madVR - high quality video renderer (GPU assisted)

[ajp_anton]

Madshi:

Do you remember when we talked about how to do an exact 2x upscale, for example 720p movie on a 1440p/1600p monitor?
Afaik, madVR preserves the center position so that every new pixel is interpolated, and I thought it would be better to center around a corner so that every source pixel is preserved. I tested the differences in Avisynth, and while there were differences, neither really looked better than the other.

I just thought about it again, and wouldn't it be much faster to preserve the source pixels? In case you're upscaling one dimension at a time, it would halve the number of interpolated pixels both times. Or if I'm missing something, at least 1/4 could be saved as that's the amount of untouched source pixels.
This might also help on other upscale factors because it pretty much guarantees that the maximum number of source pixels are still present.

[Keiyakusha]

Originally Posted by Andy o

It's in the first page of this thread, written by madshi. The basic idea is that if the player can output more than 8-bit, then it doesn't need to dither down to 8-bit after processing, so less downconversion is applied in the chain. Go read it for details though.

Here we are talking about 8bit source that was made to 10bit by x264. It is clear enough from dansrfe's post. What I say is that leaving it 10bit with no dithering doesn't give any benefits.
We don't care about "less downconversion", our 10bit source in this case is a fake. It will be downconverted anyway sooner or later. And if not, it won't look any better anyway.
Also higher than 8bit processing is good indeed but it doesn't means the source won't be dithered in the end.
Originally was a question, how it looks like, when not dithered. The answer is - it won't be any better. Lets not go any further.

[Andy o]

Still, once you've upconverted to higher bit depth, be it in encoding or in processing, it's ideal to leave it that way until final downconversion, which in this case happens on the monitor. Of course this can't be done yet, but still.

[strumf666]

Originally Posted by Keiyakusha

Maybe I'm wrong but I always though this is how it is and how it should work. Why would you care about drops if you pause anyway? And no surprise in aero actually helping. dunno why, but this is the case for many users (not only madvr). It have its own vsync

Thanks for answering. Iirc I read that aero's vsync is messing with madvr vsync hence aero should be disabled. It is possible the information was incorrect or I misunderstood it and that's why I am asking.

[Keiyakusha]

Originally Posted by strumf666

Thanks for answering. Iirc I read that aero's vsync is messing with madvr vsync hence aero should be disabled. It is possible the information was incorrect or I misunderstood it and that's why I am asking.

Don't get me wrong, I'm not entirely sure myself. I recall something like that at early stages of madvr development but desktop composition shouldn't be an issue now. Also for example on Win8 you can't turn it off anymore so if you aware of some troubles with it, they probably need to be resolved...

[Keiyakusha]

Regarding AR calculation again.
As was reported before by few users 720x480 video with 32:27 SAR results incorrect resolution 1919x1080
Now 692x448 with 103:86 SAR results 1920x1058 resolution, while it should be 1920x1038 (20px error seriously? Oo)
Is there a way to improve something about it? I don't rectally care about 2nd case, this is one and only dvd i have with letterboxed 1.85 video, but how about 720x480 case?

[HauntingShock]

Newbie here...

May I know what's the difference if I enable or disable exclusive mode in fullscreen?

[cyberbeing]

Originally Posted by Keiyakusha

Now 692x448 with 103:86 SAR results 1920x1058 resolution, while it should be 1920x1038 (20px error seriously?

Have you confirmed with screenshots that there actually is a 20px error? Normally it's just a case of madVR padding the video.
The first two numbers of madVR's Target Rectangle stats are horizontal and vertical padding.

I'd also be curious what resolution madVR uses when playing that video at 100% after anamorphic stretching.

According to the resolution and SAR you listed it, should be 829x448 (or 828x448) after anamorphic stretching, but according to what you say madVR is using @1920x1080, it would be 813x448 which doesn't make any sense. Are positive that your decoder is reporting the correct SAR to madVR?

[nx6]

Originally Posted by HauntingShock

Newbie here...

May I know what's the difference if I enable or disable exclusive mode in fullscreen?

In Fullscreen Exclusive mode you're pretty much dedicating the display output to interfacing with MadVR to provide the least chance of dropped frames or other playback issues. Other apps (or even the video player itself) trying to display other things that aren't part of your video playback like OSD displays, email arrival notifications, etc may have trouble in FSE mode or their appearance may cause MadVR to temporarily drop out of FSE mode.

If you don't have playback issues that you need FSE mode to overcome, there's no reason why you couldn't disable FSE mode if it allows these other programs to draw to the screen more easily or gets rid of the sometimes noticable "jump" to FSE mode when going to full screen on your video player.

[andybkma]

Greets, posting this to make sure that my new laptop's video card isn't broke... picked up a Lenovo Y480 Win7 64bit 8Gb Ram with Nvidia 640M-LE graphics card 2GB dedicated ram the other day.

