ffdshow tryouts project: Discussion & Development - Page 65

LoRd_MuldeR · 22nd May 2007, 21:29

I just noticed a bug with ffdshow and anamorphic video!

The video was encoded with Xvid at an resolution of 688x560 and a PAR auf 140:129.
The resulting video plays 100% fine at 4:3 in both MPlayer and VLC Media Player.

Unfortuantely this is what happens with MPC + ffdshow:

Tested with ffdshow-tryouts revision 1187. Also I use the latest Haali Renderer.
As soon as I use the XviD Decoder-Filter instead of ffdshow, it works okay in MPC too.

I uploaded a sample:
Download: http://www.mytempdir.com/1335646
Password: doom9

Episode · 22nd May 2007, 22:23

@LoRd_MuldeR, Sorry, but I cannot reproduce this with rev 1187, mpc and haali renderer. It plays perfectly fine with any renderer available in mpc.

GmorG McRoth · 22nd May 2007, 22:29

Same here works perfectly well.

LoRd_MuldeR · 22nd May 2007, 22:31

Quote:

Originally Posted by Episode

@LoRd_MuldeR, Sorry, but I cannot reproduce this with rev 1187, mpc and haali renderer. It plays perfectly fine with any renderer available in mpc.

I just tried other renderers and the problem seems to be Haali Renderer + ffdshow

ffdshow + VMR9 (Renderless): Okay
ffdshow + Overlay: Okay
ffdshow + Haali Renderer: Broken (see screenshot above)

Xvid Decoder + VMR9 (Renderless): Okay
Xvid Decoder + Overlay: Okay
Xvid Decoder + Haali Renderer: Okay

Any ideas ???

GmorG McRoth · 22nd May 2007, 22:32

Quote:

Originally Posted by LoRd_MuldeR

I just tried other renderers and the problem seems to be Haali Renderer + ffdshow

I use Haali and it works. Maybe do full reinstall of Haali filters (uninstall old, restart, install new).
Also Haali uses GPU PS2 for it's calculations if I'm not mistaken, so updating your video card drivers may help.

LoRd_MuldeR · 22nd May 2007, 22:36

Quote:

Originally Posted by GmorG McRoth

I use Haali and it works. Maybe do full reinstall of Haali filters (uninstall old, restart, install new).

I already did that before I wrote the bug report.

GmorG McRoth · 22nd May 2007, 22:43

I see you are using MPC moded by casimir, I used last MPC made by drevil_xxl, maybe here is problem.

LoRd_MuldeR · 22nd May 2007, 22:49

Quote:

Originally Posted by GmorG McRoth

I see you are using MPC moded by casimir, I used last MPC made by drevil_xxl, maybe here is problem.

1. The same problem happens with drevil_xxl's build too

2. As soon as I enable "resize" in ffdshow the problem disappears and (now it gets strange) keeps gone, even if I disable resize again. Once I re-open the video, the problem is back...

GmorG McRoth · 22nd May 2007, 22:53

Wicked XD

ok then, who binary of FFDShow you have installed?
I used ffdshow_rev1187_20070520_clsid.exe

LoRd_MuldeR · 22nd May 2007, 22:56

Quote:

Originally Posted by GmorG McRoth

Wicked XD

ok then, who binary of FFDShow you have installed?
I used ffdshow_rev1187_20070520_clsid.exe

exactly that one

// EDIT

I just went back to ffdshow_beta2a_20070322_clsid.exe and the problem is gone

// EDIT²

Another attempt with ffdshow_rev1187_20070520_clsid.exe and problem returns...

Inventive Software · 22nd May 2007, 23:22

Quote:

Originally Posted by fusion_400

Hi i tested both ffdshow_rev1187_20070520_clsid and ffdshow_rev1193_20070522_xxl builds and both still have the same 30fps bug. FFDSHOW will not encode anything in virtualdub when using 30fps framerate. The FPS is exactly 30fps it is not 29.97fps ntsc standard but exactly 30 fps from canon ixus 30 camera that uses MJPEG 640X480@30FPS.

This is the virtualdub error message:

Cannot start video compression:

An unknown error occurred (may be corrupt data)
(error code -100)

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

From the SVN changelog:

Code:

Revision 1186 - Directory Listing
Modified Sun May 20 14:47:28 2007 UTC (2 days, 7 hours ago) by h_yamagata

Bug fix
The following encoders only work for input with specific frame rates: MPEG4, MPEG 1 and MPEG 2. The accepted frame rates are 25*2^n. For example: 6.25, 12.5, 25, 50, 100.

