madVR - high quality video renderer (GPU assisted)

[pei]

For some reason when setting FFDshow OR VobSub subtitles to delay + or -, it doesn't work for me using ZoomPlayer with madVR and LAV decoder. Is this a known issue or any suggestions as to what might be going wrong

ZoomPlayer + MadVR + LAV + FFDshow Subtitles

[Lyris]

Thanks David. The results look pretty good with Jinc 3 AR here, but obviously I can't see how that compares to the other processing without access to one of those projectors.
Jinc can look a little soft with very fine details, but is preferable to aliasing or sharpening artifacts in my opinion.

Sony's DRC maybe avoids ringing a little better (hard to tell from photos) but Jinc seems to look a lot more natural with real picture content.

http://www.abload.de/img/sqrlgjuce.jpg


That looks exactly like what I ask for in the review: clean edge adaptive scaling without the other 'detail enhancement' tricks.

[dansrfe]

So, I'm going to buy a new GTX 670 but before I do I just want to make sure that it's worth it compared to getting a half price GTX 660 if all I'm going to use it for is playing bluray material. Possibly 2 blurays at the same time on a 2560 x 1440 resolution with the highest image upscaling algorithms. I don't plan to game with it at all.

The major difference I see between the 660 and 670 is that the 660 has a 192-bit vs 670's 256-bit Memory Interface Width.

[madshi]

Hi guys,
I'm David Mackenzie.

Sorry I didn't get a chance to test madVR. But, here's the original pattern I made:
http://www.mediafire.com/?8pd9y18uc5qfcdr
Thanks. FWIW, I've tried to match the photographs in your article by resizing a bit more (with Jinc3 AR), upping saturation by a monstrous 1.4 factor, then doing some rotation in photoshop. Here's the result without any added sharpening:

http://madshi.net/4k1.png

Of course the Sony projection photos are at a big disadvantage because all the limitions of the projector and digicam add up in making those photos look worse. Also my screenshot has an advantage because I had to upscale more than 2x to make the image size match your photos, so my screenshot simulates a projector with a higher than 4K resolution. I do wonder though why those claws look so much thinner (which is probably more true to life) in my screenshots compared to those projection photos. That's kinda weird...

[cyberbeing]

madVR in its current state uses only a very minimal amount of memory bandwidth, so that should be the least of your concerns.

The GTX 660 Ti should be the clear price/performance winner for madVR between the GTX 660 (~$220), GTX 660 Ti (~$290), and GTX 670 (~$360), especially if you decide to overclock.

[dansrfe]

cyberbeing,

From within this link (http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007709%20600030348%20600315498%20600311818&IsNodeId=1&name=GeForce%20GTX%20660%20Ti) which would be the best buy in your opinion? Thanks!

[cyberbeing]

If quietness is important, you should probably only look at the dual-fan non-reference models. I think nevcairiel and a couple others here own the ASUS DirectCU II model, and they have been satisfied with the noise level. As luck would have it, that model seems to be on sale (http://www.newegg.com/Product/Product.aspx?Item=N82E16814121656) at the moment as the cheapest non-reference option from NewEgg, even without the rebate.

[ryrynz]

A fine recommendation there from Cyberbeing, you really can't go wrong with that card.

[jmone]

I (and I think nevcairiel) have the ASUS DIrectCU II 660 (not the 660Ti). The noise level is OK not great in my Shuttle.... eg when pushed there is fan noise at the seating distance when the Audio is muted when working hard (eg 1080/60i material on Jinc3 AR).

[dansrfe]

Wow, the card looks great cyberbeing. Just ordered it along with the rest of the stuff for my new build! Thanks!

[Plutotype]

Thanks. FWIW, I've tried to match the photographs in your article by resizing a bit more (with Jinc3 AR), upping saturation by a monstrous 1.4 factor, then doing some rotation in photoshop.
Of course the Sony projection photos are at a big disadvantage because all the limitions of the projector and digicam add up in making those photos look worse. Also my screenshot has an advantage because I had to upscale more than 2x to make the image size match your photos, so my screenshot simulates a projector with a higher than 4K resolution. I do wonder though why those claws look so much thinner (which is probably more true to life) in my screenshots compared to those projection photos. That's kinda weird...

Thanks madshi, of course the saturation is too much ( the black is not black anymore ) and some very little sharpening would not hurt, but in general, JINC brings a very natural result.

[MSL_DK]

It's still on my list of things to look into.

:thanks:

[madshi]

It's very convenient to have mVR automatically hiding the mouse cursor after a few secs in FSE mode, but do you really find it to be mandatory for windowed playback? I mean you could let the movie play as "always on top" in the center of the screen in a small window and do other stuff simultaneously, atm it's a major PITA as the mouse cursor keeps disappearing :o
I know this is something minor but would it be possible to have a mouse pointer cursor (with the pointing finger) in exclusive mode?
madVR supports Microsoft's "IVideoWindow" interface to show/hide the mouse cursor. So it's the media player's decision when to show/hide it.

I'm not saying this is the definitive answer on ATI's/AMD's "Enforce Smooth Video Playback (ESVP)" feature, but it's an interesting read which includes feedback from AMD:

http://www.anandtech.com/show/3601/the-final-word-on-the-best-radeon-htpc-card
Ah, thanks!

Using Mad vr with Media Player Classic and FFDshow DXVA video decoder results in a crash of mpc when moving to a second screen, or setting media player classic to start playback on a secondry screen.
Do you get a crash with LAV Video Decoder? LAV should be superior to ffdshow for DXVA decoding.

