Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > New and alternative video codecs

Reply
 
Thread Tools Search this Thread Display Modes
Old 31st January 2012, 00:05   #14621  |  Link
oddball
Registered User
 
Join Date: Jan 2002
Posts: 1,264
I am having issues with ffdshow (any build as far as I can tell) hanging when playing back MKV with AAC audio. Sometimes it plays fine and other times MPC-HC (Latest xvidvideo.ru builds) just hangs with the video and audio still playing but you cannot pause, stop or exit the player. I have to end it in task manager and it ends with the audio continuing for a second like it's a stuck CD going really fast on stuck sample. This only just started with recent builds and I don't know what to make of it other than to try rolling back MPC-HC, ffdshow and MadVR, I also use ReClock with WASAPI. But never had this issue until recently. I use a extrernal USB soundcard but never had issues with that until now either.
oddball is offline   Reply With Quote
Old 31st January 2012, 13:25   #14622  |  Link
haruhiko_yamagata
Registered User
 
Join Date: Feb 2006
Location: Japan
Posts: 1,560
Quote:
Originally Posted by oddball View Post
I am having issues with ffdshow (any build as far as I can tell) hanging when playing back MKV with AAC audio. Sometimes it plays fine and other times MPC-HC (Latest xvidvideo.ru builds) just hangs with the video and audio still playing but you cannot pause, stop or exit the player. I have to end it in task manager and it ends with the audio continuing for a second like it's a stuck CD going really fast on stuck sample. This only just started with recent builds and I don't know what to make of it other than to try rolling back MPC-HC, ffdshow and MadVR, I also use ReClock with WASAPI. But never had this issue until recently. I use a extrernal USB soundcard but never had issues with that until now either.
Thanks for report, but we need more information to reproduce.
__________________
[ Download ffdshow | Wiki ]
haruhiko_yamagata is offline   Reply With Quote
Old 31st January 2012, 15:18   #14623  |  Link
oddball
Registered User
 
Join Date: Jan 2002
Posts: 1,264
Nevermind it appears to be a switching but when going fullscreen on my HDTV. Going from 60Hz to 24Hz in conjunction with ReClock appears to be the culprit.
oddball is offline   Reply With Quote
Old 1st February 2012, 12:23   #14624  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,645
Haruhiko, I can reproduce another Rundll32 error. Simply rename any default preset and press enter, DO NOT click apply, then press ALT + F4.

Last edited by ryrynz; 1st February 2012 at 12:26.
ryrynz is offline   Reply With Quote
Old 1st February 2012, 19:48   #14625  |  Link
ipanema
Registered User
 
Join Date: Apr 2009
Posts: 93
Some time between rev 3914 (Jun 29 2011) and rev (4291 Jan 31 2012) there seems to have been a bug introduced in the timestamps of the first decoded frame for the H.264/AVC decoder.

For example the older rev 3914 produces output frames with these start and end timestamps:
5575666 5575667
5909333 5909333
6243000 6243001
6576666 6576667

The difference is always correctly 333666, and the end timestamp is always 1 greater then the start timestamp.

But the newer rev 4291 gives these output timestamps:
0 399999
5909333 5909334
6243000 6243001
6576666 6576667

The timestamp of the first frame is way too low and much less than the input sample timestamps. Then there's a large jump to the timestamp values of the second frame which seems to be correct (as are all subsequent frames).

This causes a glitch or error when fed to downstream filters.

I don't know at what rev this problem first appeared, but at least rev 3914 works fine.

Is it possible to get this fixed ?

Thanks.


EDIT: I've just gone through all the previous versions of ffdshow-tryouts in

http://sourceforge.net/projects/ffds...eric%20builds/

and the problem seems to have been introduced at rev 4194. The previous rev 4174 seems to be OK (first frame has correct timestamps).

Hope that helps.

Last edited by ipanema; 1st February 2012 at 20:32.
ipanema is offline   Reply With Quote
Old 1st February 2012, 19:55   #14626  |  Link
kc7bfi
Registered User
 
Join Date: Nov 2010
Location: Chesapeake, VA USA
Posts: 38
Problems with ffdshow_rev4291_20120131_clsid_x64.exe

I just installed ffdshow_rev4291_20120131_clsid_x64.exe and now, when trying to play a video I get a popup ffdshow error window saying it failed to load ffmpeg.dll. I am running Windows7 64-bit. Thanks, David
kc7bfi is offline   Reply With Quote
Old 1st February 2012, 20:15   #14627  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,640
A fresh working x64 build will be available shortly.
__________________
MPC-HC 2.1.7.2
clsid is offline   Reply With Quote
Old 1st February 2012, 22:24   #14628  |  Link
kc7bfi
Registered User
 
