Media Player Classic Home Cinema (MPC-HC) - DXVA!

[v0lt]

...and I kind of hoped this would have solved my ticket (#341). Yeah well, maybe text time.
You can use ffdshow.

[lych_necross]

[...]I think VS2008 build should be compatible Windows SP2. At least there should be no BSODs.
No it shouldn't. Install SP3 or upgrade to Windows 7. If you are using a computer where you are not able to obtain admin privileges (i.e. work computer), you probably shouldn't be using MPC-HC on it anyways.

[nevcairiel]

If your work computer is still on XP SP2, hire a new admin :P

[Tom Keller]

Oh c'mon folks. SP2 is not THAT uncommon for XP users :rolleyes: . Stop hitting people, which are alreads lying on the ground :p ...

But besides that: i already wrote, that i found the file which causes the bsod... and the solution. And isn't it good to know this - just in case a few other stupid and immature idiots (like me) are still using SP2 and refuse to install SP3 or buy a new os :p ? So they now just have to download a tiny little hotfix instead of a whole service pack to get rid of their trouble. Hurray to the blessings of modern communication forms and "friendly" forum communities ;) .

[CruNcher]

From the performance point it makes no real sense though and interoperability wise more problems will arise as most software expects SP3 these days but everyone his own :) Depending on your CPU in SP2/SP3 the most slowing down thing is DEP and if you have a CPU with Hardware DEP its no real problem @ all and if not you can also disable DEP in SP3 if you need the better performance (and don't care about Security, or lets say don't believe really that DEP in any way can protect you without ASLR ;) ).

[entrecour]

Hi all

I have been watching Toy Story 3 blu-ray and notice after 5 or 6 minutes, the audio will have gone out of sync significantly enough for everyone to notice.

mark0077, I think there are some threads along similar lines (audio sync) over at Slysoft. The advice there is - don't play the m2ts files directly as this will cause audio / sub problems.

Have you used tsmuxer to generate a single m2ts / blue ray structure?

[LigH]

@ Ingram:

Are you certain the video to be played is DXVA compliant (with all the limitations DXVA requires)?

"MediaInfo -f ..." would not only report details about the MKV container, but also about the included video stream.

[Ingram]

@ Ingram:

Are you certain the video to be played is DXVA compliant (with all the limitations DXVA requires)?

"MediaInfo -f ..." would not only report details about the MKV container, but also about the included video stream.

Thanks but I just spent 20 hours re-encoding the video to H264 and it works now. Couldn't be stuffed trying to work out why it didn't work.

[v0lt]

No it shouldn't.
If we do not care about compatibility, then why do we need different builds except VS2010?

[tetsuo55]

From the performance point it makes no real sense though and interoperability wise more problems will arise as most software expects SP3 these days but everyone his own :) Depending on your CPU in SP2/SP3 the most slowing down thing is DEP and if you have a CPU with Hardware DEP its no real problem @ all and if not you can also disable DEP in SP3 if you need the better performance (and don't care about Security, or lets say don't believe really that DEP in any way can protect you without ASLR ;) ).MPC-HC won't work without DEP, it will give you an error about DEP being required and then exit.

[mark0077]

mark0077, I think there are some threads along similar lines (audio sync) over at Slysoft. The advice there is - don't play the m2ts files directly as this will cause audio / sub problems.

Have you used tsmuxer to generate a single m2ts / blue ray structure?

Hi. It's quite the opposite for me. Playing the m2ts files gives no problems. Playing from index.bdmv begins to shoe audio desync. Doesn't happen with power DVD but does with mpc. I'll try with other blurays that have many m2ts to see if I can confirm this is where mpc is having trouble.

[Jong]

If I remember correctly back a couple of years Madshi had to do quite a bit of work to get seamless branching titles to sync.

Edit: Yes! See here: http://forum.doom9.org/showthread.php?p=1176342#post1176342

Sounds like mpc-hc is producing a "crude" playlist from the individual files and not handling the joins properly.

[SamuriHL]

Hi. It's quite the opposite for me. Playing the m2ts files gives no problems. Playing from index.bdmv begins to shoe audio desync. Doesn't happen with power DVD but does with mpc. I'll try with other blurays that have many m2ts to see if I can confirm this is where mpc is having trouble.

Sounds like a seamless branching issue. Again. Still? :) I know that some people have worked on seamless branching before but it definitely sounds like there's an issue there still.

