else and if you do not use the Windows 7 or newer windowed overlay and don't bypass the GPU gamma ramps. This will not correct the gamut.

To correct the gamut you can make some measurements using HCFR after your calibration and enter the values into MadVR's yCMS calibration fields. You can also make these measurements without your calibration and use the "disable GPU gamma ramps" option to bypass the GPUs CLUT and let MadVR do all of the calibration when playing video.

You can also create your own 3DLUT and use that to do the calibration in MadVR and again you have the option to either use the GPU gamma ramps or not. I currently use a method similar to the one described in "How To Get Perfect Video Color With JRiverMC (http://yabb.jriver.com/interact/index.php?topic=78302.0)" (it doesn't need JRiverMC). I leave the "Auto-calibrate" option checked in TI3 Parser and bypass the GPU gamma ramps (ICM profile) as I want to do the calibration in one step using one LUT.

You can skip past the first half to the final steps:

"Download and install the following software:
dispcalGUI: DispcalGUI
Argyll Color Management System: Argyll Color Management
TI3 Parser toolset: TI3 Parser

Set dispcalGUI to video, profile type XYZ LUT + matrix. Click the advanced button to set gamut mapping options: check both boxes; select an Rec. 709 profile; set the intents to absolute colorimetric. Click on the Calibrate & Profile button. When finished install the ICM profile. Check that it is loaded as the default profile for your display in the color management control panel.

Now go to the bin folder of the Ti3 Parser toolset and run LinkICCGUI.exe (important: Uncheck Auto-calibrate, this will then make a 3D LUT that will work with the installed ICM profile). The application will browse to the default locations to load the files created by dispcalGUI. Click on generate to make the 3D LUT for madVR. Go back to madVR settings and under calibration pick the 3D LUT option. Make the link to the .3dlut file you just created. Click apply and OK. Your done!"

The first half does give some useful information but you don't need to follow the steps to get a good calibration in MadVR. I think the Spyder 4's software does walk you through tuning the manual settings on your display.


Very true. :)

[Asmodian]

While we're on it, what is "disable GPU gamma ramps for"?
I need the regular ICC in Windows though since I edit lots of photos so I'll have to keep the gamma ramps.

This is the point of the "disable gamma ramps" option. It only disables them for MadVR, not the rest of Windows. Basically it allows you to use one 3DLUT in MadVR but also have a normal calibration for everything else. If you don't bypass the gamma ramps the 3DLUT in MadVR needs to convert to the wrong colors which your ICC profile then fixes.

WARNING: Using a 3dlut generated from ArcgyllCMS/TI3parser will not produce a 100% corrected color gamut. Although the saturation/hue of targets inside the color gamut will be moved closer to the target, the luminance(Y) levels will result in huge deltaEs.

Ok, I have noticed this but as far as I know there is no better solution for color calibration for video playback. Using this method looks much better on my wide gamut display than anything else I have tried. After this videos on both my wide gamut and sRGB monitors look very similar. I believe humans are very bad at noticing differences in luma but relatively good at hue?

Any suggestions for a better option?

[dansrfe]

I wish I could just use a single 3DLUT for madVR and for Windows. It would be simpler to get rid of ICC profiles in their entirety.

[cyberbeing]

If you don't bypass the gamma ramps the 3DLUT in MadVR needs to convert to the wrong colors which your ICC profile then fixes.

That is a rather convoluted statement, which I can't make sense of.

If you create your 3DLUT from measurements when your GPU gamma ramp is active (i.e. calibrated ICC profile), you need to leave the madVR GPU gamma ramp enabled and use Windowed mode or Fullscreen Exclusive only. In this case, your GPU gamma ramp is already doing gamma and grayscale correction, making any further finetuning of gamma and grayscale correction in your 3DLUT entirely optional.

If you explicitly disable your GPU gamma ramp before taking measurments which are used for creating your 3DLUT, you need to disable the GPU gamma ramp in madVR and use Overlay mode or Fullscreen Exclusive only. In this case, you'd want to handle gamma and grayscale correction in your 3DLUT.