Don't know if this is where to post this. The folks at ZoomPlayer suggested I contact the developer of madVR for my question. Here:

There is a cool program called SoftTH which allows you to use Direct3D to span multiple monitors for video games AND you can specify the bezel/border size so the picture looks consistent like you were watching (at a drive-in, if anyone knows what that is) through a car windshield with the frame blocking part of the view but the picture/video looks normal when you move or turn. With ZP Video Wall I can stretch over 3 monitors but the video gaps to the outside monitors. Is it possible to adjust for the gaps/bezels to span multiple monitors AND allow for the bezel? tx

I know this is not common, but thought I'd ask
Maybe in a far far away future version. But this is really *very* low priority for me at the moment.

Ineteresting, I've tried using madvr on the same machine but in windows xp (i need it for work) and madvr won't even display video with 1080p content, 720p stutters like crazy lol, using lav filters, geforce 670, 6 core cpu... weird... standard renderers play with no issues o.o
That's really weird. Are we talking about windowed or fullscreen exclusive mode? Does the problem occur in both modes? Or just in one of them? If you look at the madVR debug OSD (Ctrl+J) when getting stuttering with 720p videos, which queues are less than half full? Are you using the default madVR settings, or did you change anything? Maybe you turned on Jinc8 AR? Which is total overkill and not recommended at all?

What video card is necessary from the last that it was possible to reproduce 1080p@60fps@10bit@100MHz at jinc 8 taps hroma upsc?
ati ore nv?
7850-7970 ore 660-680
You should not use Jinc8. It's no good. Use Jinc3 AR. Then a 660 is plenty fast enough.

What I also noticed when doing this though, is that if the source image is not a multiple of 8, madVR does not seem to resize it correctly.
Neither does EVR, so it's a bug in the PNG source filter.

For some reason when setting FFDshow OR VobSub subtitles to delay + or -, it doesn't work for me using ZoomPlayer with madVR and LAV decoder. Is this a known issue or any suggestions as to what might be going wrong

ZoomPlayer + MadVR + LAV + FFDshow Subtitles
I don't think this is likely to have anything to do with madVR.

[leeperry]

madVR supports Microsoft's "IVideoWindow" interface to show/hide the mouse cursor. So it's the media player's decision when to show/hide it.
IIRC, before mVR smartened up PotP was only hiding the mouse cursor in FS mode when the transport bar wasn't visible...just like KMP and MPC.

To be perfectly clear when I'll whine to the PotP coder, you are saying that PotP sends the HideCursor (http://msdn.microsoft.com/en-us/library/windows/desktop/dd377276%28v=vs.85%29.aspx) message when playback is in progress, whatever windowed or FS?

All this said, everything works like a charm here and I have nothing to whine about anymore :p ..I only wish that the "windowed/exclusive" OSD messages could be disabled someday as they're annoying.

:thanks:

The GTX 660 Ti should be the clear price/performance winner for madVR between the GTX 660 (~$220), GTX 660 Ti (~$290), and GTX 670 (~$360)
for SD to 1080p in Jinc3 AR, the 660 runs 30% load at max....make it 60% for 59.94fps but you'll need to run 4K or so to require a 660Ti IMHO.....and the best bang/bucks is most likely the forthcoming 660SE, which is a 192bit 650Ti.

[iSunrise]

All this said, everything works like a charm here and I have nothing to whine about anymore :p ..I only wish that the "windowed/exclusive" OSD messages could be disabled someday as they're annoying.
If that is considered, please make that "Disable ALL OSD messages". All the relevant info should be in the ALT+J stats, anyway.

[madshi]

IIRC, before mVR smartened up PotP was only hiding the mouse cursor in FS mode when the transport bar wasn't visible...just like KMP and MPC.

To be perfectly clear when I'll whine to the PotP coder, you are saying that PotP sends the HideCursor (http://msdn.microsoft.com/en-us/library/windows/desktop/dd377276%28v=vs.85%29.aspx) message when playback is in progress, whatever windowed or FS?
I don't know what you mean with transport bar and I don't know if PotPlayer calls HideCursor or not. All I can say is that madVR support IVideoWindow::HideCursor, so the media player should be in full control. Of course it is possible that there's a bug in madVR, but unless a media player developer tells me he's calling HideCursor and madVR doesn't react properly, there's nothing I can do. Except if all media players show the same problem for many users. That would be a sign that there's a bug in madVR somewhere...

[mikser]

Updated to v0.85.7 and noticed a problem with audio and video slowly but surely de-syncing. When I for example (don't know about lesser files) try to play a 720p@60fps file (.ts/.mkv - doesn't matter), the video starts lagging behind the audio. In one minute of watching the lagging amounts to roughly 1-5 seconds, which is a lot (especially if I try to watch a fight and hear all the strikes way before I see them). When I skip back/forward (or pause) the tracks are once again synced, only to start de-syncing again after video starts rolling.

It seems that my computer (Intel Q6600 @ 2.40GHz, Nvidia GeForce 9800 GTX+, MPC-HC) is not fast enough to handle the file. Statistics show constantly growing numbers in "dropped frames" and "delayed frames" lines, but obviously this shouldn't result in de-syncing. This happens in the normal 100% pixel-to-pixel viewing, without any upscaling/downscaling.