Join Date: Nov 2010
Location: Chesapeake, VA USA
Posts: 38
H.264 video appears to be playing only I-frames

Here is a cell phone video (http://dl.dropbox.com/u/37804952/201...-17-41_209.3gp) taken of our attempt to play H.264 video from an Optelecom encoder. It appears that only I-frames are being decoded. Any thoughts what the problem might be? I am running Windows7 64-bit and ffdshow ffdshow_rev4293_20120201_clsid_x64. Thanks, David
kc7bfi is offline   Reply With Quote
Old 2nd February 2012, 01:56   #14629  |  Link
CruNcher
Registered User
 
CruNcher's Avatar
 
Join Date: Apr 2002
Location: Germany
Posts: 4,926
you trying to playback the raw bitstream, what is the parser ?
though your chances arent high that it will work flawlessly as ffdshow isn't meant for this use case
__________________
all my compares are riddles so please try to decipher them yourselves :)

It is about Time

Join the Revolution NOW before it is to Late !

http://forum.doom9.org/showthread.php?t=168004

Last edited by CruNcher; 2nd February 2012 at 02:04.
CruNcher is offline   Reply With Quote
Old 2nd February 2012, 12:21   #14630  |  Link
haruhiko_yamagata
Registered User
 
Join Date: Feb 2006
Location: Japan
Posts: 1,560
Quote:
Originally Posted by ryrynz View Post
Haruhiko, I can reproduce another Rundll32 error. Simply rename any default preset and press enter, DO NOT click apply, then press ALT + F4.
I can reproduce the crash. Thanks.
__________________
[ Download ffdshow | Wiki ]
haruhiko_yamagata is offline   Reply With Quote
Old 2nd February 2012, 12:25   #14631  |  Link
haruhiko_yamagata
Registered User
 
Join Date: Feb 2006
Location: Japan
Posts: 1,560
Quote:
Originally Posted by ipanema View Post
Some time between rev 3914 (Jun 29 2011) and rev (4291 Jan 31 2012) there seems to have been a bug introduced in the timestamps of the first decoded frame for the H.264/AVC decoder.
I think it's rev 4179.
Would you provide the sample?
__________________
[ Download ffdshow | Wiki ]
haruhiko_yamagata is offline   Reply With Quote
Old 2nd February 2012, 17:46   #14632  |  Link
Pat357
Registered User
 
Join Date: Jun 2006
Posts: 452
I 've just updated my FFDshow from rev3972 to rev4291.
When setting up FFDshow DXVA, I noticed that there is currently only support for H264 and VC1, no option to choose MPEG2 anymore !
What has happened to the MPEG2 HW acceleration ?
This is really a pitty because I used it often with my GTX 570 (full MPEG2 HW decoding support).
Do we we also loose the HW vector adaptive deinterlacing for all MPEG2 streams ? Like for TV recordings, DVD movies, ..?
I know we have Yadiff and other softw. based options but none comes even close to HW deinterlacing (quality wise),
not to mention the extra load on the CPU.
Can anyone please enlighten me ?
Will the option be back in the near future versions ?

Last edited by Pat357; 2nd February 2012 at 17:52.
Pat357 is offline   Reply With Quote
Old 2nd February 2012, 17:49   #14633  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,336
Deinterlacing through the renderer is not related to DXVA decoding, it should work equally well with software decoding as long as the decoder is setting the flags properly.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 2nd February 2012, 21:52   #14634  |  Link
ipanema
Registered User
 
Join Date: Apr 2009
Posts: 93
Quote:
Originally Posted by haruhiko_yamagata View Post
I think it's rev 4179.
Would you provide the sample?
Do you mean the MTS file that produced the timestamps I quoted? If so where do I provide it?
ipanema is offline   Reply With Quote
Old 2nd February 2012, 22:15   #14635  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,640
Quote:
Originally Posted by Pat357 View Post
I 've just updated my FFDshow from rev3972 to rev4291.
When setting up FFDshow DXVA, I noticed that there is currently only support for H264 and VC1, no option to choose MPEG2 anymore !
What has happened to the MPEG2 HW acceleration ?
This is really a pitty because I used it often with my GTX 570 (full MPEG2 HW decoding support).
Do we we also loose the HW vector adaptive deinterlacing for all MPEG2 streams ? Like for TV recordings, DVD movies, ..?
I know we have Yadiff and other softw. based options but none comes even close to HW deinterlacing (quality wise),
not to mention the extra load on the CPU.
Can anyone please enlighten me ?
Will the option be back in the near future versions ?
There never has been such an option. There are also no plans to add it.
You should use LAV Video instead. That provides CUVID HW acceleration with HW deinterlacing.
__________________
MPC-HC 2.1.7.2
clsid is offline   Reply With Quote
Old 2nd February 2012, 22:17   #14636  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,640
Quote:
Originally Posted by ipanema View Post
Do you mean the MTS file that produced the timestamps I quoted? If so where do I provide it?
You can upload it to for example:
http://www.mediafire.com