This is most likely the limitation. The following fixes have been applied, but don't go into any detail:

Code:

Revision 1189 - Directory Listing
Modified Mon May 21 11:32:49 2007 UTC (34 hours, 47 minutes ago) by h_yamagata

minor fix of rev 1186
 (revert Tffvfw.cpp and move the change to TvideoCodecLibavcodec.cpp)

Code:

Revision 1192 - Directory Listing
Modified Tue May 22 13:12:23 2007 UTC (9 hours, 7 minutes ago) by h_yamagata

more fix about rev 1186 issue.

Leak · 23rd May 2007, 00:07

Quote:

Originally Posted by kurt

@ Leak: I got a problem with your avisynth patch. When playing an anamorphic video and enabling seesaw script, mpc (with haali renderer) ignores the PAR flag.

Yeah, something's not right - you should be able to set any aspect ratio you want in your AviSynth script (or just pass the input aspect ratio through), but no matter what values you use the aspect ratio stays the same...

That needs some investigation, but I was busy this evening with building a new version of GCC - I'll look into it tomorrow.

EDIT: I guess I'll better wait with checking in the new AviSynth filter until I've fixed that one.

np: RJD2 - Chicken-Bone Circuit (Deadringer)

Dr Pizza · 23rd May 2007, 02:14

The cpu test introduced in revision 1151 is faulty on x64.

The file in question is Tconfig.cpp.

I don't quite understand the #ifndef __GNUC__ on line 212; it seems to me that IsProcessorFeaturePresent is always the right way to test for processor features on Windows.

But no matter; I think the code as it stands is trying to enable MMX on x64. Although the SSE/SSE2/SSE3/SSSE3 features get tested explicitly, and disabled if they fail, MMX and MMXEXT are never tested, and so, I suspect, never disabled. Forcibly disabling them (available_cpu_flags &= ~FF_CPU_MMX; available_cpu_flags &= ~FF_CPU_MMXEXT

restores ffdshow to working order, as does using IsProcessorFeaturePresent (presumably because IsProcessorFeaturePresent knows that x64 has no MMX).

akupenguin · 23rd May 2007, 02:44

There aren't any x86_64 CPUs without MMX, there aren't any x86_64 OSs that don't support MMX (win64 allows MMX in userspace, just not in the kernel), and ffmpeg doesn't support using SSE* without using MMX.

B.F. · 23rd May 2007, 04:30

URL http://www.comq.ru/user/alex/pirates2/pirates2.mp4
Mpeg4AVC HighProfile video.
Picture is broken using ffdshow, mplayer or KMP.
Look normal using CoreAVC 1.2.

chros · 23rd May 2007, 07:38

Quote:

Originally Posted by LoRd_MuldeR

I just tried other renderers and the problem seems to be Haali Renderer + ffdshow

ffdshow + VMR9 (Renderless): Okay
ffdshow + Overlay: Okay
ffdshow + Haali Renderer: Broken (see screenshot above)

Xvid Decoder + VMR9 (Renderless): Okay
Xvid Decoder + Overlay: Okay
Xvid Decoder + Haali Renderer: Okay

Any ideas ???

Yes, once same was for me too, and I have reported to Haali, but I got no answer:
http://forum.doom9.org/showthread.ph...43#post1001143

Quote:

Originally Posted by Chros

The problem related to ffdshow (I'm using ffdshow_rev1133_20070504_clsid.exe): if The Raw Video is set to All Supported at the Codecs, the problem occurs. So I don't know if it's the fault of Haali's renderer or ffdshow ... (But strange is that VMR is OK.)
(tried several MPC builds (the official 6.4.9.0 too), latest haali's package).

And if ffdshow is in the DS chain, the problem occurs, if ffdshow isn't, the problem gone ...

EDIT: I found an mpg1 (from VCD) which also produces the same bug, when libavcodec is selected at the Codecs in MPG1 ...

chros · 23rd May 2007, 07:48

Quote:

Originally Posted by B.F.

URL http://www.comq.ru/user/alex/pirates2/pirates2.mp4
Mpeg4AVC HighProfile video.
Picture is broken using ffdshow, mplayer or KMP.
Look normal using CoreAVC 1.2.

Confirmed with rev 1133, and it plays fine with CoreAVC 1.3 too

(You sould report this in the coreavc thread, because this is the 1st time when coreavc is able to decode something that ffdshow isn't

)

No matter, if I remux to mkv. Does the whole file produce this, or only this splitted chunk?

PS: the AR is wrong ...

B.F. · 23rd May 2007, 08:14

Quote:

Originally Posted by chros

Confirmed with rev 1133, and it plays fine with CoreAVC 1.3 too

(You sould report this in the coreavc thread, because this is the 1st time when coreavc is able to decode something that ffdshow isn't

)
No matter, if I remux to mkv. Does the whole file produce this, or only this splitted chunk?
PS: the AR is wrong ...