On a slightly separate issue (don't know if you are aware of it and if this is a bug or a feature) it seems that the new version (in a basic 100% viewing, without upslacing/downscaling; no defragmentation issues) is also slower, as I seem to have noticebly more dropped frames than previously.


Comparison is made with v0.84.7, which seems to be faster and where there is no such de-sync issues.


-------------------------

May I also humbly suggest a few things?

1. Could you please re-think the way you put a "What's new" changelog? Now there is a lot of redundant information, when after each minor version you re-print all the features which changed since the last major version. So for example "added support for external DXVA2 decoders" and "added support for DXVA2 scaling" features are now presented seven times in a row, in info for all the versions between v0.85.1 and v0.85.7. It would be so much more convenient to only see the changes that really happened in the newest version, so a decision of (not) upgrading could be made quickly.

2. Now there is a visualization of the impact of different upscale/downscale algorithms on a picture quality. It's a super-useful feature! It would also be very useful to see the similar information regarding the performance of different algorithms and their impact on CPU/GPU. This would really help to decide what algorithms is best for each user, and you wouldn't have to spend your time answering e.g. "Don't use Jinc8 AR on your computer. Which is total overkill and not recommended at all.".

Thanks for a great codec!

[leeperry]

All I can say is that madVR support IVideoWindow::HideCursor, so the media player should be in full control.
Oops, I apologize for wasting your time as I just noticed that PotP also does it with EVR :rolleyes:

If that is considered, please make that "Disable ALL OSD messages". All the relevant info should be in the ALT+J stats, anyway.
Great idea, I really can't stand OSD messages...especially the "windowed/exclusive" ones as they are useless and appear a hundred times everyday...for nothing :o

[chuuey]

That's really weird. Are we talking about windowed or fullscreen exclusive mode? Does the problem occur in both modes? Or just in one of them? If you look at the madVR debug OSD (Ctrl+J) when getting stuttering with 720p videos, which queues are less than half full? Are you using the default madVR settings, or did you change anything? Maybe you turned on Jinc8 AR? Which is total overkill and not recommended at all?


I was playing in windowed mode and tried fullscreen exclusive - and I see that the display refresh rate simply drops to 0 for a few seconds sometimes - then the terrible slideshow starts - after a few seconds the refresh rate bumps up to 60 again and everything is ok, so it's a roundabout like that ;) I've been using Jinc3 AR on 7 but on XP i restored default settings just to be safe, but the problem is with the detected refresh rate falling to 0, for some reason. Maybe it's an Nvidia driver bug? Using the latest drivers and MPC-HC with LAV filters.

[xabregas]

do you have any suggestions which setting better to disable or enable in CCC, starting from default values?


I disabled everything except auto pulldown detection in video settings in CCC

Everything else disabled, new CCC 13.1 has better 3d settings, we can disable everything if we want ;)

In video settings, amd steady video disappeared which is good :D

still powerplay is on by default, but if u create a profile and edit the minimum clocks u r fine, it can be loaded at startup so every clock is correct now...

i`ve tested madvr now with coreavc and lav video (software decoding)= + lav audio and splitter and madvr at default settings is smooth, no drops for an entire episode of elementary...

i tested in some action movies in the most active scenes and no drops...

ah madvr settings at default but in upscaling chroma and luma i enabled AR, still i left bicubic 75 in chroma and lanczo 3 in luma...

Tried with jinc 3 taps, everything smooth also, no drops, but i didnt see any improvement whasoever, only thing that changed was gpu temps :sly:

but probably is because i upscale 1280x720 files to 1360x768 only, its a small upscale :scared:

[LaUs3r]

Hey guys,

I have the following setup for movie playback:
PC: i5-2500, 16GB RAM, Intel HD-Graphics (to AV-receiver via HDMI) & GTX560i (to monitor via DVI)

For movie playback I switch in Win7 to dual-screen mode to activate the LCD-TV which is connected to the av-receiver. Now when I start playing a movie in fullscreen on the LCD-TV I encounter frame droppes with movies in 24p. I configured mpc-hc to play 24p-content with 24p.
madvr shows correctly that the movie has 24p and that the LCD-TV display has 24Hz as well. But unfortunately I have lots of dropped frames resulting in a very stuttering playback.

When I disable the option on mpc-hc to play 24p-content with 24hz everything runs smoothly with 60Hz.

Any ideas on that?

[madshi]

Updated to v0.85.7 and noticed a problem with audio and video slowly but surely de-syncing. When I for example (don't know about lesser files) try to play a 720p@60fps file (.ts/.mkv - doesn't matter), the video starts lagging behind the audio. In one minute of watching the lagging amounts to roughly 1-5 seconds, which is a lot (especially if I try to watch a fight and hear all the strikes way before I see them). When I skip back/forward (or pause) the tracks are once again synced, only to start de-syncing again after video starts rolling.

It seems that my computer (Intel Q6600 @ 2.40GHz, Nvidia GeForce 9800 GTX+, MPC-HC) is not fast enough to handle the file. Statistics show constantly growing numbers in "dropped frames" and "delayed frames" lines, but obviously this shouldn't result in de-syncing. This happens in the normal 100% pixel-to-pixel viewing, without any upscaling/downscaling.
Hmmmm... Is this with interlaced videos? Does the madVR debug OSD (Ctrl+J) show that deinterlacing is turned on?