Either way, the 3DLUT is always being used for gamut correction, never the ICC profile.

[n3w813]

Ok, I have noticed this but as far as I know there is no better solution for color calibration for video playback. Using this method looks much better on my wide gamut display than anything else I have tried. After this videos on both my wide gamut and sRGB monitors look very similar. I believe humans are very bad at noticing differences in luma but relatively good at hue?

Any suggestions for a better option?

I think you got that backwards...

"... human vision has finer spatial sensitivity to luminance ("black and white") differences than chromatic differences..." (link (http://en.wikipedia.org/wiki/Luma_(video)))

Right now, there is no better option unfortunately. :( I think Madshi mentioned implementing a better 3dlut correction system in MadVR but not sure where that is in his plans.

[Asmodian]

That is a rather convoluted statement, which I can't make sense of.

rephrased:
If you bypass the gamma ramps MadVR does all of the correction with the 3DLUT, ignoring the ICC profile.

I should have said MadVR's output would need further correction by the ICC profile instead of "incorrect colors". The <100% colors are affected by the ICC profile to some degree but it is true the gamut is not.

Either way, the 3DLUT is always being used for gamut correction, never the ICC profile.

Right, but I like MadVR to do as much as it can so nothing is left for the video card but to send the data to the monitor. I trust MadVR to do the math right, at higher bit depth, with dither, fewer rounding steps, etc. but I also believe I don't notice the difference. :)

I think you got that backwards...

"... human vision has finer spatial sensitivity to luminance ("black and white") differences than chromatic differences..." (link (http://en.wikipedia.org/wiki/Luma_(video)))

I am not sure how spatial sensitivity relates to luma vs hue sensitivity to be honest. You can tell the difference between bands of very slightly different luma when they are next to each other but not if you see one after the other with almost any time between. What does this mean when talking about gamuts? I am not sure. :p

[Aleksoid1978]

madshi
Hi. I found a bug that's happened only with madVR, with other renderer all ok. I use IMFGetService::GetService() to get a pointer to IDirect3DDeviceManager9. It's need to Hook call CreateVideoDecoderMine() for detect DXVA2 mode.
Well - some commercial decoder like ArcSoft & Cyberlink init DXVA2 decoder while connect to the renderer, so I need set a Hook before decoder connect to the renderer. But - madVR give a error.
here this a code:

if (CComQIPtr<IMFGetService, &__uuidof(IMFGetService)> pMFGS = pBF) {
CComPtr<IDirectXVideoDecoderService> pDecoderService;
CComPtr<IMFGetService> pGetService;
CComPtr<IDirect3DDeviceManager9> pDeviceManager;
HANDLE hDevice = INVALID_HANDLE_VALUE;

// Query the pin for IMFGetService.
hr = pMFGS->QueryInterface(__uuidof(IMFGetService), (void**)&pGetService);

// Get the Direct3D device manager.
if (SUCCEEDED(hr)) {
hr = pGetService->GetService(
MR_VIDEO_ACCELERATION_SERVICE,
__uuidof(IDirect3DDeviceManager9),
reinterpret_cast<void**>(&pDeviceManager));
}
...


error on call pGetService->GetService(). Here is a bugreport.txt - http://aleksoid.voserver.net/MPC-BE/madVR_bugreport.txt

[truexfan81]

madshi
Hi. I found a bug that's happened only with madVR, with other renderer all ok. I use IMFGetService::GetService() to get a pointer to IDirect3DDeviceManager9. It's need to Hook call CreateVideoDecoderMine() for detect DXVA2 mode.
Well - some commercial decoder like ArcSoft & Cyberlink init DXVA2 decoder while connect to the renderer, so I need set a Hook before decoder connect to the renderer. But - madVR give a error.
here this a code:

if (CComQIPtr<IMFGetService, &__uuidof(IMFGetService)> pMFGS = pBF) {
CComPtr<IDirectXVideoDecoderService> pDecoderService;
CComPtr<IMFGetService> pGetService;
CComPtr<IDirect3DDeviceManager9> pDeviceManager;
HANDLE hDevice = INVALID_HANDLE_VALUE;

