ffdshow tryouts project: Discussion & Development

[Inventive Software]

Did you also enable "Queueing", so that both cores are utilised (one decodes / renders, the other queues the frames)?

[Keepitsimple]

I dont know what neither "Inloop Deblocking" nor "Queueing" is.

I got 20% less utilization by not forcing rgb as output. However without RGB my movies hang when I move them between screens. :(

[_xxl]

I got 20% less utilization by not forcing rgb as output. However without RGB my movies hang when I move them between screens.
Can you provide a sample?

[_xxl]

Is anybody working on enhancements to the x264 decoder? I have to overclock my core2duo to 3Ghz to play 1080p with ffdshow/mpc right now. The problem is mostly that only one core is utilized.
No multithreaded support for h264 libavcodec decoder.

[foxyshadis]

The second-screen stuttering is an ATI driver problem, it can't deal with splitting a video across two screens. (And has similar problems with GDI windows being placed on top of a VMR9 surface - notably a right-click menu or codec config window.) The 20% overhead for RGB sounds pretty high though.

[Keepitsimple]

it was pretty much every movie. seems to have vanished with haalis new renderer. however it stilll takes a lot longer to switch screens. with rgb its not even a black picture if you move it fast. with nv12 it sort of stops at the edge for 1 second before it moves to the next screen.

[fastplayer]

The volume levels are updated very slowly in the latest 969 build. In previous builds they were updated "real-time"-like. Maybe this has something to do with bug fix rev963...

[LoRd_MuldeR]

The volume levels are updated very slowly in the latest 969 build. In previous builds they were updated "real-time"-like. Maybe this has something to do with bug fix rev963...

I can confirm that with r971 build. Also effects the "Info & Debug" dialog.
You can notice that when you play VBR audio and watch "Current input bitrate"

Maybe those values are just updated less often to save some CPU cycles?

[fastplayer]

AFAIK the bug was only in the audio mixer section. At least I couldn't reproduce it in the volume section...

Edit: No, it wasn't the mixer section. It was just enough to have the audio dialog opened for CPU usage to increase.

[haruhiko_yamagata]

The volume levels are updated very slowly in the latest 969 build. In previous builds they were updated "real-time"-like. Maybe this has something to do with bug fix rev963...
Thank you. I can confirm. The dialog is updated every 500ms, because "Mixer" is still heavy at 100ms. The code judging by CPU monitor may help.

[madshi]

One question to you ffdshow tryouts gurus:

How can I make "WVC1 -> wmv9" work? If I activate "WVC1 -> libavcodec" I can successfully decode VC-1 EVO files (with the latest Haali Media Splitter). But as soon as I choose "WVC1 -> wmv9" the Haali Media Splitter and ffdshow tryouts can't connect, anymore.

FWIW, I have Microsoft's WVC1 decoder (wvc1dmod.dll) downloaded and registered.

Thank you guys!

[KoD]

Does WVC1 -> disabled work ?

[madshi]

Does WVC1 -> disabled work ?
If I set it to "disabled" or "wmv9", Haali Media Splitter and ffdshow cannot connect. There's no difference between "disabled" and "wmv9". Haali Media Splitter and ffdshow tryouts can connect only if I set it to "WVC1" to "libavcodec".

[_xxl]

Works for me.Please install WMP11.
http://img150.imagevenue.com/loc21/th_89542_Capture_122_21lo.jpg (http://img150.imagevenue.com/img.php?image=89542_Capture_122_21lo.jpg)

[madshi]

Thank you - after install WMP11 it works! It's not really fluid, though, although I have a quite fast PC (Merom Dual-Core). Too bad... :scared:

Edit: After muxing to MKV things are fluid - yey!!

[Delerue]

I found a strange thing: if a subtitle open a italic tag ('<i>') but doesn't close it in the same line (dialog), so all subtitles after that remain italic until you write a close italic tag ('</i>'). This thing doesn't happen with VobSub, DirectVobSub and Subtitle Workshop.

EDIT: Forgot to say that the build tested was 971.