1. Could you please re-think the way you put a "What's new" changelog? Now there is a lot of redundant information, when after each minor version you re-print all the features which changed since the last major version. So for example "added support for external DXVA2 decoders" and "added support for DXVA2 scaling" features are now presented seven times in a row, in info for all the versions between v0.85.1 and v0.85.7.
I don't know where you get your changelogs from. In every announcement post in this thread I always only mention the changes that apply to each new version. There's nothing redundant. The "changelog.txt" shipping with madCodeHook works the same way, nothing redundant, either. However, the first post in this thread shows all changes compared to the previous functionality upgrade.

2. Now there is a visualization of the impact of different upscale/downscale algorithms on a picture quality. It's a super-useful feature! It would also be very useful to see the similar information regarding the performance of different algorithms and their impact on CPU/GPU. This would really help to decide what algorithms is best for each user, and you wouldn't have to spend your time answering e.g. "Don't use Jinc8 AR on your computer. Which is total overkill and not recommended at all.".
This has been asked and anwered before. Short answer: The performance cost of the different algorithms depends on a million of factors and there's no simple graph I could possibly reduce this to.

I was playing in windowed mode and tried fullscreen exclusive - and I see that the display refresh rate simply drops to 0 for a few seconds sometimes - then the terrible slideshow starts - after a few seconds the refresh rate bumps up to 60 again and everything is ok, so it's a roundabout like that ;) I've been using Jinc3 AR on 7 but on XP i restored default settings just to be safe, but the problem is with the detected refresh rate falling to 0, for some reason. Maybe it's an Nvidia driver bug? Using the latest drivers and MPC-HC with LAV filters.
The refresh rate measurement going to 0 and then to 60 again indicates that your GPU/driver doesn't report the VSync scanline position information to madVR in any reliable way. This is a rather weird problem because nobody else seems to suffer from it. I've seen this reported from some people with 120Hz display modes, but never yet with just 60Hz. Is it possible that you disabled the "present several frames in advance" option in madVR? In win7 having this enabled (which is the default setting) might fix your problem. In XP it won't help, though. Is there anything special about your hardware/driver setup? I'm really wondering why only you seem to have this problem and nobody else. Have you tried different GPU driver versions?

I have the following setup for movie playback:
PC: i5-2500, 16GB RAM, Intel HD-Graphics (to AV-receiver via HDMI) & GTX560i (to monitor via DVI)

For movie playback I switch in Win7 to dual-screen mode to activate the LCD-TV which is connected to the av-receiver. Now when I start playing a movie in fullscreen on the LCD-TV I encounter frame droppes with movies in 24p. I configured mpc-hc to play 24p-content with 24p.
madvr shows correctly that the movie has 24p and that the LCD-TV display has 24Hz as well. But unfortunately I have lots of dropped frames resulting in a very stuttering playback.

When I disable the option on mpc-hc to play 24p-content with 24hz everything runs smoothly with 60Hz.

Any ideas on that?
There's quite a difference between 23.976 and 24.000. Is your content 23.976 or 24.000? And your refresh rate? If possible, please make a screenshot of the madVR debug OSD (Ctrl+J) in the situation when playback stutters.

[madshi]

madVR now has a bug tracker:

http://madVR.bugs.madshi.net

You can browse the bug reports without being logged in. In order to create bug reports you need to login. Please in the future if you have a clear bug to report, use the bug tracker. If you're not sure if your problem is a bug in madVR you can ask here in this thread first. Currently please use the tracker only for bug reports. Feature wishes will be silently deleted.

[LaUs3r]

There's quite a difference between 23.976 and 24.000. Is your content 23.976 or 24.000? And your refresh rate? If possible, please make a screenshot of the madVR debug OSD (Ctrl+J) in the situation when playback stutters.

here are the mpc-hc settings:
http://www11.picfront.org/picture/EW2Ysbse/thb/mpc-hc_options.jpg (http://www.picfront.org/d/8Vqu)

and here's the madvr info:
http://www2.picfront.org/picture/wRrXMPYTPan/thb/screenshot.jpg (http://www.picfront.org/d/8Vqv)

cheers

[madshi]

@LaUs3r, the composition rate is wrong. It's still at 60Hz, although the refresh rate is 24Hz. Try using the madVR display mode switcher instead of MPC-HC. That should fix the problem.

[chuuey]

The refresh rate measurement going to 0 and then to 60 again indicates that your GPU/driver doesn't report the VSync scanline position information to madVR in any reliable way. This is a rather weird problem because nobody else seems to suffer from it. I've seen this reported from some people with 120Hz display modes, but never yet with just 60Hz. Is it possible that you disabled the "present several frames in advance" option in madVR? In win7 having this enabled (which is the default setting) might fix your problem. In XP it won't help, though. Is there anything special about your hardware/driver setup? I'm really wondering why only you seem to have this problem and nobody else. Have you tried different GPU driver versions?


Well I don't have any of these problems on this machine using win7, only happens in xp. And the software configuration is pretty much the same on both operating systems, same driver versions etc. I was thinking to change the V-sync settings in the nvidia control panel - i tried adaptive and on, even off... Same damn thing I don't know what the heck is wrong. I tried earlier versions of madVR and the only that works fine on this xp install is 0.09. From version 0.10 the same issue occurs. Ofcourse i've disabled all the options that require Vista and newer. I wouldn't have even bothered reporting this but for the time being I use xp more often and I can always watch something on win7. But would be nice to be able to figure this out ;) By the way - I just changed the size of cpu and gpu queues (using latest madVR) to 4 - and 720p videos now play without choppiness, however the refresh rate falls to around 52-54 sometimes, but at least the slideshow's are gone, but 1080p video just displays a black screen, there's audio though ;)

[mikser]

Hmmmm... Is this with interlaced videos? Does the madVR debug OSD (Ctrl+J) show that deinterlacing is turned on?
Just a normal progressive 720p@60fps video. Deinterlacing is turned off. Here are the details of one of the files. But the problem exists with other files as well. If the file is small and computing resources are enough, then it's unnoticeble (but probably still exists). Like I said, problem didn't and doesn't exist in the older version.