// Query the pin for IMFGetService.
hr = pMFGS->QueryInterface(__uuidof(IMFGetService), (void**)&pGetService);

// Get the Direct3D device manager.
if (SUCCEEDED(hr)) {
hr = pGetService->GetService(
MR_VIDEO_ACCELERATION_SERVICE,
__uuidof(IDirect3DDeviceManager9),
reinterpret_cast<void**>(&pDeviceManager));
}
...


error on call pGetService->GetService(). Here is a bugreport.txt - http://aleksoid.voserver.net/MPC-BE/madVR_bugreport.txt

please post bug reports on the bug tracker (http://bugs.madshi.net/my_view_page.php) thanks

[leeperry]

That is a terrible video for testing judder. The white bar is far too thin, and not only is there no overlap between frames, but there are actually gaps between them.

That is going to look like it's juddering no matter what you play it back on.
Thought so! Well, setting up a good audio rig took me years but doing the same with video just seems to be an endless source of headaches, I'm this close to giving up.

If Reclock's judder test is really a nasty hack(as it would appear) and I can't find any useful test video then I'm basically SOL.

I found an old version of juddertest.exe that might do what I want but it also hiccups a bit, OTOH it's an old beta.......and that video you just linked doesn't play at all on my PC for some reason, I will try to change its frame rate to 24p and/or remux it to MKV together with an audio track.

All I know is that 24p movies look dead smooth in 96Hz on my CRT(and so did they in 48Hz on a 46" Hitachi LCD set), I think I'll leave it at that. The more it goes the more I prefer a small high contrast screen to the opposite....Yesgrey once told me that vision takes most of the human brain activity processing power and I happen to care a lot more about audio than video anyway. That 25" broadcast OLED would be right up my alley, but its price tag is :scared:

SONY claim that UHD Oled's will be available by the end of the year, all I want is a 1080p 25".....and SONY do know how to handle 24p for a change :devil:

I have noticed this but as far as I know there is no better solution for color calibration for video playback.
Did you try/measure this script (http://www.avsforum.com/t/912720/color-correction-with-a-htpc-simpler-solution-and-now-it-really-works)? I added a few digits after the coma (http://forum.doom9.org/attachment.php?attachmentid=13320&d=1362359744) and it has always looked amazing to me, one of those things that never fails making me happy http://forum.slysoft.com/images/smilies/agreed.gif

[mindbomb]

What are the benefits of full screen exclusive mode?

[6233638]

What are the benefits of full screen exclusive mode?Better performance and no chance of notifications from other programs overlaying the video.

[aufkrawall]

And better control over system clocks to avoid dropped frames, I think.

[namaiki]

It should help with multiple monitors and Aero in Windows... 7, I think. If you find the screen flash annoying, I think it's less distracting if you untick the "wait 3 seconds" option.

[Niyawa]

Funny because exclusive mode here always makes things worse. My rendering doubles when it's enabled. I used to get a lot of dropped frames with it. One funny thing I noticed about queues is that the higher the CPU queue, the lower the rendering ms and as soon as my GPU queue goes above 6 my rendering time goes "sky's the limit". In the end I'm using 32/6 respectively.

[namaiki]

What GPU is that? Did you try with madVR default settings?

[Niyawa]

What GPU is that? Did you try with madVR default settings?
I'm using an onboard Intel 4500MHD and yes, I used with various settings since default but none of them worked.

[karamancho]

does anyone know what a presentation glitch is and how can it be avoided?

[namaiki]

I'm using an onboard Intel 4500MHD and yes, I used with various settings since default but none of them worked.

Since that is integrated, maybe try reduce GPU queues to the lowest and disable Aero while in fullscreen.