If the file is very large and you want to cut a sample, then use DGSplit:
http://neuron2.net/dgsplit/dgsplit12.zip
__________________
MPC-HC 2.1.7.2
clsid is offline   Reply With Quote
Old 3rd February 2012, 14:35   #14637  |  Link
kc7bfi
Registered User
 
Join Date: Nov 2010
Location: Chesapeake, VA USA
Posts: 38
Problem playing Optelecom H.264 video

Quote:
Originally Posted by kc7bfi View Post
Here is a cell phone video (http://dl.dropbox.com/u/37804952/201...-17-41_209.3gp) taken of our attempt to play H.264 video from an Optelecom encoder. It appears that only I-frames are being decoded. Any thoughts what the problem might be? I am running Windows7 64-bit and ffdshow ffdshow_rev4293_20120201_clsid_x64. Thanks, David
I have confirmed that the problem with the Optelecom H.264 video was created somewhere between version 4174 and 4194. Version 4174 plays the video fine but 4194 seems to only be playing I-frames. Any ideas what may have changed in those versions? If there are any intermediate Windows 64 bit installs I can also test them to try and narrow down the exact revision that caused the problem. Just let me know if they are available. Any help in resolving this issue would be greatly appreciated. David
kc7bfi is offline   Reply With Quote
Old 3rd February 2012, 15:06   #14638  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,640
I am making preparations for a new official release that is planned for later this month, once remaining regressions have been fixed.

Everyone please review this changelog for omissions and mistakes:
Quote:
(revision 3154 to 4296)

Bugfixes:
* Lots of updates for libavcodec that bring bug fixes, performance enhancements, and security fixes.
* Disable "Detect soft telecine and average frame durations" if 3:2 pulldown is being applied in AviSynth filter.
* IAMStreamSelect: fix duplicated subtitle list in case of multiple ffdshow instances
* If a matroska container had hidden leading chapter and Haali's splitter was used and tray icon was right-clicked, ffdshow crashed.
* Various fixes regarding input mediatype handling, strides, and pin re-connection.
* Various fixes regarding H.264 SPS parsing
* Fixed decoding of 96khz 32bit audio with libavcodec
* Fixed decoding of AAC audio with libavcodec when extradata is missing
* Better handling of non-mod4 resolutions
* Workaround for freeze in Powerpoint 2010
* AC3 encoder was producing improper bitstream when connected to file writer filter
* Various fixes for audio bitstreaming

New and improved features:
* Support for H.264 10-bit 4:2:0/4:2:2/4:4:4/RGB
* Support for H.264 8-bit 4:2:2/4:4:4/RGB, 4:2:0 was already supported
* Support for various 10-bit and 16-bit colorspaces
* Optimized YV12 to NV12 conversion
* Enabled high quality RGB32 conversion by default
* Added option to select a preferred output colorspace
* Added Intel QuickSync decoder. This provides hardware acceleration decoding of H.264, VC-1 and MPEG-2. Requires the integrated GPU of Intel Sandy Bridge processors.
* Added DXVA video filter for hardware accelerated decoding. Supports H.264 and VC-1 video. Also includes subtitle support.
* Various additions to the application blacklist and whitelist
* Added support for Blu-ray PGS subtitles
* Removed support for USF subtitles
* Various improvements for the handling of ASS/SSA subtitles. More tags are supported now, but not yet all of them.
* Various improvements for the rendering of text-based subtitles
* By default now only subtitle files are shown that match the video filename. This behavior can be adjusted through an
option on the "Try, dialog & Paths" page in ffdshow settings.
* Created new format option in ffdshow audio decoder for the QT PCM variants (such as TWOS). They were previously part if the Uncompressed audio option.
* ffmpeg-mt has been merged into libavcodec
* Multi-threaded decoding is now used by default for H.264, MPEG-1/2, FFV1, and DV video.
* Added decoding support for Indeo 4 and Indeo 5 video
* Floating point output for libavcodec AAC, AC3, E-AC3, DTS, Vorbis, and Nellymoser decoders.
* Added option to disable jitter correction for audio decoder
* Improved performance of Deband filter
* Removed some encoders for which the interface that ffdshow provided was outdated and unmaintained, such as Xvid and
H.264. You should use the official codecs instead.
* Removed some encoders that were considered obsolete or unstable.