General
Unique ID : 223481042210418997022924404197243259278 (0xA820E21420236B7D9AA40EE45748898E)
Complete name : C:\Bellator.85.720p.HDTV.x264-Secludedly.mkv
Format : Matroska
Format version : Version 2
File size : 4.79 GiB
Duration : 1h 35mn
Overall bit rate mode : Variable
Overall bit rate : 7 176 Kbps
Encoded date : UTC 2013-01-19 00:42:41
Writing application : mkvmerge v5.8.0 ('No Sleep / Pillow') built on Sep 2 2012 15:37:04
Writing library : libebml v1.2.3 + libmatroska v1.3.0

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : Main@L4.0
Format settings, CABAC : Yes
Format settings, ReFrames : 4 frames
Format settings, GOP : M=4, N=32
Muxing mode : Header stripping
Codec ID : V_MPEG4/ISO/AVC
Duration : 1h 35mn
Bit rate mode : Variable
Maximum bit rate : 20.0 Mbps
Width : 1 280 pixels
Height : 720 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 60.000 fps
Original frame rate : 59.940 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Default : Yes
Forced : No
Color primaries : BT.709
Transfer characteristics : BT.709
Matrix coefficients : BT.709

Audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : A_AAC
Duration : 1h 35mn
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 KHz
Compression mode : Lossy
Delay relative to video : -11ms
Default : Yes
Forced : No



I don't know where you get your changelogs from. In every announcement post in this thread I always only mention the changes that apply to each new version. There's nothing redundant. The "changelog.txt" shipping with madCodeHook works the same way, nothing redundant, either. However, the first post in this thread shows all changes compared to the previous functionality upgrade.
Yes, seems like I was wrong. Sorry for that. I usually check the changelog here:
http://www.videohelp.com/tools/madVR/version-history#changelog


This has been asked and anwered before. Short answer: The performance cost of the different algorithms depends on a million of factors and there's no simple graph I could possibly reduce this to.
So there is no way to even generalize on a scale of "algorithm X is usually 1.5-2 times slower than algorithm Y, and 2-2.5 times slower if AR is turned on"? Such a pity. :(

[cyberbeing]

for SD to 1080p in Jinc3 AR, the 660 runs 30% load at max....make it 60% for 59.94fps but you'll need to run 4K or so to require a 660Ti IMHO.....and the best bang/bucks is most likely the forthcoming 660SE, which is a 192bit 650Ti.
Maybe for scaling to 1920x1080 only, something like a 660SE would be sufficient, but that's not what dansrfe was doing. He had a 2560x1440 monitor which raises the processing requirement quite a bit, especially when deinterlacing 1080i to 60fps. Downscaling 3840x2160 to 2560x1440 with AR + linear light, I also expect would be quite rough. I maintain that the GTX 660 Ti is the best price/performance value for 2560x1440 in dansrfe's price range.

[LaUs3r]

@LaUs3r, the composition rate is wrong. It's still at 60Hz, although the refresh rate is 24Hz. Try using the madVR display mode switcher instead of MPC-HC. That should fix the problem.

I think I found the cause of this behavior. When I start mpc-hc on the LCD-TV (dual screen enabled) the composition rate always stays at 60Hz. This is the refresh rate of my PC-display (primary display). the refresh rate of the movie doesn't matter. madVR recognizes it correctly, the display refresh rate is changes accordingly but the composition rate always stays at 60Hz.
If I switch to single screen, so only using the LCD-TV, everything works great and the composition rate is changed accordingly.

[madshi]

Well I don't have any of these problems on this machine using win7, only happens in xp.
Ah, interesting. Maybe there's some sort of problem with the driver. You could try installing a different version, or maybe just uninstall and reinstall the same, hoping for an improvement. But I can't really say if it will help.

Just a normal progressive 720p@60fps video. Deinterlacing is turned off.
Weird. Then I can't explain why A/V sync gets lost. Maybe a debug log could help getting to the bottom of this. Alternatively it could also help if you could check older madVR builds to see which exact version introduced the problem.

So there is no way to even generalize on a scale of "algorithm X is usually 1.5-2 times slower than algorithm Y, and 2-2.5 times slower if AR is turned on"?
No. It behaves very differently, depending on GPU model, scaling factor and many other things.

[madshi]

I think I found the cause of this behavior. When I start mpc-hc on the LCD-TV (dual screen enabled) the composition rate always stays at 60Hz. This is the refresh rate of my PC-display (primary display). the refresh rate of the movie doesn't matter. madVR recognizes it correctly, the display refresh rate is changes accordingly but the composition rate always stays at 60Hz.
If I switch to single screen, so only using the LCD-TV, everything works great and the composition rate is changed accordingly.
I'm not sure if Windows is able to use different composition rates for primary and secondary monitors. Anyway, I'd still recommend trying the madVR display mode changer instead of the MPC-HC one, maybe the madVR one will work. If not, there are 2 or maybe 3 alternative solutions:

(1) Use fullscreen exclusive mode, then composition rate doesn't matter.
(2) Tell madVR to disable Aero.
(3) Don't know if this works, but you could try if Overlay mode works without the frame drops. No idea, though, really. It might not help.