Try use the lowest scalers first (bilinear) and then try raise image upscaling slowly to see what your GPU is capable of.

Make sure not to use any form of DXVA in LAV Video since we will assume that GPU is very low in resources.


does anyone know what a presentation glitch is and how can it be avoided?

Need more information about everything in your setup.

[Niyawa]

Since that is integrated, maybe try reduce GPU queues to the lowest and disable Aero while in fullscreen.

Try use the lowest scalers first (bilinear) and then try raise image upscaling slowly to see what your GPU is capable of.

Make sure not to use any form of DXVA in LAV Video since we will assume that GPU is very low in resources.
Like I said, I've already tried all of that. It simply doesn't work here. I'm using Windows 8 so Aero is long gone, using the lowest queue would reduce the ms but only by a insignificant margin, I'm already using bilinear for everything (aside downscaling which is Catmull-Rom) and I don't touch DXVA. I have no idea why I would even use it when my GPU is this weak hahaha.

[karamancho]

Need more information about everything in your setup.

default settings + bilinear on all 3 scaling options
I've pretty much reduced dropped and delayed frames that I was getting, the only thing left are presentation glitches so I was asking is there something I can change in the settings to get rid of them too.

[namaiki]

default settings + bilinear on all 3 scaling options
I've pretty much reduced dropped and delayed frames that I was getting, the only thing left are presentation glitches so I was asking is there something I can change in the settings to get rid of them too.

What GPU, version of Windows? Did you change any other madVR settings?

Like I said, I've already tried all of that. It simply doesn't work here. I'm using Windows 8 so Aero is long gone, using the lowest queue would reduce the ms but only by a insignificant margin, I'm already using bilinear for everything (aside downscaling which is Catmull-Rom) and I don't touch DXVA. I have no idea why I would even use it when my GPU is this weak hahaha.


Probably can't help you there. When I had my 4500MHD, I changed the two settings I mentioned above. I can't remember which scalers though I think I had lanczos3 for upscaling SD to 720p. This was before the time of smooth motion. I didn't have that enabled.

[karamancho]

What GPU, version of Windows? Did you change any other madVR settings?

laptop with integrated NVIDIA Quadro NVS 135M GPU, windows 7 ultimate x86, no options changed in madvr except for the scaling algorithms (fullscreen exclusive mode enabled by default)

@madshi: can a volume indicator be added to fullscreen exclusive mode interface?

[Niyawa]

laptop with integrated NVIDIA Quadro NVS 135M GPU, windows 7 ultimate x86, no options changed in madvr except for the scaling algorithms (fullscreen exclusive mode enabled by default)
Presentation glitches huh? Never had them but someone mentioned overlay helps with that.

@madshi: can a volume indicator be added to fullscreen exclusive mode interface?
madshi is not accepting any features requests until version 1.0, so keep that with you for now. :p

[e-t172]

Unfortunately, I don't have an example of a good 24p judder test video. I've yet to come across one.

You gave me the idea to write a shell script (https://docs.google.com/file/d/0B4SHVPm2DfK5OXZiN0x4Qllsenc/edit?usp=sharing) to generate this kind of videos (moving bar). Here is a series of 24p judder tests (https://drive.google.com/folderview?id=0B4SHVPm2DfK5RnZIMDdINTFBN2s&usp=sharing) generated using this script with different parameters. For example "juddertest-25s-50.mkv" means a 50-pixel wide bar with a cycle duration of 25 seconds. The script is written in such a way that the bar always progresses by the exact same number of pixels at each frame. I included a -60dBFS white noise soundtrack as dummy audio. The result is encoded as H.264+AC3 in MKV.

Anyone with access to a Unix box with the necessary dependencies can use the script to generate his own videos with different parameters, such as frame rate, colors, etc.

I would be interested in knowing the set of parameters that work best in most situations, especially to make 3:2 pulldown obvious.