Updates:
* Updated Boost library to v1.48
* Updated zlib library to v1.2.6
* Updated minilzo library to v2.06
* Updated unrar library to v4.10
* Updated libsamplerate to v0.1.8
* Removed mp3lib
* Removed xvidcore
__________________
MPC-HC 2.1.7.2
clsid is offline   Reply With Quote
Old 3rd February 2012, 19:36   #14639  |  Link
ipanema
Registered User
 
Join Date: Apr 2009
Posts: 93
Quote:
Originally Posted by clsid View Post
You can upload it to for example:
http://www.mediafire.com

If the file is very large and you want to cut a sample, then use DGSplit:
http://neuron2.net/dgsplit/dgsplit12.zip
Thanks. I have just been doing some more checks for 3 MTS files chosen at random and they all have similar output timestamp problems which were not present in the older revisions of ffdshow tryouts. And it isn't just the timestamp of the first frame. There are some very strange timestamp values jumping forward and then backward (this is the output remember), and the end timestamp value is sometimes strange aswell. In comparison, the older revisions always produce correct values with the end timestamp always 1 greater than the start.

Here are the start and end timestamps for the 3 files I tried. The input timestamps are shown, followed by the "correct" output timestamps from the older revision 4174, followed by the weird values produced by the newer revision.

File 1

Input Timestamps
5575666 5575667
4908333 4908334
5242000 5242001
6576666 6576666
5909333 5909334
6243000 6243001
7577666 7577667
6910333 6910334
7244000 7244001
8578666 8578667
7911333 7911334
8245000 8245001

Output Timestamps from revision 4174
5575666 5575667
6576666 6576667
6910333 6910334
7244000 7244001
7577666 7577667
7911333 7911334

Weird Output Timestamps from newer revision
0 399999
6576666 6576667
6910333 6910334
7244000 7244001
7644000 8043999
7911333 7911334


File 2

Input Timestamps
2918111 2918112
2118111 2118112
2518111 2518112
3718111 3718112
3318111 3318112
4518111 4518112
4118111 4118112
5318111 5318112
4918111 4918112
6118111 6118112
5718111 5718112
6918111 6918112
6518111 6518112

Output Timestamps from revision 4174
2118111 2118112
2518111 2518112
2918111 2918112
3318111 3318112
3718111 3718112
4118111 4118112
4518111 4518112
4918111 4918112
5318111 5318112
5718111 5718112
6118111 6118112

Weird Output Timestamps from newer revision
2118111 2118112
3718111 3718112
4118111 4518110
3318111 3318112
3718111 3718112
5318111 5318112
4518111 4518112
6118111 6118112
5318111 5318112
6918111 6918112
6118111 6118112


File 3

Input Timestamps
7770333 7770334
7103000 7103001
7436666 7436667
8771333 8771334
8104000 8104001
8437666 8437667
9772333 9772334
9105000 9105001
9438666 9438667
10773333 10773333
10106000 10106001
10439666 10439667
11774333 11774334
11107000 11107001
11440666 11440667
12775333 12775334
12108000 12108001
12441666 12441667
13776333 13776334
13109000 13109001
13442666 13442666

Output Timestamps from revision 4174
7770333 7770334
8771333 8771334
9105000 9105001
9438666 9438667
9772333 9772334
10106000 10106001
10439666 10439667
10773333 10773334
11107000 11107001
11440666 11440667
11774333 11774334
12108000 12108001
12441666 12441667
12775333 12775334
13109000 13109001

Weird Output Timestamps from newer revision
0 399999
8771333 8771334
9438666 9438667
10773333 10773334
11173333 11573332
10439666 10439667
10439666 10439667
10773333 10773334
11107000 11107001
12775333 12775334
13175333 13575332
12441666 12441667
12441666 12441667
12841666 13241665
13109000 13109001


Given that all 3 of the files I selected at random show the problem within the first few frames, I would think that you would be able to reproduce this quite easily using one of your own files (all 3 were from different AVCHD camcorders).

However, if you still think you need an example file, let me know and I will upload one for you.
ipanema is offline   Reply With Quote
Old 4th February 2012, 01:26   #14640  |  Link
haruhiko_yamagata
Registered User
 
Join Date: Feb 2006
Location: Japan
Posts: 1,560
Quote:
Originally Posted by clsid View Post
I am making preparations for a new official release that is planned for later this month, once remaining regressions have been fixed.

Everyone please review this changelog for omissions and mistakes:
Good log.
One thing I want for the next release is fast "Accurate rounding" in resize filter. Actually it's a regression.
__________________
[ Download ffdshow | Wiki ]
haruhiko_yamagata is offline   Reply With Quote
Reply

Tags
ffdshow, ffdshow tryouts, ffdshow-mt, ffplay, icl

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 06:17.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.