ffdshow tryouts project: Discussion & Development

[Mercury_22]

ffdshow (libmpeg2) rev 2661 & 2664 it's crashing when seeking !
Version rev 2653 it's OK !

[@Tron@]

Latest build that I've seen is 2661.

http://ffdshow-tryout.svn.sourceforge.net/viewvc/ffdshow-tryout?view=rev&revision=2665 Then its not ffdshow update ( or path ) ???

[_xxl]

Not every revision gets to be build. I compile ffdshow when time is available.

[clsid]

ffdshow (libmpeg2) rev 2661 & 2664 it's crashing when seeking !
Version rev 2653 it's OK !
Either the problem occurs randomly, or it is a difference between builds. Nothing has changed in the code in those revisions that can cause a crash.
Do the 'broken' builds work if you copy libmpeg2_ff.dll from the 'working' build?

[_xxl]

I think that those crashes come from 2660. I've done a test build with 2659 and no crashes. I should test more, maybe I'm wrong.

[@Tron@]

Not every revision gets to be build. I compile ffdshow when time is available.

And how do you do it ( compile ffdshow ) ???

What programs need ??

[Mercury_22]

Either the problem occurs randomly, or it is a difference between builds. Nothing has changed in the code in those revisions that can cause a crash.
Do the 'broken' builds work if you copy libmpeg2_ff.dll from the 'working' build?

I don't have that mpeg2 file any more but now I get a crash with ffdshow.ax Problem Event Name: BEX
Application Name: mplayerc.exe
Application Version: 1.2.988.0
Application Timestamp: 49861768
Fault Module Name: ffdshow.ax
Fault Module Version: 1.0.5.2462
Fault Module Timestamp: 498dc3c5
Exception Offset: 001f56f6
Exception Code: c0000417
Exception Data: 00000000
OS Version: 6.0.6002.2.2.0.256.1
Locale ID: 1033
Additional Information 1: 7ec4
Additional Information 2: 165a3f55fb11351ed7df118f8a20d5b4
Additional Information 3: 177c
Additional Information 4: ecc854e453e687b4d733365bb227284d
with several other mpeg2 files when seeking or just when I start playing the files
Again ffdshow_rev2653_20090202_sse_icl10 it's last working rev i can get my hands on !

P.S. copying libmpeg2_ff.dll from the 'working' build it's NOT helping but as i've said now it's ffdshow.ax crashing so... If you need I'll do more tests tomorrow . Let me know !
AVC files are OK with the 'broken' (ALL) builds


EDIT: Found the problem ! It's the combination between MPC-HC's MPEG splitter and rev 2661, 2664, 2666 because when I'm using other splitter or rev 2653 everything it's OK ! no more crashing !

[alexins]

I don't have that mpeg2 file any more but now I get a crash with ffdshow.ax with several other mpeg2 files when seeking or just when I start playing the files
Again ffdshow_rev2653_20090202_sse_icl10 it's last working rev i can get my hands on !

P.S. copying libmpeg2_ff.dll from the 'working' build it's NOT helping but as i've said now it's ffdshow.ax crashing so... If you need I'll do more tests tomorrow . Let me know !
AVC files are OK with the 'broken' (ALL) builds


EDIT: Found the problem ! It's the combination between MPC-HC's MPEG splitter and rev 2661, 2664, 2666 because when I'm using other splitter or FFD rev 2653 everything it's OK ! no more crashing !

This problem appeared, starting with the audit of 2660, 2659 last working version.

[Mercury_22]

This problem appeared, starting with the audit of 2660, 2659 last working version.

It's NOT manifesting in VOB file just in ts and m2ts so far
More testing with other type of files later !

@ alexins can you PLEASE compile rev 2659 and 2660 for testing ?:helpful:

EDIT: AVC in ts or m2ts files crashing too (previously tested only AVC in MKV) !! So after more testing I've discover that it's the AUDIO which it's crashing with the "new" rev and MPC-HC's mpeg splitter because when i'm using MPC-HC's internal audio filters and FFD (all rev) video (avc, mpeg.....) everything it's OK with the same files which previously were crashing on me !!!!!!!!