[SamuriHL]

If I remember correctly back a couple of years Madshi had to do quite a bit of work to get seamless branching titles to sync.

Edit: Yes! See here: http://forum.doom9.org/showthread.php?p=1176342#post1176342

Sounds like mpc-hc is producing a "crude" playlist from the individual files and not handling the joins properly.

You're faster than me, Jong. :D

[entrecour]

Use clown bd (eac3to / tsmuxer) to extract the film + audio track that you want to keep. It will generate a file that shoudl not have any audio synch issues.

[Jong]

You're faster than me, Jong. :D:D Pure luck! First time here in months :o

[SamuriHL]

Use clown bd (eac3to / tsmuxer) to extract the film + audio track that you want to keep. It will generate a file that shoudl not have any audio synch issues.

That's a good workaround, but, it'd be nice if this could be fixed. There's a lot of seamless branching titles out there. If I rent a title, I don't necessarily want to have to remux the title just to watch it, but, MPC-HC provides an AWESOME movie only viewing scenario that skips all the previews and menus. Just a thought.

[SamuriHL]

:D Pure luck! First time here in months :o

LOL :) You provided better info. I forgot about madshi's experience with eac3to on this one. It's why ClownBD uses it to extract the title instead of tsMuxer because that doesn't properly join, either.

[clsid]

...and I kind of hoped this would have solved my ticket (#341). Yeah well, maybe text time.
Supply a sample file, and I might be able to add MJPEG-B support to ffdshow.

[mark0077]

Sounds like a seamless branching issue. Again. Still? :) I know that some people have worked on seamless branching before but it definitely sounds like there's an issue there still.

Yeah definitely. Using vlc, i get no audio delay at all but its not seamless at all, it gives an ugly blip on transitions from one m2ts to the other. Anyone interested in renting out Toy Story 3 or trying other blu-rays to see can they fix it :)

[v0lt]

Supply a sample file, and I might be able to add MJPEG-B support to ffdshow.
4 examples are here - http://samples.mplayerhq.hu/V-codecs/MJPEGs/mjpa-b.mov/

Also I ask to see the file - http://samples.mplayerhq.hu/V-codecs/MJPEGs/dmb1/Dimitri%20Walking.mov ('dmb1' - Matrox MJPEG)

All these files are decoded LEAD MCMP/MJPEG Codec 2.0.

[lych_necross]

If we do not care about compatibility, then why do we need different builds except VS2010?
Because VS2010 is new and some programmers haven't upgraded to it yet. Besides, I wasn't referring to VS2008 and VS2010 builds, I was referring to clients running XP SP2. Microsoft doesn't support SP2 anymore. Even VS2008 requires SP3 to function. If a user doesn't want to apply SP3, they should be forced to use an older build of MPC-HC.

[betaking]

MPC-HC won't work without DEP, it will give you an error about DEP being required and then exit.

Are you sure?
Revision 1366 - Directory Listing
Modified Sat Nov 28 21:04:28 2009 UTC (12 months, 2 weeks ago) by Spec-Chum
Added /GS, /NXCOMPACT and /DYNAMICBASE compile options, where available

Patch by XhmikosR
This version changes the code in the DataExecutionPrevention = "1" to DataExecutionPrevention = "2" feedback I've caused the problem! Today, I put this back to 1, re-compiled under the latest version of mpc-hc, and DEP has not closed. mpc-hc work as usual. Of course the system into 2 XPSP3 have no problem. but VISTASP2 and WIN7 including WIN7SP1 the code DataExecutionPrevention = "2" does not change DataExecutionPrevention = "1" compilation of mpc-hc is easy to cause. But if you disable the two systems DEP, mpc-hc is little cause! I do not know the player turn DEP on the development of any help, at least I did not play with other DEP encountered problems, such as potplayr or Mplyer! Unstable recommended if the mpc-hc code can be DataExecutionPrevention = "2", but if it is stable in mpc-hc code is best DataExecutionPrevention = "1", because it affects the user's impression!
PS:There was consulted to do this under Why do not we change the user's advice?
Maybe back at that time in accordance with XhmikosR me, this is because the corresponding decoder is not compatible and caused isolated problems, if the decoder and separator is open source also easy to handle. Can be resolved by modifying the source code. But if not open source, or only separated by the decoder and decoder play. It is not without any solution can only disable the DEP?