I am using madVR with defaults with Zoom Player. When I set Zoom Player to be used with the onboard Intel 4000 graphics everything is fine with mVR in Fullscreen Exclusive Mode. But when I use Zoom Player with the Nvidia graphics card, and with "Present Several Frames In Advance" (I do believe that's default) checked, in FSE or without FSE, videos stop playing (they just pause by themselves) around 20 seconds after start but the sound continues. And then when I come out fullscreen (doesn't matter whether out of FSE or not) into window mode the video will look like its in Fast Forward by itself trying to catch up. If I uncheck "Present Several Frames In Advance " then it works fine in FSE or without FSE. Ummm... have any of you guys experienced this? Am worried that my new laptop's nvidia video card is hosed since this is new technology (compared to me old 8600M-GT) and ought to work right?

Edit: Doesn't matter what container or decoder, any vid I play using the nvidia graphics card with mVR "Present Several Frames In Advance" checked freezes playback after about 20 seconds. Reclock on or off doesn't matter either

[vivan]

andybkma,
that happens with all Optimus systems. Solution is to untick "use a separate device for presentation".

[makakam]

Hi everyone
I have this really important question that's been buggin' me for some time now. My HTPC (Radeon 6570) is connected to onkyo avr with hdmi and then to tv(panasonic vt30) with hdmi as well. I have been using full rgb pixel format and mpc-hc with madvr set to 16-235 (otherwise i get crushed blacks). I have calibrated my tv using htpc pattern generator and greyscale ramp in it looks like the one from avs hd calibration disc when I play it in mpc-hc which is not perfectly smooth. I have noticed that when I change pixel format to ycbcr 4 2 2 the ramp in htpc pattern generator becomes smooth but thin vertical lines are still there in mpc-hc. Am I doing sth wrong or is it just impossible to have it completely smooth in mpc-hc? Thanks in advance.

[fairchild]

Originally Posted by makakam

Hi everyone
I have this really important question that's been buggin' me for some time now. My HTPC (Radeon 6570) is connected to onkyo avr with hdmi and then to tv(panasonic vt30) with hdmi as well. I have been using full rgb pixel format and mpc-hc with madvr set to 16-235 (otherwise i get crushed blacks). I have calibrated my tv using htpc pattern generator and greyscale ramp in it looks like the one from avs hd calibration disc when I play it in mpc-hc which is not perfectly smooth. I have noticed that when I change pixel format to ycbcr 4 2 2 the ramp in htpc pattern generator becomes smooth but thin vertical lines are still there in mpc-hc. Am I doing sth wrong or is it just impossible to have it completely smooth in mpc-hc? Thanks in advance.

I have mine smooth with slightly different settings. I use RGB Limited as the pixel format, then in MadVR I have it set to the default 0-255. I have my Radeon 7870 going from it's HDMI to the HDMI in on my Panasonic S2 42" plasma. Maybe it's something to do with your receiver.

[Ver Greeneyes]

Since it seems the deringing craze has died down a bit (very interesting stuff!), reposting this as I think it got lost at the end of a page. Can someone advise me on how this works?

Originally Posted by Ver Greeneyes

I have a question with regards to the color correction built into madVR. My monitor is calibrated using an X-Rite ColorMunki Photo, using a LUT together with a v4 ICC profile. If I setup my monitor in madVR so that it's set to "this display is already calibrated", will it use the ICC profile assigned to my monitor? Or do I somehow need to convert my ICC profile to a 3DLUT file? The LUT itself should already be loaded into my GPU.

Oh, and should I "enable gamma processing" on the color & gamma tab? Would it do anything if it's set to the same gamma as my ICC profile (2.20)?

If there's a readme with detailed descriptions for each setting out there, I'd love to read it.

[pie1394]

Originally Posted by makakam

Hi everyone
I have noticed that when I change pixel format to ycbcr 4 2 2 the ramp in htpc pattern generator becomes smooth but thin vertical lines are still there in mpc-hc. Am I doing sth wrong or is it just impossible to have it completely smooth in mpc-hc? Thanks in advance.

With 4:2:2, the chroma U/V are sampled with half horizontal resolution if compared to 4:4:4 video signal format. If you see the vertical line or color banding, it has the possibility that non-interpolated U/V pixel down-sampling is performed somewhere. I saw some similar problems when designing an ARGB 4:4:4 source to 4:2:2 UYVY+Alpha OSD frame buffer converting many years ago.

[blackjack12]

Originally Posted by madshi

Generally, if you report problems with specific samples with LAV + madVR, please always check if madVR reports deinterlacing to be on/off, and whether overriding that decision by pressing Ctrl+Alt+Shift+D fixes things. You would noticably increase the usefulness of your reports if you mentioned right away what the real problem is. There are 3 basic possibilities:

(1) Image quality is bad with deinterlacing off, but good with deinterlacing on. The problem is that it's not turned on automatically.
(2) Image quality is bad with deinterlacing on, but good with deinterlacing off. The problem is that it's turned on automatically.
(3) Image quality is bad in any case, no matter if deinterlacing is activated or not.

For every problem you see, it would be nice if you could specify which of the 3 problems we're talking about. Also, it would be helpful to know if EVR plays the same file correctly with the same decoder settings.