EDIT 2 : Vob files are crashing too ! but after more "intensing" seeking

[alexins]

@ alexins can you PLEASE compile rev 2659 and 2660 for testing ?:helpful:


2659, 2660, 2666 x86/x64 (http://www.xvidvideo.ru/clsid/for_Mercury_22.zip)

[STaRGaZeR]

clsid, have you take a look at the several decoders that revert back to the closest option if the current one is not in the installer's list? Example: WMV3, wmv9 selected as decoder, after a new install it reverts to libavcodec. It's really annoying, I have to manually reconfigure it each time I install a new build to test.

Haruhiko, how do you bench exactly? We need to know it if we want accurate results.

[Mercury_22]

2659, 2660, 2666 x86/x64 (http://www.xvidvideo.ru/clsid/for_Mercury_22.zip)

Thanks :thanks: :D

I can confirm that the problem it's starting with Revision 2660 - Directory Listing
Modified Sat Feb 7 08:55:27 2009 UTC (40 hours, 42 minutes ago) by h_yamagata
Use MSVC's STL instead of uClibc++. and it has something to do with ffdshow's audio component and MPC-HC's MPEG PS/TS/PVA splitter :confused::helpful:

[madshi]

Guys who have Core2 Duo or Pentium D, could you tell me if it is faster with "High quality ..." checked than unchecked?
Is there an easy and reliable way to get proper test results? Just playing a USA 1080i60 broadcast with "DScaler with IVTC mod -> ffdshow raw video processor" and looking at the task manager's CPU consumption graphs it seems to me that "High quality" is still slower by maybe 10% on my Core 2 Duo (Merom), but that's a very rough estimate...

[haruhiko_yamagata]

Is there an easy and reliable way to get proper test results? Just playing a USA 1080i60 broadcast with "DScaler with IVTC mod -> ffdshow raw video processor" and looking at the task manager's CPU consumption graphs it seems to me that "High quality" is still slower by maybe 10% on my Core 2 Duo (Merom), but that's a very rough estimate...
You already have one ;)
C:\Program Files\Haali\MatroskaSplitter\timeCodec.exe
dfps is the result that I need.

[yesgrey]

haruhiko,
Here are my results with Intel E2160@3.0GHz (Core2 duo) XPSP3:
Test Clip 1 - xvid 592x256
ffdshow 2653 enabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 824.0, dfps: 144.3

ffdshow 2653 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 824.0, dfps: 148.7

Test Clip 2 - h264 1920x816
ffdshow 2653 enabled High Quality YV12 to RGB conversion:
User: 1s, kernel: 0s, total: 1s, real: 4s, fps: 116.5, dfps: 41.7

ffdshow 2653 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 1s, real: 4s, fps: 165.7, dfps: 46.3

Test Clip 1 - xvid 592x256
ffdshow 2666 enabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 988.8, dfps: 145.4

ffdshow 2666 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 941.7, dfps: 147.6

Test Clip 2 - h264 1920x816
ffdshow 2666 enabled High Quality YV12 to RGB conversion:
User: 1s, kernel: 0s, total: 1s, real: 4s, fps: 115.4, dfps: 42.8

ffdshow 2666 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 1s, real: 4s, fps: 159.6, dfps: 47.9

[haruhiko_yamagata]

I can confirm that the problem it's starting with and it has something to do with ffdshow's audio component and MPC-HC's MPEG PS/TS/PVA splitter :confused::helpful:
I'm not sure if I reproduced your problem.
This includes some bug fixes. (Patch (http://ffdshow-tryout.sourceforge.net/doom9/multithreading_of_ffdshow_converters20090209.patch), Build (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2668_20090209_Q.exe))
Does this fix the problem?

[Mercury_22]

I'm not sure if I reproduced your problem.
This includes some bug fixes. (Patch (http://ffdshow-tryout.sourceforge.net/doom9/multithreading_of_ffdshow_converters20090209.patch), Build (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2668_20090209_Q.exe))
Does this fix the problem?