[jffulcrum]

Found bug in Resize&Aspect filter.
When i select 2xSai, 2xSuperSai or hq2x methods and leave in "New size" filed vertical size as 0 (automatic), resize actually not performed.
FFDShow-tryout build 967 (also checked 927), MPC latest build by celtic_druid (with BSPlayer also wrong).

[diizzy]

Hi,
First of all I have to admit that I've been a "silent" user for over a year so I'm not really a new member. I also would like the developers of ffdshow/ffdshow-tryouts for their work and time spent on the project, its really appreciated.

Anyhow I have a little bug report/feature request.

Issue: "RAW" interlaced/deinterlaced video doesnt get detected as interlaced using scalers resulting in improper scaling.

How to reproduce: Play a deinterlaced source using an external filter and upscale.

Fix: Instead of letting ffdshow decide for itself (having the Interlaced box gray) make sure to check it.

Software used:

ffdshow_rev906_20070210_clsid.exe
YV12 input and YV12 output
Lanczos used as scaler algorithm

Dscaler5.IVTC.251206.zip
Can be found here
http://www.avsforum.com/avs-vb/showthread.php?p=9266528&&#post9266528

MPC - mplayerc.rev611-3.2kxp.7z
http://ffdshow.faireal.net/
VMR7 (windowed) and WMR9 mixer mode disabled as output

So the flow is the following: Splitter --> Dscaler5 (decoding) --> ffdshow (scaling) --> MPC

I'm also wondering if its possible to include this modified/optimized MPEG audio decoder into ffdshow.
http://otachan.com/in_!mpg123.html

I would love to contribute but my programming skills arent that great yet :-/

//Danne

[mariner]

Hi drevil_xxl,

"Media Player 2 (MPC) + Haali Media Splitter + ffdshow ac3 decoder."

Greetings and thanks for the ffdshow decoders., but was not able to get it to decode DD+. It won't connect.

Thanks for you kind assistance.

[foxyshadis]

Found bug in Resize&Aspect filter.
When i select 2xSai, 2xSuperSai or hq2x methods and leave in "New size" filed vertical size as 0 (automatic), resize actually not performed.
FFDShow-tryout build 967 (also checked 927), MPC latest build by celtic_druid (with BSPlayer also wrong).
How do you leave anything in new size? The entire left side of the resize panel should be greyed out when you choose one of those three methods, because it's always set to 2x. If this isn't actually the "resize if" section, you might have found some GUI bug.

Oh yeah, almost forgot, you have to reload the video after setting it to one of the game resizers or back. They aren't on-the-fly capable at all.

[haruhiko_yamagata]

Found bug in Resize&Aspect filter.
When i select 2xSai, 2xSuperSai or hq2x methods and leave in "New size" filed vertical size as 0 (automatic), resize actually not performed.
FFDShow-tryout build 967 (also checked 927), MPC latest build by celtic_druid (with BSPlayer also wrong).
What is "2xSai, 2xSuperSai or hq2x methods" ?

[haruhiko_yamagata]

Issue: "RAW" interlaced/deinterlaced video doesnt get detected as interlaced using scalers resulting in improper scaling.
Hi,
Does checking "Resize & aspect->Settings->Interlaced" fix your problem?
If your report means ffdshow does not do it automatically, I can reproduce your issue.

[kikker]

Hi, I have the latest 971 build, but I can't seem to switch my presets. I have 4 besides the "default", which is the only one in bold type (active preset?). Ffdshow remembers all my settings and are entered in the registry, but the "active preset" reg entry never changes from "default", so none of the other presets can be applied.

I am running Vista and the ffdshow preset settings are at HKCU\software\GNU. I am also running as a limited user, but this shouldn't matter as all the preset settings are entered just fine.
Ffdshow was installed as an admin, but when I try to run it now as an admin, I get an error "ffdshow.ax module couldn't be found."

Why can't that reg entry "active preset" be changed?

[JarrettH]

aren't those resize modes used in zSNES?