[clsid]

4 examples are here - http://samples.mplayerhq.hu/V-codecs/MJPEGs/mjpa-b.mov/

Also I ask to see the file - http://samples.mplayerhq.hu/V-codecs/MJPEGs/dmb1/Dimitri%20Walking.mov ('dmb1' - Matrox MJPEG)

All these files are decoded LEAD MCMP/MJPEG Codec 2.0.
I already tried with those samples, but playback does not work properly. Decoding seems to work OK, but playback just freezes after decoding the first frame. Seeking causes other frames to be decoded successfully. The freezing could be a problem in ffdshow, but might also be some kind of splitter issue.
A different, longer, sample would help to verify if this is a general problem, or specific to those samples.

The video in the sample with the 'dmb1' FourCC isn't recognized by the MP4 splitter, so I can't add that one yet.

[clsid]

DEP can easily be disabled if desired:
Index: src/apps/mplayerc/mplayerc.vcproj
===================================================================
--- src/apps/mplayerc/mplayerc.vcproj (revision 2781)
+++ src/apps/mplayerc/mplayerc.vcproj (working copy)
@@ -79,7 +79,7 @@
SubSystem="2"
LargeAddressAware="2"
RandomizedBaseAddress="2"
- DataExecutionPrevention="2"
+ DataExecutionPrevention="0"
SupportUnloadOfDelayLoadedDLL="true"
TargetMachine="1"
/>

[sansnom05]

Are you sure?
Revision 1366 - Directory Listing
Modified Sat Nov 28 21:04:28 2009 UTC (12 months, 2 weeks ago) by Spec-Chum
Added /GS, /NXCOMPACT and /DYNAMICBASE compile options, where available

Patch by XhmikosR
This version changes the code in the DataExecutionPrevention = "1" to DataExecutionPrevention = "2" feedback I've caused the problem! Today, I put this back to 1, re-compiled under the latest version of mpc-hc, and DEP has not closed. mpc-hc work as usual. Of course the system into 2 XPSP3 have no problem. but VISTASP2 and WIN7 including WIN7SP1 the code DataExecutionPrevention = "2" does not change DataExecutionPrevention = "1" compilation of mpc-hc is easy to cause. But if you disable the two systems DEP, mpc-hc is little cause! I do not know the player turn DEP on the development of any help, at least I did not play with other DEP encountered problems, such as potplayr or Mplyer! Unstable recommended if the mpc-hc code can be DataExecutionPrevention = "2", but if it is stable in mpc-hc code is best DataExecutionPrevention = "1", because it affects the user's impression!
PS:There was consulted to do this under Why do not we change the user's advice?
Maybe back at that time in accordance with XhmikosR me, this is because the corresponding decoder is not compatible and caused isolated problems, if the decoder and separator is open source also easy to handle. Can be resolved by modifying the source code. But if not open source, or only separated by the decoder and decoder play. It is not without any solution can only disable the DEP?

You can disable DEP, that doesn't mean you should do it. Also enable DEP should be the normal, not the other way around, after all, DEP,ASLR and the like are there for a reason: security. If something need to disable DEP to run, that's theirs fault, not mpc-hc's.

But as you put it, this is open source project, you're free to do anything you feel like, as a user myself, I have no need to do that, so disable DEP is definitely not my advice. :)

[bozek]

Supply a sample file, and I might be able to add MJPEG-B support to ffdshow.

http://www.easy-share.com/1913324598/mjpeg_samples.rar

Those are the two samples that were linked from ticket #341. I do have some longer stuff as well if you need...

[v0lt]

The video in the sample with the 'dmb1' FourCC isn't recognized by the MP4 splitter, so I can't add that one yet.
You can use LAVFSplitter. It is not limited list of formats.
Also I compiled a test build (http://www.mediafire.com/?inedw07gc34m9a0).

[adam777]

Hello all,
I've just managed to set FFDShow to output 24 bit DTS audio track as expected (that is, using libdts to output 32 bit FP and not libavcodec that outputs 16 bit).
Now, I would like to take this 32 bit stream and play it with as little processing as possible on windows 7 x64.

My first attempt was using MPC Audio Renderer that uses WASAPI, but I was not able to connect to it when feeding 32 bit (or even 24 bit) stream.
My second attempt was using WASAPI with the help of ReClock audio renderer, again with no success.
My current approach is kind of cheating by setting the default format in the audio card properties to what I expect (24 bit / 48000 Hz).
I think the latter is probably able to achieve the needed goal, but isn't very flexible (for example, when playing 16 bit AC3 streams).