[karamancho]

Presentation glitches huh? Never had them but someone mentioned overlay helps with that.

overlay mode is not supported by my GPU :(

[mindbomb]

I have no idea why I would even use it when my GPU is this weak hahaha.

you know, in cases of a weak gpu like this, perhaps it would be better to do the rgb conversion in lav video and the scaling with ffdshow video processor. This way, it is all done on the cpu.

[6233638]

You gave me the idea to write a shell script (https://docs.google.com/file/d/0B4SHVPm2DfK5OXZiN0x4Qllsenc/edit?usp=sharing) to generate this kind of videos (moving bar). Here is a series of 24p judder tests (https://drive.google.com/folderview?id=0B4SHVPm2DfK5RnZIMDdINTFBN2s&usp=sharing) generated using this script with different parameters. For example "juddertest-25s-50.mkv" means a 50-pixel wide bar with a cycle duration of 25 seconds. The script is written in such a way that the bar always progresses by the exact same number of pixels at each frame. I included a -60dBFS white noise soundtrack as dummy audio. The result is encoded as H.264+AC3 in MKV.

Anyone with access to a Unix box with the necessary dependencies can use the script to generate his own videos with different parameters, such as frame rate, colors, etc.

I would be interested in knowing the set of parameters that work best in most situations, especially to make 3:2 pulldown obvious.Awesome, thank you. One suggestion I'd make would be to calculate the distance based on pixels rather than dividing up a certain amount of time. The current videos are showing a grey line on the left edge intermittently, which I'm guessing might be related to that.

I'm also wondering if inverted patterns (white on black) or black/white on grey might work better - with the faster patterns I see edge flickering, which is distracting when you're trying to identify judder. If changing the colors doesn't help, there's nothing else that could be done about that really, other than having patterns which had some kind of motion blur encoded in them - that would remove the flicker but hopefully not mask any judder.

That's really only an issue at 24fps though - at higher framerates (such as 60fps) moving in single pixel increments is fast enough to identify judder without flickering.

[Niyawa]

you know, in cases of a weak gpu like this, perhaps it would be better to do the rgb conversion in lav video and the scaling with ffdshow video processor. This way, it is all done on the cpu.
No need for that. It's not like I need fullscreen exclusive mode, I simply can't use it as it was supposed to be. Also I want to avoid any sort of workaround like that, I like madVR the best since I want all the quality I can get.

[madshi]

JFMI: Why making new judder test patterns? What's wrong with the madTestPatternSource smooth motion test patterns?

[e-t172]

Awesome, thank you. One suggestion I'd make would be to calculate the distance based on pixels rather than dividing up a certain amount of time. The current videos are showing a grey line on the left edge intermittently, which I'm guessing might be related to that.

I'm not sure I understand what you're referring to.

I'm also wondering if inverted patterns (white on black) or black/white on grey might work better - with the faster patterns I see edge flickering, which is distracting when you're trying to identify judder. If changing the colors doesn't help, there's nothing else that could be done about that really, other than having patterns which had some kind of motion blur encoded in them - that would remove the flicker but hopefully not mask any judder.

Black on white: I could try that.
Black/White on grey: the issue with that is that it will probably look worse with monitors that have bad RTC (Response Time Compensation), i.e. overshoot.
Motion blur: I should be able to generate a pattern where the moving bar is a narrow horizontal gradient instead of a solid line, which would make it appear blurry.

Why making new judder test patterns? What's wrong with the madTestPatternSource smooth motion test patterns?

They don't have sound, so they cannot be used to reproduce real-world scenarios with respect to clock-related issues (e.g. ReClock won't be used). Besides, I'm not sure how effective your patterns are compared to the simple moving bar pattern. I personally find yours too complex and distracting.

[cyberbeing]

They don't have sound, so they cannot be used to reproduce real-world scenarios with respect to clock-related issues (e.g. ReClock won't be used).
If you wanted sound, all you had to do was create a 1 minute FLAC file, name it the same as the *.ytp files, and MPC-HC will automatically load it as external audio.