[Syris2k4]

They are methods for smoothing resized images, so yes... things like zSNES, Exult and Pentagram have them to upscale the wonderfully pixilated early 90's bitmaps :D

@haruhiko_yamagata the methods he refers to are found in :
Resize and aspect -> Settings -> Luma Method
Check the drop down list, they are near the bottom... if you didnt find them :)

[HeadBangeR77]

@ kikker
In case it could be of any use, double-clicking on a chosen preset (I've got more than ten of them) makes it bold, and it's loaded as the active one after re-opening ffdshow /player. One-click just changes the preset sort of temporarily, until you close ffdshow /player - should work for you too. Generally speaking, no problems here with rev 971, but I'm running WinXP Pro SP2.

[diizzy]

@ haruhiko_yamagata
Yes, checking interlaced does help.
Is it possible to make this option work automatically?
//Danne

[kikker]

@ kikker
In case it could be of any use, double-clicking on a chosen preset (I've got more than ten of them) makes it bold, and it's loaded as the active one after re-opening ffdshow /player. One-click just changes the preset sort of temporarily, until you close ffdshow /player - should work for you too. Generally speaking, no problems here with rev 971, but I'm running WinXP Pro SP2.

:) I don't know if I hate it or love it that such answers can be so simple. Who would have thought to *double*-click on it?

[HeadBangeR77]

:) I don't know if I hate it or love it that such answers can be so simple. Who would have thought to *double*-click on it?
Wish all the answers concerning this forum were so simple. ;) :D
Does one-click work for you also, in the way I described it above?

[zambelli]

Would be possible to integrate some sort of standard performance test into the build compile process? Commonly used functions such as XviD decoding could be benchmarked, for example. The results might not mean much on their own, but if perf results were available for every new tryout build, then it'd be possible to track regressions in performance and answer the often asked question "Is this build faster than xyz build?"

[Keepitsimple]

Levels in ffdshow

As I have understood it:

The histogram has intensity on the y axis and grayscale on the x-axis.

Posterize determines how many levels the grayscale will have.

Now some questions:

Fullrange makes the picture gets darker. The histogram also compresses to 16-235 levels instead of 0-255. But what is fullrange?

What do the different modes and modify only luminance mean?

How do I know what levels a particular media file uses? Ie should I look at the histogram with fullrange enabled or not?

[foxyshadis]

@ haruhiko_yamagata
Yes, checking interlaced does help.
Is it possible to make this option work automatically?
//Danne

If the solution was as simple as the problem, it would have been done a very long time ago. Interlaced detection is a very hard problem, because there's no good realtime method without a false positive/negative rate of at least 5%.

[foxyshadis]

Levels in ffdshow

As I have understood it:

The histogram has intensity on the y axis and grayscale on the x-axis.
Well, the relative count of pixels having that intensity.

Fullrange makes the picture gets darker. The histogram also compresses to 16-235 levels instead of 0-255. But what is fullrange?

What do the different modes and modify only luminance mean?

How do I know what levels a particular media file uses? Ie should I look at the histogram with fullrange enabled or not?

Fullrange is 0-255, Video range is 16-240. Most video isn't fullrange, in fact it's quite rare to see them unless you create them yourself. However, because of the uselessness of different video drivers, expansion to fullrange is sometimes needed. By checking fullrange you get the real histogram and can expand it to the full levels. Otherwise you get one without the (often empty) extremes, the input will be clamped to (16,240) and the output scaled to (16,240), removing any BWB or WTW content.

It's the output scaling that makes the difference in brightness when you change the fullrange state, if you don't change the output sliders yourself.

[haruhiko_yamagata]

Found bug in Resize&Aspect filter.
When i select 2xSai, 2xSuperSai or hq2x methods and leave in "New size" filed vertical size as 0 (automatic), resize actually not performed.
FFDShow-tryout build 967 (also checked 927), MPC latest build by celtic_druid (with BSPlayer also wrong).
I understood, but I can't reproduce this. It works with vertical size 0 (automatic).