Sorry but it doesn't fix the problem (NOT just mine !)
Problem signature:
Problem Event Name: BEX
Application Name: mplayerc.exe
Application Version: 1.2.988.0
Application Timestamp: 49861768
Fault Module Name: ffdshow.ax
Fault Module Version: 1.0.5.2402
Fault Module Timestamp: 4990199a
Exception Offset: 0020202b
Exception Code: c0000417
Exception Data: 00000000
OS Version: 6.0.6002.2.2.0.256.1
Locale ID: 1033
Additional Information 1: 7ec4
Additional Information 2: 165a3f55fb11351ed7df118f8a20d5b4
Additional Information 3: 08a2
Additional Information 4: 992dfccb1af028196e2c9b335aa3f011
Please try to play any ts or m2ts using MPC-HC's internal mpeg splitter and ffd's audio (any ac3 or dts or.....) and then try seeking :( :helpful:

EDIT: there is some improvement because the files that were crashing from the start now are crashing only when I'm seeking

[haruhiko_yamagata]

haruhiko,
Here are my results with Intel E2160@3.0GHz (Core2 duo) XPSP3:

Almost!
Is your E2160 heavily overclocked Pentium DC?
Maybe different on 45nm Core2 Duo.

[haruhiko_yamagata]

Sorry but it doesn't fix the problem (NOT just mine !)

Please try to play any ts or m2ts using MPC-HC's internal mpeg splitter and ffd's audio (any ac3 or dts or.....) and then try seeking :( :helpful:

EDIT: there is some improvement because the files that were crashing from the begining now are crashing only when I'm seekingThanks, I found a sample that reproduced the problem.

[Mercury_22]

Thanks, I found a sample that reproduced the problem.

Ok :thanks: Let me know if you need more testing

[madshi]

build 2664, XPSP2, Core 2 Duo Mobile 2.0GHz

USA broadcast, decoded by ffdshow libavcodec, rendered to Overlay Mixer (rendering to NULL didn't work for me).

YV12 dfps: 127,5
Low Quality RGB dfps: 65,6
High Quality RGB dfps: 65,5
When playing back in MPC HC, the CPU consumption with High Quality is a bit higher than without High Quality, though, I think.

[yesgrey]

Is your E2160 heavily overclocked Pentium DC?
Yes. The name is Pentium DC, but it's a Core2 Duo based cpu, with 1MB cache.
Standard clock: 9x200MHz, I overclocked to 9x333MHz.:cool:
Maybe it's performance is lower due to the small cache...

[haruhiko_yamagata]

I have committed all my local changes. (Build (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2671_20090209_Q.exe))

[_xxl]

haruhiko,
Here are my results with Intel E2160@3.0GHz (Core2 duo) XPSP3:
Test Clip 1 - xvid 592x256
ffdshow 2653 enabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 824.0, dfps: 144.3

ffdshow 2653 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 824.0, dfps: 148.7

Test Clip 2 - h264 1920x816
ffdshow 2653 enabled High Quality YV12 to RGB conversion:
User: 1s, kernel: 0s, total: 1s, real: 4s, fps: 116.5, dfps: 41.7

ffdshow 2653 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 1s, real: 4s, fps: 165.7, dfps: 46.3

Test Clip 1 - xvid 592x256
ffdshow 2666 enabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 988.8, dfps: 145.4

ffdshow 2666 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 0s, real: 2s, fps: 941.7, dfps: 147.6

Test Clip 2 - h264 1920x816
ffdshow 2666 enabled High Quality YV12 to RGB conversion:
User: 1s, kernel: 0s, total: 1s, real: 4s, fps: 115.4, dfps: 42.8

ffdshow 2666 disabled High Quality YV12 to RGB conversion:
User: 0s, kernel: 0s, total: 1s, real: 4s, fps: 159.6, dfps: 47.9

I think that no MT is used for a reason with 2666 and your cpu.

[Mercury_22]

I have committed all my local changes. (Build (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2671_20090209_Q.exe))

Yes the problem has been fixed ! :thanks:

Any chance for a 64-bit build too ? PLEASE ! :helpful:

[@Tron@]

So ... How to compile ffdshow and what programs need ??????