Madshi,

MadVR deinterlacing: MPEG2 DVD “interlaced” issues with direct DVD material rips using MakeMKV ver 1.7.6 or 1.7.7 - Problem now understood
Seen and tested on both AMD Radeon and NVIDIA GPU systems

MadVR is “BAD”
Reports deinterlacing on : force off and is “GOOD” … (2)

EVR/EVR-CP is “GOOD”
Can actually see FPS jumps a little at beginning that looks like interlaced then goes to 24fps. An interlaced beginning blip that triggers MadVR and MediaInfo reads but EVR ignores or plays through.

MediaInfo shows material as interlaced but the reason that I did not include samples is because when I used MKVMerge to create clip of credits where easy to see, the clip now indicates progressive.

Even tried re-ripping with absolute newest MakeMKV to see if that was issue but all show same result.

So the interlaced flag that looks like is a quick blip in the beginning is being seen and responded to by MadVR but EVR “ignores”? and plays correctly.

Did not upload samples but here are some DVD’s that I had that easily showed the issues: (ripped with MakeMKV ver 1.7.7)

• 13 Going on 30
• xXx
• The Back to the Future Series
• … Many, many more

I then used the latest MKVMerge and clipped the first 30 seconds off the ripped material for a quick test to see if could get rid of the “interlaced” blip and it did! Now the files tested play as progressive without issue.

The “interlaced” blip at the beginning causes the entire video to be seen as “interlaced” by MadVR when in fact it is not. ... A lot of issues in the beginning logo stuff.

Note:
See the exact same behavior with Radeon 6570 and NVidia/CUVID with 560Ti and GT525M systems.

UPDATE NOTE:
Clipping off 1.0-39 sec at the beginning of the MKV files got rid of the "interlace blip" on most files and they then indicated deinterlacing off with MadVR and progressive in MediaInfo in MPC-HC - and played without issue.

The "old" intro logos and stuff are crappy, have interlaced content and cause most of the problems ...

Chopping out is not practical but it worked.

Really need MadVR to accommodate this.

EVR/EVR-CP allows these same files to play correctly and there are a lot of them with a DVD direct rip to MKV using MakeMKV. If you process with something like HandBrake all is converted to progressive and you never see the issue ... but you do lose information.

The results when working are extremely good.

[pando2]

I have a question about yCMS in madVR:
How can we move to a target SMPTE-C while the concervant yCMS, the two can not be checked ??

[ThurstonX]

Originally Posted by vivan

andybkma,
that happens with all Optimus systems. Solution is to untick "use a separate device for presentation".

Ah! Thank you very much :-) I have a Toshiba Qosmio with a GTX 560M and a Core i5 (HD 3000), and out of the box madVR didn't work, with playback being massively stuttered (EVR cp was fine). I checked the nVidia settings, and MPC-HC was set to use the nVidia. When I forced it to use the Intel GFX and switched LAV to Quick Sync, playback worked, but render times seemed pretty high.

Now I've switched back to forcing MPC-HC to use the nVidia, and set LAV to use CUVID. Render times improved greatly.

And I love the Anti-ringing switches in test build 6! I have a lot of low-res vids, mostly stuff dumped from a Comcast DVR to an old Hauppauge via S-Video (I know! I know! Don't laugh ;-) Just needed to preserve a lot of stuff ) madVR's ability to tweak the upscaling is bliss.

I'm a total noob to LAV and madVR, having been content with CCCP's implementation with ffdshow, but when reading they're going to ditch it eventually for LAV, I decided to be proactive.

My HTPC runs a Radeon HD 5670 and a Core i5, and all is perfect with LAV + madVR, whether I use no HW acceleration or DXVA2 copy-back. Actually, GPU-Z shows lower GPU % used with no HW acceleration, at least with a brief logging test. I'm not exactly clear if DXVA2 copy-back gives some benefit over no HW acceleration. But then I'm still reading, absorbing and learning.

Anyway, thanks a ton, vivan, for the quick fix, and of course madshi and nev for madVR and LAV. Y'all rule :-)

[pie1394]

Originally Posted by blackjack12

The "old" intro logos and stuff are crappy, have interlaced content and cause most of the problems ...

This kind of trouble-some mixed fps contents did exist.

I remembered that interlaced title / introduction at the beginning and then fully progressive 24 fps FILM was seen on the old Region-1 DVDs.

It also happens to some Japanese broadcast anime programs encoded with MPEG-2MP@HL.
i.e. 29.97 fps Opening & Ending encoded with interlaced mode while the other parts are still 23.976 fps progressive.

If interlacing flag and frame duration information are correctly passed to the video post-processor(deinterlacer) per video frame, it should not be a problem to decide whether deinterlacing should be performed or not. (I had did that for the standalone media player + TV recorder products 6 years ago)

If the display refresh rate is set to 59.94 fields or frames per second, it is going to have the judder effect on the 24 fps progressive contents. Only the high-end LCD and PDP TV's engines can detect and perform IVTC and restore the 60 telecined fields back to 24 progressive frames. (On the cost-down TV models, it is claimed capable of such processing, but it often does not show good results.)

On the good LCD / PDP, the panel driving mechanism is also often adjusted for the 24 fps FILM content's display optimization.

[Thunderbolt8]

what is recommended for madvr, leaving vsync to application controlled in CCC settings or rather change it to always on?