[haruhiko_yamagata]

@ haruhiko_yamagata
Yes, checking interlaced does help.
Is it possible to make this option work automatically?
//Danne
Checking "Interlaced" intermidiate state (half checked) help automatic detection. It works correctly as long as the inputing information is correct.
ffdshow can't detect weave or bob. So it's hard to detect automatically in DVD.

What the default setting should be?
Perhaps ffdshow resizer is often used for DVD upscaling, and DVD is weave in most cases.
If it is bob (truely interlaced), it is recommended to use software de-interlacer before resizing.
So I think the default setting is better to be off (as it is now).

[tekNerd]

ffdshow 971, icl 9

Even if Resize&Aspect is not checked I still can override the Aspect ratio. Is this normal?

About the Audio, delay filter: Is it possible to have something more like MPC's "Audio time shift"? I am not using it that much [actually almost zero], but...just asking.

Anywho, thanks to all for this beautiful piece of software and for all the help that I find in this great forum.

[kikker]

Wish all the answers concerning this forum were so simple. ;) :D
Does one-click work for you also, in the way I described it above?

Yes, it works like you describe.

[cc979]

Would be possible to integrate some sort of standard performance test into the build compile process? Commonly used functions such as XviD decoding could be benchmarked, for example. The results might not mean much on their own, but if perf results were available for every new tryout build, then it'd be possible to track regressions in performance and answer the often asked question "Is this build faster than xyz build?"

do you mean to benchmark a compile after each different compiler and test to see what each module/or setting work best ?

i did benchmark most builds earlier but not that much of late

ffdshow_rev971_20070227.exe -o3 pentium3
null: 9s, kernel: 0s, total: 9s, real: 9s, fps: 71.8, dfps: 70.6
vmr9: 9s, kernel: 2s, total: 12s, real: 13s, fps: 53.9, dfps: 49.0

ffdshow-913-sse1-gcc4.1.2
vmr9: 8s, kernel: 1s, total: 10s, real: 12s, fps: 60.8, dfps: 52.5

ffdshow-20060929-rev655-cc979-sse.exe
null: User: 8s, kernel: 0s, total: 8s, real: 10s, fps: 74.5, dfps: 63.0
vmr9: User: 9s, kernel: 1s, total: 11s, real: 12s, fps: 58.3, dfps: 51.0

cc979 rev584 -o3
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 81.3, dfps: 74.7
vmr9: User: 7s, kernel: 0s, total: 8s, real: 10s, fps: 81.8, dfps: 61.1
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 75.8, dfps: 72.7

clsid
ffdshow_rev584_20061123_clsid.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 75.1, dfps: 72.9
vmr9: User: 8s, kernel: 0s, total: 8s, real: 11s, fps: 75.8, dfps: 58.0
ovrl: User: 9s, kernel: 0s, total: 9s, real: 9s, fps: 71.4, dfps: 69.4

clsid
ffdshow_rev494_20061103_clsid.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 75.1, dfps: 74.0
vmr9: User: 8s, kernel: 0s, total: 8s, real: 11s, fps: 77.4, dfps: 58.3
ovrl: User: 8s, kernel: 0s, total: 9s, real: 9s, fps: 72.4, dfps: 69.4

clsid
ffdshow_rev348_20061008_clsid.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 77.7, dfps: 75.8
vmr9: User: 8s, kernel: 0s, total: 8s, real: 10s, fps: 79.0, dfps: 61.1
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 73.6, dfps: 71.3

clsid
ffdshow_rev187_20060915_clsid.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 76.4, dfps: 73.7
vmr9: User: 8s, kernel: 0s, total: 8s, real: 10s, fps: 77.8, dfps: 59.8
ovrl: User: 8s, kernel: 0s, total: 9s, real: 9s, fps: 72.4, dfps: 70.3

clsid
ffdshow_rev187_20060915_clsid_generic.exe
null: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 74.5, dfps: 70.1
vmr9: User: 8s, kernel: 0s, total: 8s, real: 11s, fps: 75.0, dfps: 58.1
ovrl: User: 8s, kernel: 0s, total: 9s, real: 9s, fps: 72.7, dfps: 70.1