[haruhiko_yamagata]

I think that no MT is used for a reason with 2666 and your cpu."High quality ..." was much slower before multithreading, now it's close on his special Pentium DC. Considering the heavy overclocking, I think his result is reasonable.
If his CPU was not overclocked, CPU would be more bottleneck than memory and result might be different.

[haruhiko_yamagata]

So ... How to compile ffdshow and what programs need ??????
This (http://ffdshow-tryout.sourceforge.net/wiki/playground:playground) was the secret place where we were planing for our new wiki.
I'm not sure if this will be our formal wiki.

[@Tron@]

This (http://ffdshow-tryout.sourceforge.net/wiki/playground:playground) was the secret place where we were planing for our new wiki.
I'm not sure if this will be our formal wiki.

Thank you.

But it is possible to explain the details that must be installed and how, how to install that and in what directory, and then how to make exe file, etc. ...

[haruhiko_yamagata]

Any chance for a 64-bit build too ? PLEASE ! :helpful:
OK, here (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2671_20090209_Q_x64.exe) you go.

[haruhiko_yamagata]

Thank you.

But it is possible to explain the details that must be installed and how, how to install that and in what directory, and then how to make exe file, etc. ...
No, I don't want to explain how to install GCC...

[STaRGaZeR]

With a 45nm Core2Duo, E8400@4GHz (445x9)

H.264 1920x1080 ~37Mbps Blu-ray:

High quality YV12 to RGB conversion enabled --> dfps: 48.8
High quality YV12 to RGB conversion disabled --> dfps: 46.9

H.264 1440x1080 ~15Mbps x264:

High quality YV12 to RGB conversion enabled --> dfps: 85.2
High quality YV12 to RGB conversion disabled --> dfps: 80.9

Tested 3 times each, just to be sure :)

Using Haruhiko's post 6574 build (http://forum.doom9.org/showpost.php?p=1247337&postcount=6574).

[Mercury_22]

OK, here (http://ffdshow-tryout.sourceforge.net/doom9/ffdshow_rev2671_20090209_Q_x64.exe) you go.

:thanks:

[whurlston]

This patch will allow the audio decoder to connect to the ArcSoft Demux and decode EAC3 audio tracks. It also defines the wFormatTag as WAVE_FORMAT_DTS_HD instead of WAVE_FORMAT_DTS_W when the media subtype is MEDIASUBTYPE_ARCSOFT_DTSHD.

Index: src/ffdshow_mediaguids.h
===================================================================
--- src/ffdshow_mediaguids.h (revision 2672)
+++ src/ffdshow_mediaguids.h (working copy)
@@ -394,6 +394,7 @@