So my question is if there's anything else I can try to make WASAPI work (seems to be the most convenient option overall).
Also, when playing 16 bit AC3 which is decoded to 32 bit FP by liba52 and then output as 24 bit by DirectSound, am I losing anything along the way or the "actual data" is always 16 bits padded with 0's to 32 or 24 bits, so nothing gets lost?

Thanks in advance, Adam.

[panda-34]

Is there any way of forcing mpc to respect video crop settings (PixelCropXXX values) in mkv header with mpeg-2 video?

[clsid]

http://www.easy-share.com/1913324598/mjpeg_samples.rar

Those are the two samples that were linked from ticket #341. I do have some longer stuff as well if you need...
Those samples already play fine with ffdshow.

[ilkertezcan]

There is a aspect ratio issue in MKV Splitter. (Tested with 2499 and SVN 2781).

Source MKV video: http://rapidshare.com/files/438311485/MKV_Aspect_Ratio.rar

Original Resolution: 640x352 but wrong aspect ratio when playing. Strange because Virtualdub (with ffcHandler's MKV input-plugin) can open properly resolution. I saved to AVI (with direct stream copy). MPC-HC can play properly with aspect ratio (the AVI file).

[Aleksoid1978]

There is a aspect ratio issue in MKV Splitter. (Tested with 2499 and SVN 2781).

Source MKV video: http://rapidshare.com/files/438311485/MKV_Aspect_Ratio.rar

Original Resolution: 640x352 but wrong aspect ratio when playing. Strange because Virtualdub (with ffcHandler's MKV input-plugin) can open properly resolution. I saved to AVI (with direct stream copy). MPC-HC can play properly with aspect ratio (the AVI file).

All play fine:
AVI - http://s011.radikal.ru/i317/1012/a5/11b2f799db29t.jpg (http://radikal.ru/F/s011.radikal.ru/i317/1012/a5/11b2f799db29.png.html)

MKV - http://i003.radikal.ru/1012/4d/2c5d43805a0at.jpg (http://radikal.ru/F/i003.radikal.ru/1012/4d/2c5d43805a0a.png.html)

[tetsuo55]

Good news for ati users the 10-bit RGB input it's back with the new AMD Catalyst™ Display Driver 8.80 (http://forums.guru3d.com/showthread.php?t=334165) so no more black screen (with DXVA) or those "burn colors" (without DXVA) when enable "Force 10-bit RGB Input" for Surface A2R10G10B10 !

Now all that remains to do so we can get back the Full 10-bit RGB = "Force 10-bit RGB Input" for Surface A2R10G10B10 + "10-bit RGB Output" for Backbuffer A2R10G10B10 + "D3D Full screen Mode" for Display A2R10G10B10 is to revert or fix the rev 1826 so we can get Display A2R10G10B10 too


Edit: For those still getting those "burn colors" when enable "Force 10-bit RGB Input" remember not all filters are 10-bit RGB input "capable" e.g. MPC-HC's internal xvid/MPEG-4 filter (Windows 7's Mpeg4s Decoder DMO it's working !)Can you run some tests and tell me exactly which file, or even better which line needs to be reverted?

[ilkertezcan]

@Aleksoid1978: Thank you for interest. You're right, MPC-HC can play as 827x352 aspect ratio (MKV or AVI video file that I sent). This is wrong ratio of width. Example screenshots:

Virtualdub: http://i.imgur.com/JzgHx.png

GOM Player - MKV (mpc-hc MKV splitter is actived): http://i.imgur.com/NB8l9.jpg

GOM Player - AVI (mpc-hc AVI splitter is actived): http://i.imgur.com/ss6Om.jpg

Do I need to write a ticket topic for this issue?

[bozek]

Those samples already play fine with ffdshow.

Well, the thing is I dont use ffdshow, just MPC-HC...

[Mercury_22]

Can you run some tests and tell me exactly which file, or even better which line needs to be reverted?