[cyberbeing]

I have the same problem a LaUs3r with my GT440 & 310.90 drivers.

Main monitor = CRT (1920x1080 @119.88Hz or 1600x1200 @95.904Hz Analog BNC)
Secondary monitor = TV (1920x1080 @23.976Hz HDMI).

Whenever I play a video on the TV @23.976Hz, most of the time my composition rate will show as 119.88Hz or 96Hz of my main monitor. Other times I've seen madVR report a composition rate of 59.94Hz or 29.97Hz (especially when exiting Fullscreen Exclusive), yet never 23.976Hz. It makes no difference if start playing from the TV, drag a playing video from main monitor to TV, already have 23.976Hz set, use the madVR refresh rate switcher to change to 23.976Hz, nor reloading MPC-HC & madVR with CTRL+E.

From what you're saying, it sounds like composition rate isn't something madVR has control over? Assuming madVR is always reporting composition rate correctly, would having a higher multiple composition rate have a negative effect on madVR's vsync logic in Windowed mode?

[madshi]

I don't know who ultimately decides the composition rate. I'd guess it's somewhere in the depth of the OS, or maybe the GPU drivers have something to do with it, too. At least there's no way to control it for normal applications. All madVR does is disable and reenable Desktop Composition whenever it switches display modes. That usually helps "updating" the composition rate. I suppose Windows just doesn't like different composition rates for primary and secondary monitors, but I don't really know. As I said, using fullscreen exclusive mode should help, as should disabling Aero while playing back video.

I'm not sure exactly what happens if composition rate and refresh rate don't match, but it seems to result in a lot of trouble.

[madshi]

First of all, thanks for your continued hard work on MadVR. A really good piece of software. :cool:

I've encountered a rather strange bug, happens randomly and with quite specific settings. Here's how to reproduce:

Enable both "use separate device for presentation" and "use separate device for DXVA processing".
Set "DXVA2" for image up-/downscaling.
Play a h264 10bit file in windowed mode, while using downscaling (or upscaling).
The video is "jumping", as if instead of next frame, the previous frame gets rendered.
What I mean is instead of frames 1 2 3 4, you get frames 1 2 1 4 or something.


Same file works fine with one of the "use separate device ..." options disabled or when using either other scaling algorithms or no scaling. Also other files work perfectly.

Link to sample:
www.mediafire.com/?ca88a1buue7g2dy
and log:
www.mediafire.com/?z72h4726xh4v859

Basic system info:
Intel Core i5 3570K
Intel HD 4000 + drivers v2875
Windows 7 (x64)
MadVR v0.85.7
MPC-HC Lite 1.6.6.6391 (32bit of course)
LAV Filters 0.54.1, but also happens if using MadVR's internal decoding
Unfortunately I don't seem to be able to reproduce the problem. Anyway, does this test build fix the problem?

http://madshi.net/madVR16635.rar

[LaUs3r]

I'm not sure if Windows is able to use different composition rates for primary and secondary monitors. Anyway, I'd still recommend trying the madVR display mode changer instead of the MPC-HC one, maybe the madVR one will work. If not, there are 2 or maybe 3 alternative solutions:

(1) Use fullscreen exclusive mode, then composition rate doesn't matter.
(2) Tell madVR to disable Aero.
(3) Don't know if this works, but you could try if Overlay mode works without the frame drops. No idea, though, really. It might not help.

I use the madvr display mode changer. But if I'm in dual screen mode, madVR does not to apply the correct refresh rate for the LCD-TV. It stays at 60Hz for the composition and display rate. As mentioned before, if I only use the LCD-TV (single screen) everything works fine. Seems to me that madVR doesn't recognize that the playback happens on screen 2 and not on screen 1.

[madshi]

That seems weird. Maybe a madVR debug log could help finding out why madVR doesn't switch.

[Qotscha]

I have a problem watching interlaced content in secondary display when using dxva2 decoding. When I have fullscreen exclusive mode enabled in the secondary display and I then open a new file in my main monitor, madVR changes the refresh rate to 25/30 fps instead of 50/60 fps and playback gets very stuttery.

However, the problem can be avoided with "if in doubt, activate deinterlacing" setting. But the strange thing is that even with "if in doubt, deactivate deinterlacing" setting, madVR OSD tells that deinterlacing is on. And if I change source type setting during playback from auto-detect (Video) to Video, madVR changes to correct refresh rate. Or if source type is set to video instead of auto-detection, changing it to auto-detect (Video) will change to correct refresh rate.

I have tested this with LAV (native), MPC HC, and ffdshow dxva decoders with h.264 and MPEG2 content. Software decoders work fine.

I'm using 64 bit Windows 7 and my graphics card is Radeon HD 6850 with Catalyst 12.11 beta 11 drivers.

Here (https://www.dropbox.com/s/kio66yeb24ae0z5/madVR%20-%20log.txt) is a log containing opening a file, changing to fullscreen in secondary display, reopening it, chancing source type from auto-detect (Video) to Video and quitting MPC HC.

[Dodgexander]

Do you get a crash with LAV Video Decoder? LAV should be superior to ffdshow for DXVA decoding.