DEFINE_GUID(MEDIASUBTYPE_NERO_MLP ,0x1e889be7, 0xb276, 0x4064, 0x9a, 0x39, 0x16, 0x0a, 0x06, 0x89, 0x5b, 0x52);
DEFINE_GUID(MEDIASUBTYPE_ARCSOFT_MLP ,0x4288b843, 0x610b, 0x4e15, 0xa5, 0x3b, 0x43, 0x00, 0x7f, 0xcf, 0xf6, 0x14);
+DEFINE_GUID(MEDIASUBTYPE_ARCSOFT_DDPLUS ,0x71cfa727, 0x37e4, 0x404a, 0xae, 0xc0, 0x34, 0x84, 0x25, 0x32, 0xef, 0xf7);
DEFINE_GUID(MEDIASUBTYPE_SONIC_MLP ,0x4094a857, 0x7891, 0x44ac, 0x92, 0xb5, 0xc1, 0xcf, 0xf3, 0x7a, 0xf2, 0xe7);
DEFINE_GUID(MEDIASUBTYPE_ARCSOFT_DTSHD ,0xf6498f57, 0xb399, 0x4a43, 0xa6, 0xfa, 0xf6, 0x94, 0xad, 0x42, 0xb9, 0xbe);
DEFINE_GUID(MEDIASUBTYPE_DOLBY_DDPLUS ,0xa7fb87af, 0x2d02, 0x42fb, 0xa4, 0xd4, 0x5, 0xcd, 0x93, 0x84, 0x3b, 0xdd);
Index: src/TffDecoder_reg.cpp
===================================================================
--- src/TffDecoder_reg.cpp (revision 2672)
+++ src/TffDecoder_reg.cpp (working copy)
@@ -435,6 +435,7 @@
{ &MEDIATYPE_Audio , &MEDIASUBTYPE_Vorbis2 },
{ &MEDIATYPE_Audio , &MEDIASUBTYPE_VorbisIll },
{ &MEDIATYPE_Audio , &MEDIASUBTYPE_ARCSOFT_MLP },
+ { &MEDIATYPE_Audio , &MEDIASUBTYPE_ARCSOFT_DDPLUS },
{ &MEDIATYPE_Audio , &MEDIASUBTYPE_ARCSOFT_DTSHD },
//{ &MEDIATYPE_Audio , &MEDIASUBTYPE_EAC3 },
{ &MEDIATYPE_Audio , &MEDIASUBTYPE_PCM },
Index: src/TffdshowDecAudio.cpp
===================================================================
--- src/TffdshowDecAudio.cpp (revision 2672)
+++ src/TffdshowDecAudio.cpp (working copy)
@@ -212,7 +212,7 @@
wFormatTag=WAVE_FORMAT_VORBIS;
else if (mt.subtype==MEDIASUBTYPE_DOLBY_AC3)
wFormatTag=WAVE_FORMAT_AC3_W;
- else if (mt.subtype==MEDIASUBTYPE_DTS || mt.subtype==MEDIASUBTYPE_ARCSOFT_DTSHD)
+ else if (mt.subtype==MEDIASUBTYPE_DTS)
wFormatTag=WAVE_FORMAT_DTS_W;
else if (mt.subtype==MEDIASUBTYPE_DVD_LPCM_AUDIO)
wFormatTag=WAVE_FORMAT_LPCM;
@@ -233,9 +233,9 @@
else if (mt.subtype==MEDIASUBTYPE_DOLBY_TRUEHD || mt.subtype==MEDIASUBTYPE_NERO_MLP
|| mt.subtype==MEDIASUBTYPE_ARCSOFT_MLP || mt.subtype==MEDIASUBTYPE_SONIC_MLP)
wFormatTag=WAVE_FORMAT_MLP;
- else if (mt.subtype==MEDIASUBTYPE_DOLBY_DDPLUS)
+ else if (mt.subtype==MEDIASUBTYPE_DOLBY_DDPLUS || mt.subtype==MEDIASUBTYPE_ARCSOFT_DDPLUS)
wFormatTag=WAVE_FORMAT_EAC3;
- else if (mt.subtype==MEDIASUBTYPE_DTS_HD)
+ else if (mt.subtype==MEDIASUBTYPE_DTS_HD || mt.subtype==MEDIASUBTYPE_ARCSOFT_DTSHD)
wFormatTag=WAVE_FORMAT_DTS_HD; // TODO : define a separate codecId for DTSHD when available
else
{

[ikarad]

Ok I reproduce the problem : it is working but the jitter causes problem. I remember I disabled jitter correction for MLP because it was causing problems. I will digg this around.

Concerning LPCM, I have no problem playing LPCM 2.0 but you mentioned LPCM coming from blurays, maybe the bandwidth is higher or the sample rate is different. Anyway, if you are able to extract a sample it will help. Also it may be interesting if you can confirm that you also have no sound using SPDIF output of your motherboard and not HDMI of your graphic card.
for the sample : This sample have three soundtracks: one mlp and two lpcm 2.1.
http://www.zshare.net/info.html?54307328-f4cf1e029ba268fd6fc1e3907e3ab1dc

a second example with lpcm 2.1 and lpcm 6.1 soudntrack
http://www.zshare.net/info.html?55241905-3d52058746470a1fa90c6eed8b81c6ea


For spidf, I'm sorry but I can't try because my speakers only offers analogic output and not digital output. I d'ont use HDMI from my grphic card.
I use vga connector from my graphic card (I have CRT monitor) and for sound I use analogic connector from my soundcard (xfi titanium)


Albain, have you succeed to download my two examples?