clsid
ffdshow_rev181_20060913_clsid.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 75.7, dfps: 74.6
vmr9: User: 8s, kernel: 0s, total: 8s, real: 11s, fps: 75.4, dfps: 59.0
ovrl: User: 9s, kernel: 0s, total: 9s, real: 9s, fps: 70.2, dfps: 68.7

videomixer9
ffdshow-tryouts-rev155.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 81.6, dfps: 80.7
vmr9: User: 7s, kernel: 1s, total: 9s, real: 10s, fps: 72.0, dfps: 61.2
User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 78.2, dfps: 75.0

drevil_xxl
ffdshow-tryouts-rev155-AMD-XP-K8.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 79.6, dfps: 78.7
vmr9: User: 8s, kernel: 1s, total: 9s, real: 10s, fps: 68.3, dfps: 59.8
ovrl: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 76.8, dfps: 73.0

videomixer9
ffdshow-tryouts-rev148.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 81.4, dfps: 79.0
vmr9: User: 8s, kernel: 0s, total: 9s, real: 10s, fps: 68.4, dfps: 60.8
ovrl: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 77.2, dfps: 74.6

drevil_xxl
ffdshow-tryouts-rev145-amd-xp.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 79.3, dfps: 76.5
vmr9: User: 8s, kernel: 0s, total: 9s, real: 10s, fps: 67.1, dfps: 59.5
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 74.9, dfps: 72.2
...
ffdshow-tryouts-rev145-k8.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 78.5, dfps: 75.7
vmr9: User: 8s, kernel: 1s, total: 9s, real: 10s, fps: 66.2, dfps: 59.9
ovrl: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 75.3, dfps: 72.8
...
ffdshow-tryouts-rev145-p4.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 79.7, dfps: 76.8
vmr9: User: 8s, kernel: 1s, total: 9s, real: 10s, fps: 70.2, dfps: 59.9
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 74.5, dfps: 72.7

clsid
ffdshow_rev142_20060906.exe
null: User: 8s, kernel: 0s, total: 8s, real: 8s, fps: 78.8, dfps: 77.7
vmr9: User: 8s, kernel: 1s, total: 9s, real: 10s, fps: 65.5, dfps: 59.8
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 75.1, dfps: 72.5

drevil_xxl
ffdshow-tryouts-rev138-sse.exe
null: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 78.7, dfps: 70.7
vmr9: User: 8s, kernel: 0s, total: 9s, real: 10s, fps: 69.4, dfps: 59.7
ovrl: User: 8s, kernel: 0s, total: 8s, real: 9s, fps: 75.3, dfps: 72.0

ffdshow-tryouts-rev138-sse2.exe
null: User: 9s, kernel: 0s, total: 10s, real: 10s, fps: 65.4, dfps: 64.7
vmr9: User: 10s, kernel: 0s, total: 11s, real: 12s, fps: 58.3, dfps: 51.3
ovrl: User: 10s, kernel: 0s, total: 10s, real: 10s, fps: 63.1, dfps: 61.2

there has been changes up/down in different revs

[_xxl]

Decoding speed is provided by libavcodec.
Libavcodec compiled with -msse/sse2 is slower.
Please test generic builds only.

[hdboy]

Using Haali's latest splitter and celtic_druid's latest MPC build and ffdshow tryouts 3/01/07 nightly build.

I get video playing an EVO file, but no audio. Looking at haali's pin info, the audio subtype is

{0000EAC3-0000-0010-8000-00AA00389B71}

which must be E-AC3. I've set ffdshow audio decoder to "prefer" in MPC, and even added the subtype above to the list for ffdshow audio decoder under MEDIATYPE_Audio, but no sound and ffdshow audio is not being invoked. MPC's filter menu shows

Video Renderer
Overlay Mixer
WMVVideo Decoder DMO (WMP11 installed)
Haali Media Splitter (AR)
filename.evo

I've also tried to use Cyberlink audio decoder PDVD6.x, PDVD7, PDVD7 UPNP. No go.