It just happens with FFDSHOW dxva, not lav. Strangely only on a secondry screen.

I have a problem watching interlaced content in secondary display when using dxva2 decoding. When I have fullscreen exclusive mode enabled in the secondary display and I then open a new file in my main monitor, madVR changes the refresh rate to 25/30 fps instead of 50/60 fps and playback gets very stuttery.

However, the problem can be avoided with "if in doubt, activate deinterlacing" setting. But the strange thing is that even with "if in doubt, deactivate deinterlacing" setting, madVR OSD tells that deinterlacing is on. And if I change source type setting during playback from auto-detect (Video) to Video, madVR changes to correct refresh rate. Or if source type is set to video instead of auto-detection, changing it to auto-detect (Video) will change to correct refresh rate.

I have tested this with LAV (native), MPC HC, and ffdshow dxva decoders with h.264 and MPEG2 content. Software decoders work fine.

I'm using 64 bit Windows 7 and my graphics card is Radeon HD 6850 with Catalyst 12.11 beta 11 drivers.

Here (https://www.dropbox.com/s/kio66yeb24ae0z5/madVR%20-%20log.txt) is a log containing opening a file, changing to fullscreen in secondary display, reopening it, chancing source type from auto-detect (Video) to Video and quitting MPC HC.

I had this problem before, now it doesn't happen anymore. Originally I fixed it by disabling desktop composition in Madvr settings, but now using separate device for presentation in options does the trick. That is of course if the cause of your problem was the same as mine.

Oh, also you can try disabling/enabling fullscreen exclusive mode.

[michkrol]

Unfortunately I don't seem to be able to reproduce the problem. Anyway, does this test build fix the problem?

http://madshi.net/madVR16635.rar

Thanks for your time.

For 10bit videos, it's worse than before - the windowed mode doesn't work with "separate DXVA device" and DXVA scaling selected, also the fullscreen mode stopped working with DXVA scaling. By not working I mean the player quits silently, almost as if closed by user - can't get a crashlog. Fullscreen without scaling works normally (tested by setting player to start in fullscreen).

For 8bit videos it works without scaling in both windowed and fullscreen. Doesn't work with DXVA scaling (both windowed and fullscreen). Doesn't matter whether I use software or hardware decoding.

For the record, since my last post, I've update MPC-HC to 6609, LAVFilters to 0.55.1 and Intel drivers to 2932.

A pity I can't provide more useful info :(

[madshi]

By not working I mean the player quits silently, almost as if closed by user
Argh, sorry, my fault. How about this one?

http://madshi.net/madVRdxvaScalingTest.rar

I have a problem watching interlaced content in secondary display when using dxva2 decoding. When I have fullscreen exclusive mode enabled in the secondary display and I then open a new file in my main monitor, madVR changes the refresh rate to 25/30 fps instead of 50/60 fps and playback gets very stuttery.

However, the problem can be avoided with "if in doubt, activate deinterlacing" setting. But the strange thing is that even with "if in doubt, deactivate deinterlacing" setting, madVR OSD tells that deinterlacing is on. And if I change source type setting during playback from auto-detect (Video) to Video, madVR changes to correct refresh rate. Or if source type is set to video instead of auto-detection, changing it to auto-detect (Video) will change to correct refresh rate.

I have tested this with LAV (native), MPC HC, and ffdshow dxva decoders with h.264 and MPEG2 content. Software decoders work fine.

I'm using 64 bit Windows 7 and my graphics card is Radeon HD 6850 with Catalyst 12.11 beta 11 drivers.

Here (https://www.dropbox.com/s/kio66yeb24ae0z5/madVR%20-%20log.txt) is a log containing opening a file, changing to fullscreen in secondary display, reopening it, chancing source type from auto-detect (Video) to Video and quitting MPC HC.
Ok, I understand why madVR is reacting this way. Let me see if I can fix it for the next build.

[madshi]

I have a problem watching interlaced content in secondary display when using dxva2 decoding. When I have fullscreen exclusive mode enabled in the secondary display and I then open a new file in my main monitor, madVR changes the refresh rate to 25/30 fps instead of 50/60 fps and playback gets very stuttery.

However, the problem can be avoided with "if in doubt, activate deinterlacing" setting. But the strange thing is that even with "if in doubt, deactivate deinterlacing" setting, madVR OSD tells that deinterlacing is on. And if I change source type setting during playback from auto-detect (Video) to Video, madVR changes to correct refresh rate. Or if source type is set to video instead of auto-detection, changing it to auto-detect (Video) will change to correct refresh rate.

I have tested this with LAV (native), MPC HC, and ffdshow dxva decoders with h.264 and MPEG2 content. Software decoders work fine.

I'm using 64 bit Windows 7 and my graphics card is Radeon HD 6850 with Catalyst 12.11 beta 11 drivers.

Here (https://www.dropbox.com/s/kio66yeb24ae0z5/madVR%20-%20log.txt) is a log containing opening a file, changing to fullscreen in secondary display, reopening it, chancing source type from auto-detect (Video) to Video and quitting MPC HC.
P.S: Just thought about it. Although I might find a way to fix that, you could already fix it yourself right now. Your list of display modes contains 25/29/30 refresh rates. I'm surprised your display even supports those. Most displays do not. My suggestion would be to remove those. 25p content can be played back at 50Hz with no problems. 29p/30p content can be played back at 59.940Hz/60Hz with no problems. And there is very very few native 29p/30p content out there to begin with.