[dandy]

re merging mt and main branch

tested mt "just for educational purposes" - and saw
the luma sharpen (resize/setting page) option has different scales (0-2 for normal / 0-1 for mt) - and
luma sharpen doesn't seeem to work with mt (mpeg2 spline resize)

this may only one example for subtle differences between mt and main
for me it wouldn't be nice to loose the resize/sharpen functionality

thanks

[clsid]

The functionality of the trunk will stay the same. So don't worry. The only differences will be in libavcodec, which will have two variants: normal and MT.

[albain]

Albain, have you succeed to download my two examples?

Yes, and I reproduced the problem which is different from the former issue.

However, I am quite busy at work right now so it may take a while until I come back with a fix.

[mark0077]

Hi guys.

Testing the colorspace conversions as I am about to update my dvd quality thread with all of my new findings and suggestions. It seems I have found a bug in ffdshows recent builds.

I am testing using the following clip. I notice in some recent builds of ffdshow and mpc-hc I see nothing when viewing this clip outside of its original DVD.

http://www.mediafire.com/?9g9ddlfzxhv

Here are my results using the mpc-hc mpeg2 decoder (so i can set it to feed out yv12 and yuy2 from my yv12 dvds), version 988, and ffdshow 2671. Only good thing in the new ffdshow versions from my tests weeks ago is the new excellent yv12 - rgb32 conversion. The yv12 to yuy2 conversion is still very bad.

mpc-hc yv12 -> ffdshow yv12-rgb32 -> nvidia rgb32 (VERY GOOD, EXCELLENT WORK GUYS)
mpc-hc yv12 -> ffdshow yv12-yuy2 -> nvidia yuy2-rgb32 (VERY BAD AS ALWAYS)
mpc-hc yuy2 -> ffdshow yuy2-rgb32 -> nvidia rgb32 (CRASH)
mpc-hc yuy2 -> ffdshow yuy2-yv12 -> nvidia yv12-rgb32 (USUALLY EXCELLENT QUALITY BUT NOW CRASH)

[BatKnight]

for the sample : This sample have three soundtracks: one mlp and two lpcm 2.1.
http://www.zshare.net/info.html?54307328-f4cf1e029ba268fd6fc1e3907e3ab1dc

a second example with lpcm 2.1 and lpcm 6.1 soudntrack
http://www.zshare.net/info.html?55241905-3d52058746470a1fa90c6eed8b81c6ea


For spidf, I'm sorry but I can't try because my speakers only offers analogic output and not digital output. I d'ont use HDMI from my grphic card.
I use vga connector from my graphic card (I have CRT monitor) and for sound I use analogic connector from my soundcard (xfi titanium)
Ikarad, i've tested your second sample using Media Player Classic HomeCinema (x86), svn 988 + Beliyaal Fixes Ver 8 http://forum.doom9.org/showpost.php?p=1238556&postcount=5934 and using the internal LPCM filter of MPC-HC it plays ok.
I then succesfully AC3 encoded it and SPDIFed it with AC3 Filter. This allowed me to have proper surround channel mapping from a multichannel LPCM on a digital-only output.

Do you have other problems?

Bat

[ikarad]

Ikarad, i've tested your second sample using Media Player Classic HomeCinema (x86), svn 988 + Beliyaal Fixes Ver 8 http://forum.doom9.org/showpost.php?p=1238556&postcount=5934 and using the internal LPCM filter of MPC-HC it plays ok.
I then succesfully AC3 encoded it and SPDIFed it with AC3 Filter. This allowed me to have proper surround channel mapping from a multichannel LPCM on a digital-only output.

Do you have other problems?

Bat
here is the ffdshow topic and not MPC-HC topic.
Yes, with internal LPCM filter of MPC-HC it works but not with ffdshow audio.
The problem is that FFDSHOW doesn't detects and decode lpcm soundtrack from blu-ray.