Sorry but from a quick look at rev 1826 it will take more time for me than I have right now to pin-point the file / line to revert, but I'm sure there are others more familiar with the code for whom it will take much less time to do that

[janos666]

I made some screenshots to demonstrate the remaining problems with the CMS feature (ticket #862 addendum). I used GIMP 2.6.11 for comparison (as much as I know it uses lcms as well as MPC-HC).
Black crush: CM OFF (http://img12.tar.hu/janos666/img/96331770.png) ; Perceptual (http://img12.tar.hu/janos666/img/96331771.png) ; Relative (http://img12.tar.hu/janos666/img/96331772.png) ; Saturation (http://img12.tar.hu/janos666/img/96331773.png) ; Absolute (http://img12.tar.hu/janos666/img/96331769.png) ;;;;; MPC-HC vs GIMP - Absolute (http://img12.tar.hu/janos666/img/96331768.png)
As you can see, it's independent from the rendering intent mode, it's very noticeable and it doesn't exist in GIMP.
Something else: CM OFF (http://img12.tar.hu/janos666/img/96331764.png) ; Perceptual (http://img12.tar.hu/janos666/img/96331765.png) ; Relative (http://img12.tar.hu/janos666/img/96331766.png) ; Saturation (http://img12.tar.hu/janos666/img/96331767.png) ; Absolute (http://img12.tar.hu/janos666/img/96331763.png) ;;;;; MPC-HC vs GIMP - Absolute (http://img12.tar.hu/janos666/img/96331762.png)
I don't know what it is exactly but it's very odd (it looks like a problem with the near-white scale).
Here is my display profile if you want to play with it: CLICK (http://www.tar.hu/janos666/abstest.rar)

[adam777]

Hello all,
I've just managed to set FFDShow to output 24 bit DTS audio track as expected (that is, using libdts to output 32 bit FP and not libavcodec that outputs 16 bit).
Now, I would like to take this 32 bit stream and play it with as little processing as possible on windows 7 x64.

My first attempt was using MPC Audio Renderer that uses WASAPI, but I was not able to connect to it when feeding 32 bit (or even 24 bit) stream.
My second attempt was using WASAPI with the help of ReClock audio renderer, again with no success.
My current approach is kind of cheating by setting the default format in the audio card properties to what I expect (24 bit / 48000 Hz).
I think the latter is probably able to achieve the needed goal, but isn't very flexible (for example, when playing 16 bit AC3 streams).

So my question is if there's anything else I can try to make WASAPI work (seems to be the most convenient option overall).
Also, when playing 16 bit AC3 which is decoded to 32 bit FP by liba52 and then output as 24 bit by DirectSound, am I losing anything along the way or the "actual data" is always 16 bits padded with 0's to 32 or 24 bits, so nothing gets lost?

Thanks in advance, Adam.

OK, played some more and figured out the following.
- Using MPC audio renderer (configured to use WASAPI), I can not connect when outputting 24 bit, 32 bit or 32 bit FP from FFDShow, only 16 bit.

- Using ReClock audio renderer, outputting the 32 bit FP from FFDShow as is, and modifying the output format in ReClock will not connect to WASAPI with 32 bit FP, 32 bit or 24 bit. It will connect using 16 bit or 24 bit padded to 32.

Now, I believe I read somewhere that although MediaInfo reports the audio track as being 24 bit, it is most likely 16 bit, as are most soundtracks.
So, what path should I use?
- FFDShow (16 bit) -> MPC audio renderer (16 bit)
- FFDShow (32 bit FP) -> ReClock (24 bit padded to 32)
Both utilize WASAPI, so the OS stuff isn't relevant.

Adam.

PS.
The whole point is purely out of interest, I really do not expect any noticeable difference in audio quality :)

PSS.
How can I check the bit depth of a given file other than using mediainfo?

[Porsche_fan]

Having an issue which appears related to the use of DVD Navigator as the problem occurs in MPC and Windows Media Center but not with a Cyberlink player. The issue also doesn’t occur if I play a vob file using mpeg-2 splitter.

While playing a dvd, the disc drive will suddenly start spinning at high speed (8x+) for a couple of minutes and then revert to 1x. This is rather irritating as it usually occurs multiple times when watching a film.

It happens on every pressed dvd I have tried both single and dual layers. The timing of the increased drive speed is different for each dvd but if the dvd is replayed it occurs at approximately the same runtime.

I have read a number of people on the web having the same problem. My guess is it is buffering related. Oddly, I didn’t have this issue with the exact same model HP which used an nVidia card…my current computer uses an ATI.

Are there some programming changes that could be made to the way DVD Navigator works in MPC based on this info http://msdn.microsoft.com/en-us/library/dd388611%28VS.85%29.aspx

I know currently MPC only works with Microsoft’s DVD Navigator but is it possible to allow the use of Cyberlink’s DVD Navigator with MPC? An alternative navigator would probably eliminate the Macrovision failure error with Haali’s and madVR too.