This is not my file.
I was just surprised about ffdshow can't play it normally, so I report about this bug here.

akupenguin · 23rd May 2007, 08:27

Quote:

Originally Posted by B.F.

http://www.comq.ru/user/alex/pirates2/pirates2.mp4

I get artifacts when playing that in MPlayer, and I still get artifacts after remuxing to mkv, but if I demux to raw .264 and play that then it works. (had to demux with MP4Box; mkvextract refuses.) So... maybe a bug in lavc's AVC1 bitstream parser?

chros · 23rd May 2007, 08:27

Quote:

Originally Posted by haruhiko_yamagata

I cannot reproduce. 1080i MPEG2, core2duo, build 1133.
If it is sample specific, please upload a sample.

Thanks for the answer, I have tried with several content (mpg1, mpeg2 DVD, mpeg2 720p, mpeg2 1080i) and it works indeed.
Altough my MPC is sometimes crashing with the 1080i content (so maybe my Windows is rather old...)

22nd May 2007, 21:29	#1281 \| Link
LoRd_MuldeR Software Developer Join Date: Jun 2005 Location: Last House on Slunk Street Posts: 13,248	I just noticed a bug with ffdshow and anamorphic video! The video was encoded with Xvid at an resolution of 688x560 and a PAR auf 140:129. The resulting video plays 100% fine at 4:3 in both MPlayer and VLC Media Player. Unfortuantely this is what happens with MPC + ffdshow: Tested with ffdshow-tryouts revision 1187. Also I use the latest Haali Renderer. As soon as I use the XviD Decoder-Filter instead of ffdshow, it works okay in MPC too. I uploaded a sample: Download: http://www.mytempdir.com/1335646 Password: doom9 __________________ Go to https://standforukraine.com/ to find legitimate Ukrainian Charities 🇺🇦✊ Last edited by LoRd_MuldeR; 22nd May 2007 at 21:34.

22nd May 2007, 22:29	#1283 \| Link
GmorG McRoth Registered User Join Date: Jan 2006 Posts: 137	Same here works perfectly well. __________________ I'm sick of following my dreams. I'm just going to ask them where they're goin', and hook up with them later. ---Mitch Hedberg

22nd May 2007, 22:43	#1287 \| Link
GmorG McRoth Registered User Join Date: Jan 2006 Posts: 137	I see you are using MPC moded by casimir, I used last MPC made by drevil_xxl, maybe here is problem. __________________ I'm sick of following my dreams. I'm just going to ask them where they're goin', and hook up with them later. ---Mitch Hedberg

22nd May 2007, 22:53	#1289 \| Link
GmorG McRoth Registered User Join Date: Jan 2006 Posts: 137	Wicked XD ok then, who binary of FFDShow you have installed? I used ffdshow_rev1187_20070520_clsid.exe __________________ I'm sick of following my dreams. I'm just going to ask them where they're goin', and hook up with them later. ---Mitch Hedberg

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

22nd May 2007, 22:23	#1282 \| Link
Episode n00b Join Date: Jan 2003 Location: Tokyo, Japan Posts: 122	@LoRd_MuldeR, Sorry, but I cannot reproduce this with rev 1187, mpc and haali renderer. It plays perfectly fine with any renderer available in mpc.

23rd May 2007, 02:14	#1293 \| Link
Dr Pizza Registered User Join Date: Feb 2007 Posts: 65	The cpu test introduced in revision 1151 is faulty on x64. The file in question is Tconfig.cpp. I don't quite understand the #ifndef __GNUC__ on line 212; it seems to me that IsProcessorFeaturePresent is always the right way to test for processor features on Windows. But no matter; I think the code as it stands is trying to enable MMX on x64. Although the SSE/SSE2/SSE3/SSSE3 features get tested explicitly, and disabled if they fail, MMX and MMXEXT are never tested, and so, I suspect, never disabled. Forcibly disabling them (available_cpu_flags &= ~FF_CPU_MMX; available_cpu_flags &= ~FF_CPU_MMXEXT restores ffdshow to working order, as does using IsProcessorFeaturePresent (presumably because IsProcessorFeaturePresent knows that x64 has no MMX).

23rd May 2007, 02:44	#1294 \| Link
akupenguin x264 developer Join Date: Sep 2004 Posts: 2,392	There aren't any x86_64 CPUs without MMX, there aren't any x86_64 OSs that don't support MMX (win64 allows MMX in userspace, just not in the kernel), and ffmpeg doesn't support using SSE* without using MMX.

23rd May 2007, 04:30	#1295 \| Link
B.F. Registered User Join Date: Jul 2004 Location: Russia, Novosibirsk Posts: 92	URL http://www.comq.ru/user/alex/pirates2/pirates2.mp4 Mpeg4AVC HighProfile video. Picture is broken using ffdshow, mplayer or KMP. Look normal using CoreAVC 1.2.