Powerdvd gives me incompatible graphic device (I have a 7600GT non HDCP card and non-HDCP monitor). I'm close with MPC. suggestions? thanks.

[fastplayer]

ffdshow doesn't support E-AC3.

[zambelli]

do you mean to benchmark a compile after each different compiler and test to see what each module/or setting work best ?
Sort of. It doesn't necessarily need to be a comparison of different compilers; it could be just a way to track performance over time too.

One could pick most commonly used components - i.e. libavcodec decoder, resizer, deinterlacer, sharpener - and then benchmark how long it takes to decode, resize, or whatnot a standard piece of test content.

If this was automatically run post-build and results posted with the downloads, then users would have some sort of idea what to expect from particular builds in terms of performance. It would also help the devs spot performance bugs when they creep up due to other checkins.

i did benchmark most builds earlier but not that much of late
That's pretty interesting. What did you measure exactly and how did you measure it?

[cc979]

That's pretty interesting. What did you measure exactly and how did you measure it?

i use haali splitter and his time-codec - with a hd-xvid - using no resize, no filters or anything - default ffdshow settings - and it tells you the fps for null/overlay mixer/vmr9

[_xxl]

Does Haali Media Splitter supports E-AC3?
ATSC A/52b is an extension to A/52a and was approved for the Bluray and HD-DVD standard. It is the same as Enhanced AC3 or EAC3 for short. EAC3 is also supposed to be backwards compatible with AC3.
EAC3 is backwards compatible with AC3?

[cc979]

Does Haali Media Splitter supports E-AC3?

EAC3 is backwards compatible with AC3?

it does now i think, but you need an EAC3 filter eg. windvd's

[_xxl]

http://img.photobucket.com/albums/0903/pakthedude/Screenshots/greenlines.jpg
http://sourceforge.net/tracker/index.php?func=detail&aid=1669045&group_id=53761&atid=471489
Tested:
directvobsub 2.37 + ffdshow-rev899_20070209_gcc_xxl +WMP

[haruhiko_yamagata]

http://sourceforge.net/tracker/index.php?func=detail&aid=1669045&group_id=53761&atid=471489
Tested:
directvobsub 2.37 + ffdshow-rev899_20070209_gcc_xxl +WMP
What is the holizontal size?
Does unchecking YV12 help?

[nurbs]

Does Haali Media Splitter supports E-AC3?

EAC3 is backwards compatible with AC3?

It is not backwards compatible per se, but hardware that supports E-AC3 has to be able to convert it to AC3.

[Peuj]

Hi,

I have 2 questions about the ffdshow WMV codecs:

in the ffdshow video there are:
WMV1/7
WMV2/8
WMV3/9

and in the ffdshow audio there are:
WMA 7
WMA 8/9

Does the WMA 7 and WMA8/9 is them than WMA 1 and WMA 2/3?
If yes then you should update the names (as it is with the video) with something like:
WMA 1/7
WMA 2/3 8/9

If I set libavcodec for WMA8/9 and a play a movie with an audio stream WMA2, then the ffdshow codec is used.
But if I play a movie with an audio stream WMA3 the ffdshow codec is not used.

So maybe it's normal but then the information should be updated too and separate WMA 2/8 and WMA 3/9.

Last question, are there some known interests to use libavcodec for VMW/A. I just use them because I wanted to try the "set all stable formats to libavcodec" but i don't see better performance or quality.

Thanks

[wyrd]

Hi,authors
Thank you for your efforts.

I can't playback(weird color/block noise?) this sample in wmv3(libavcodec).(ff_wmv9 is fine)
http://tirnanog.fate.jp/tmp/sample/wmv3/FullMetal%5bwmv+AAC%5d_remux-006.mkv
As for this sample works fine with libavcodec/ff_wmv9 either.
http://tirnanog.fate.jp/tmp/sample/wmv3/Bebop%5bwmv+AC3%5d_remux-004.mkv
I have tested with mpc/wmp10 and haali/gabest splitter.
Can anyone reproduce?

Regards.