[nevcairiel]

- Using ReClock audio renderer, outputting the 32 bit FP from FFDShow as is, and modifying the output format in ReClock will not connect to WASAPI with 32 bit FP, 32 bit or 24 bit. It will connect using 16 bit or 24 bit padded to 32.

24 bit padded to 32 is the only really reliable mode, short of 16 bit, which sacrifices quality. Consumer sound cards only work with up to 24 bits of precision (i'm not even sure if they use it fully), even professional audio gear does not typically use the full 32-bits.

If FFDShow can output 32fp natively from the decoder in question, then stick with that. Its desireable to change the format as few times as possible, so try to stick to whatever ffdshow outputs (enable all output formats, it should select the best matching), and let ReClock resample it, and output it as 24-padded-to-32 to the sound card.

[Kazuya]

Hum... interesting indeed.

About this 32bits output ffdshow, I found something strange yesterday.
I was comparing Waves from AC3filter and ffdshow (with graphedit as wave writer), and this is what I get for libavcodec in AC3 and DTS, output 32 bits :

http://img130.imageshack.us/img130/5037/libav.png

then I tried with liba52 :

http://img101.imageshack.us/img101/7193/liba52.png


Is there an explanation ?

The mixer was in 2/0/0 with only L and R channels in custom matrix for both.

[pankov]

Kazuya,
libavcodec in ffdshow is limited to 16bit output.
That's the reason why it's recommended to not use it for HD audio sources and prefer libdts or liba52.

[Kazuya]

And this is what I get with libavcodec output 24bits :

http://img690.imageshack.us/img690/1901/libav24bits.png

No problem with libdts or liba52, we really get the 24bits...

[Kazuya]

Kazuya,
libavcodec in ffdshow is limited to 16bit output.
That's the reason why it's recommended to not use it for HD audio sources and prefer libdts or liba52.

Ooooooooh ok !!!
I didn't know ! :p
:thanks:

[Kazuya]

Other thing. I updated my MPC with 1.4.2788 (and 2748 before), and I get some macroblocking with internal x264 decoder.
Look :

http://img638.imageshack.us/img638/4106/2788w.th.png (http://img638.imageshack.us/i/2788w.png/)

And I tried with 2499 and 1845, no problem :

http://img269.imageshack.us/img269/4099/2499andbelow.th.png (http://img269.imageshack.us/i/2499andbelow.png/)

I can make a sample if needed.
Not a real issue for me, I use coreAVC, but for some maybe ?

[adam777]

24 bit padded to 32 is the only really reliable mode, short of 16 bit, which sacrifices quality. Consumer sound cards only work with up to 24 bits of precision (i'm not even sure if they use it fully), even professional audio gear does not typically use the full 32-bits.

If FFDShow can output 32fp natively from the decoder in question, then stick with that. Its desireable to change the format as few times as possible, so try to stick to whatever ffdshow outputs (enable all output formats, it should select the best matching), and let ReClock resample it, and output it as 24-padded-to-32 to the sound card.

:thanks:

[CruNcher]

Kazuya,
libavcodec in ffdshow is limited to 16bit output.
That's the reason why it's recommended to not use it for HD audio sources and prefer libdts or liba52.

And i was surprised that it takes so less cpu cycles compared to the other decoders is that the case for every libavcodec audio decoder mp1/2/3,aac,ac3,eac3,vorbis ? and why does it show 32bit integer in the info display @ all then ? if its only rendering 16 bit

[pankov]

If I remember correctly madshi was talking about this issue either here or in the ffdshow thread and said that libavcodec is decoding the stream correctly and molesting it only at the end when it downconverts to 16bit. He made a patch which skips the this conversion and uses it in eac3to.
I could be totally wrong here so anybody with more knowledge, please, correct me.
If I'm not mistaken it was tetsuo55 that also participated in the discussion.

[hoborg]

Problem report:

"MPC - Avi Splitter" doesn't like this *.AVI sample (http://rapidshare.com/files/438687210/Samsam_-_1x38_-_L_ombre_qui_en_dit_trop.avi) AVI is splitted, graph created correctly, but there is only black screen undepended on decoder - tested with FFDshow and MPC video decoder.
Sample working fine if default OS AVI splitter is in use.

I think the problem is already in TRACK (https://sourceforge.net/apps/trac/mpc-hc/ticket/19).