[ryrynz]

Madshi, do you have any intention of allowing chapter markers on the seekbar?

[hannes69]

I want to calibrate my Infocus X9 DLP 720p projector. Question about "the display expects the following RGB output levels" option in devices properties section of madvr: How do I determine what levels a device expects? In datasheet of projector nothing is to find about that. The projector is connected by HDMI so both sort of levels are possible AFAIK. In projectors menu there is NO option to select HDMI levels. I use the AVC test pattern with the vertical bars to setup black level. I have an AMD HD4550 card and in CCC 4:4:4 RGB Full and Full Dynamic Range (0-255) is selected. When selecting PC Levels in madvr I never see bars below 16 or above 235 (BTB and WTW lost, no problem for me). When using TV levels in madvr I have to adjust brightness and contrast level in the projector to cut BTB and WTW.
And thats the question: Have both variants really the same result? I dont´t remember exactly, but I have read somewhere else that the result will NOT be the same, there is some kind of level compression in one case.
What do you think of the following idea: My projector also has a VGA input and my graphic card a VGA output. VGA is always PC levels. By using that connection (and madvr PC levels) I adjust brightness and contrast with the AVC test pattern. And then switch to HDMI connection and then I´ll see the right setting. That would be possible if the projector handles both signal types (HDMI and VGA) in the ame way. Hope you understand what I mean...:devil:
Or is there another possibility I´m not aware of to check out by an objective way the expected level range?
Hannes.

[michkrol]

How about this one?

http://madshi.net/madVRdxvaScalingTest.rar


It works correctly, thank you!

As a side-note, the average 'scale' times go way up, (from 0,23ms to 1,45ms for the file I've tested) while using 'separate DXVA device'. No frame drops or anything. I don't think it's worth more of your time, maybe Intel('s driver) doesn't like it. I'll just disable it until I have time for more extensive tests. Thanks again.

[huhn]

Madshi, do you have any intention of allowing chapter markers on the seekbar?
it is on very low priority if i'm not mistaken

u mean the fullscreen exclusive seekbar right?

[madshi]

Madshi, do you have any intention of allowing chapter markers on the seekbar?
Asked before. Not planned for the near future.

When selecting PC Levels in madvr I never see bars below 16 or above 235 (BTB and WTW lost, no problem for me). When using TV levels in madvr I have to adjust brightness and contrast level in the projector to cut BTB and WTW.
And thats the question: Have both variants really the same result?
After calibration is done, you should *not* see BTB. There are conflicting opinions on WTW, but the classic approach is to "calibrate it away", too. Sounds to me PC levels output in madVR is the correct setting in your case. I would *not* calibrate VGA and then switch to HDMI.

As a side-note, the average 'scale' times go way up
That's because I've added a "flush" after the scaling when you "use a separate device for DXVA processing". With the old version there was no flush, which means that the "scale" time measurements were probably not correct. I'm always saying that you shouldn't put too much stock into the measured rendering times. The only thing that counts is if playback is smooth.

[pururin]

I've experienced what michkrol said as well with only "use separate device for DXVA processing" option enabled, video is very jumpy.
EDIT: A bit weird but I can barely reproduce it now after some re-instalings and configuring, very few jumps or most of the times none at all.

IIRC DXVA2 is no good for both Nvidia and ATI, on the contrary some suggest Intel iGPU owner to use this.
madshi did you add this scaling method just as another option for testing, or if it has advantages for Intel iGPU?

Any comment/info about this scaling? Maybe it's really good quality to performance ratio wise.

[hannes69]

After calibration is done, you should *not* see BTB. There are conflicting opinions on WTW, but the classic approach is to "calibrate it away", too. Sounds to me PC levels output in madVR is the correct setting in your case. I would *not* calibrate VGA and then switch to HDMI.

Yes, i want to kill BTB and WTW. But I can achieve this by 2 setups:
1. madvr outputs 0-255. On the projector no BTB and WTW appear, regardless of contrast and brightness setting. So I adjust brightness in such a way that 17 is barely visible and contrast so that 234 is barely visible.
2. madvr outputs 16-235. On the projector BTB and WTW appear. I kill BTB and WTW with right contrast and brightness setting.

Of course the desktop level is different in both cases, but I use the projector only for movies so I don´t care about desktop level.
The only question is if both variants deliver the SAME video levels or one of the two variants is the right and the other one the wrong one.

[nevcairiel]

Personally i would prefer 0-255 because that means you actually have 255 values to spread the RGB out over, and with 16-235 you need to squeeze it in 220 different values, it may not be much in terms of bits, but maybe it helps to avoid banding once in a while. :)

[Qotscha]

P.S: Just thought about it. Although I might find a way to fix that, you could already fix it yourself right now. Your list of display modes contains 25/29/30 refresh rates. I'm surprised your display even supports those. Most displays do not. My suggestion would be to remove those. 25p content can be played back at 50Hz with no problems. 29p/30p content can be played back at 59.940Hz/60Hz with no problems. And there is very very few native 29p/30p content out there to begin with.

Thanks for your answer. In fact, I tested removing those display modes from the list some weeks ago, but I forgot to mention that now, sorry. However, as you wrote, that fixed the problem.

My secondary display is a LG HDTV (42LM620T), which supports those refresh rates, so I have enabled them. I have some 25p/29p TV rips, but of course they are from 50i/59i sources. If I play them at 50/59 Hz, then every frame is just presented twice and the result is the same, right?