[madshi]

Just received information from Microsoft about how to pass TrueHD/DTS-HD to a DirectShow audio renderer in Windows 7. If anybody is interested, please PM me.

[leeperry]

if anyone's interested, I'm doing some betatesting for Christian-W. Budde and his VST>winamp2 bridge plugin :
http://www.savioursofsoul.de/Christian/?page_id=248

the newest beta doesn't freeze/crash anymore on opening new files :eek:

still polishing it a bit(like having no GUI visible when opening a file, etc...), should he rename the DLL or please Haruhiko, can you remove the blacklist on "dsp_vst.dll" ?

anyway, w/ Effect Chainer mixing VST/DX plugins, it's a riot :D

http://www.image-load.eu/out.php/i144947_vst.png

[haruhiko_yamagata]

mpc-hc yv12 -> ffdshow yv12-yuy2 -> nvidia yuy2-rgb32 (VERY BAD AS ALWAYS)
I can't reproduce. Please make sure "Set interlaced flags in output media type" in MPC's MPEG-2 properties is unchecked (or force progressive in ffdshow).

mpc-hc yuy2 -> ffdshow yuy2-rgb32 -> nvidia rgb32 (CRASH)
mpc-hc yuy2 -> ffdshow yuy2-yv12 -> nvidia yv12-rgb32 (USUALLY EXCELLENT QUALITY BUT NOW CRASH)
I can't reproduce. Please try the latest (2672 or newer) if you are using a build of a few days ago.

[bwlonsdale]

Hi all - does someone know if there should be any difference in having FFDShow decode AC3 against passing it through to my AMP and have it do the decoding?

I've noticed that the bass is reduced when I let FFDShow decode the stream.

I'm using an ATI 4600, set for multichannel out (5.1). I've made sure there are no options selected in the sound control panel applet, i.e. no effects or any bass management settings are selected, it shouldn't be doing anything to the audio.

My amp is a Denon 1909. I only noticed this the other day when I was messing with Slysoft's Reclock. I had an issue with it so I uninstalled it and then had FFDShow temporarily pass through the AC3 direct to the amp and then I noticed the bass volume either seemed to be louder or at least it gave more of a thump. It's not my imagination at work, I had someone else check too :)

I have it in my head that there was some issue around bass being 15db quieter when decoded by FFDShow, but that was something I either read in passing or have just made up in my head.

I have FFDShow configured to decode AC3 and the mixer set for 5.1 output, no other options selected for bass management etc. Pretty much everything is at defaults in the audio side of things.

Can anyone confirm or deny I need to be doing anything with bass management if I let FFDShow do the decoding?

This is in Vista Media Center - didn't try anything else yet. I like to let FFDShow do the decoding because then I can use the volume control on the VMC remote rather than reach for another one when I need to turn it down/mute.

[mark0077]

You were right, build 2676 fixed my crashes with colorspace conversion.

The yv12 to yuy2 conversion problem was narrowed down to nvidia doing bad conversion from yuy2 to rgb32....

My suggestions for everyone I guess are always output rgb32hq from ffdshow ;) Thanks alot.

[Brazil2]

This patch will allow the audio decoder to connect to the ArcSoft Demux and decode EAC3 audio tracks. It also defines the wFormatTag as WAVE_FORMAT_DTS_HD instead of WAVE_FORMAT_DTS_W when the media subtype is MEDIASUBTYPE_ARCSOFT_DTSHD.
Aaaah, that would be really great! http://www.lfsforum.net/images/smilies/thumb.gif

[Carpo]

I did search (maybe not that well ;p) but is anyone here getting stuttering video using latest versions of x64 and x86 ffdshow when viewing full screen in vista media center, i have googled for a fix but none seem to work, plays fine in wmp full screen media player classic, well any player for that matter untill i use vista mce and seeing as i use the pc as a HTPC its a bit of a pain in the rear

[clsid]

Update your graphics driver.

[wozio]

My suggestions for everyone I guess are always output rgb32hq from ffdshow ;) Thanks alot.
... or keep colors in nv12 all the way up to renderer which will do conversion to rgb in graphic card.