[QBhd]

I have yet to get the *.ytp files to play in PotPlayer... I have always wanted to run them, but it seems to be a no go. Is there something I am missing?

QB

[namaiki]

laptop with integrated NVIDIA Quadro NVS 135M GPU, windows 7 ultimate x86, no options changed in madvr except for the scaling algorithms (fullscreen exclusive mode enabled by default)

@madshi: can a volume indicator be added to fullscreen exclusive mode interface?

If you have older drivers installed, maybe try disable "present frames in advance". In the same case of having older drivers installed, you can also try enabling smooth motion which technically is not the same thing at all, but helped me when I did not want to update my drivers. You could try either one of the options for testing, but based on my own configuration, I don't think enabling both is a good idea. In the case of the 'present frames in advance' option, you will need to close and reopen the video for any changes to take effect.

You can get a volume indicator in madVR if you are using MPC-HC by enabling the OSD. (View-> Options-> Player-> tick Show OSD)
edit: If you mean a way to change the volume by using the mouse only while in exclusive mode, you can also try use your scroll wheel or two finger scroll if you have a touchpad.

[drmpeg]

If anyone is interested, I have some judder test patterns.

http://www.w6rz.net/judder.zip

http://www.w6rz.net/juddertest.zip

Ron

[6233638]

JFMI: Why making new judder test patterns? What's wrong with the madTestPatternSource smooth motion test patterns?MP4 files work in anything, I can only get the YTP patterns working in MPC-HC - they don't load in JRiver at all.

And as mentioned, there's no audio track so ReClock/VideoClock wouldn't be loaded either.

Other than that, they are fine - better than fine actually, as they compare multiple speeds at once, include motion blur which helps reduce flicker, and aren't just white-on-black.

[madshi]

There's one thing I don't like about these "moving bar" 24fps test patterns: They don't include any motion blur. Camera pans in 24fps movies are usually shot with some motion blur which allows much faster motion without visible 24Hz judder. These 24fps "moving bar" test patterns either move too slowly, or if they move fast enough they judder like crazy even if you display them natively at 24Hz because they don't have any motion blur. In order to truely judge whether motion is smooth or not we need a test pattern which moves relatively fast, but still looks smooth when displayed with framerate = refreshrate. And that's exactly what the madVR smooth motion test patterns do.

Of course the ideal test pattern would be a mix of some smooth but fast motion movie scenes...

[madshi]

MP4 files work in anything, I can only get the YTP patterns working in MPC-HC - they don't load in JRiver at all.

And as mentioned, there's no audio track so ReClock/VideoClock wouldn't be loaded either.

Other than that, they are fine - better than fine actually, as they compare multiple speeds at once, include motion blur which helps reduce flicker, and aren't just white-on-black.
Well, isn't there a DirectShow MPEG2 or h264 encoder available somewhere? You could build a "test pattern -> encoder" chain in GraphEdit and this way convert the madVR test patterns into an MPEG2/h264 video.

[6233638]

Well, isn't there a DirectShow MPEG2 or h264 encoder available somewhere? You could build a "test pattern -> encoder" chain in GraphEdit and this way convert the madVR test patterns into an MPEG2/h264 video.I'll have to look into that, thanks for the suggestion.

[cvrkuth]

I noticed that when playing 1080p24 (mkv, m2ts, ts .. full screen exclusive mode, default settings) my screen goes to 23hz. Because of this, there is a drop frames every 30-40 sec.

https://lh4.googleusercontent.com/-WCZJVHwvo0E/UXzRP6TYSLI/AAAAAAAABXE/lpPBshci9ac/s400/fse_newpath.jpg

If I uncheck 'Present Several frames in advance "(old path), my screen stays in 24hz.

https://lh5.googleusercontent.com/-oCUIT5D2MC4/UXzRUg5udOI/AAAAAAAABXM/jnfJUQWz_o8/s400/fse_oldpath.jpg

Is it related to Windows 8, Desktop Window Manager (DWM)?