HOWTO: Install and use the latest FFmpeg and x264

[FakeOutdoorsman]

qyot27,

Thanks for the input. Did you try installing zlib1g-dev from the repository instead of compiling it? The package from the repository is up to date. I used to have gpac as a dependency for this guide, but I removed it since I didn't think it was necessary.

I may add a sentence about the patches, but I want to keep things simple.

[fishyboy]

Thanks for the guide. I'm using Ubuntu 8.10 and the installation seemed to work without any problems. However, I now get an encoding error where I didn't before, and I was wondering if anyone else has seen it. I'm trying to encode a DVD vob file to an xvid avi (I haven't got around to using the x264 stuff yet!).

ffmpeg -i VTS_01_3.VOB -vcodec libxvid -f avi -vtag XVID -s 320x128 -b 300k
-acodec libmp3lame -ar 44100 -ab 16k -ac 2 -vsync 0 -async 1 out.avi

I know for sure that the above worked with the ffmpeg I had previously from the Medibuntu repos. The error I now get from ffmpeg is:

[libxvid @ 0xa340f50]error, non monotone timestamps 20 >= 20
av_interleaved_write_frame(): Error while opening file

This error occurs only for one of the vobs, but still, it didn't occur before I upgraded ffmpeg. I have found some workarounds, and it might be that my original command was wrong?

The first workaround is to use 'vsync 1' instead of 'vsync 0' (or omitting the vsync option altogether). I only used 'vsync 0' for historical reasons, i.e. I copied it from somewhere. I've yet to find an explanation of the ffmpeg vsync option that I understand:

ffmpeg -i VTS_01_3.VOB -vcodec libxvid -f avi -vtag XVID -s 320x128 -b 300k
-acodec libmp3lame -ar 44100 -ab 16k -ac 2 -vsync 1 -async 1 out.avi

Another thing that avoids the error is to skip the start of the file using '-ss 0.001', which is not ideal because I want all of it:

ffmpeg -i VTS_01_3.VOB -ss 0.001 -vcodec libxvid -f avi -vtag XVID -s 320x128 -b 300k
-acodec libmp3lame -ar 44100 -ab 16k -ac 2 -vsync 0 -async 1 out.avi

Here's some ffmpeg output for those of you that can find problems hidden within gobbledygook:

FFmpeg version SVN-r17855, Copyright (c) 2000-2009 Fabrice Bellard, et al.
  configuration: --enable-gpl --enable-postproc --enable-pthreads --enable-libfaac
--enable-libfaad --enable-libmp3lame --enable-libtheora --enable-libx264 --enable-libxvid
  libavutil     49.15. 0 / 49.15. 0
  libavcodec    52.21. 0 / 52.21. 0
  libavformat   52.31. 1 / 52.31. 1
  libavdevice   52. 1. 0 / 52. 1. 0
  libswscale     0. 7. 1 /  0. 7. 1
  libpostproc   51. 2. 0 / 51. 2. 0
  built on Mar  6 2009 19:14:35, gcc: 4.3.2

I'd usually put this down to my source file being nonsense, but in this case I know it worked before.

Has anyone got any suggestions? Is ffmpeg broken? Is vsync useful?

Thanks in advance.

[pedrotuga]

I hope I don't break the discussion. Just a quick question, aren't there meta-packages for this anywhere?

[FakeOutdoorsman]

The error I now get from ffmpeg is:

[libxvid @ 0xa340f50]error, non monotone timestamps 20 >= 20
av_interleaved_write_frame(): Error while opening file

I was unable to duplicate this problem. Can you make a sample of the offending VOB available? This will copy the first 10 seconds:

ffmpeg -i VTS_01_3.VOB -acodec copy -vcodec copy -t 10 test.vob

I've never used async or vsync (and I don't use libxvid anymore). From man ffmpeg:

-vsync parameter
Video sync method. Video will be stretched/squeezed to match the timestamps, it is done by duplicating and dropping frames. With -map you can select from which stream the timestamps should be taken. You can leave either video or audio unchanged and sync the remaining stream(s) to the unchanged one.

-async samples_per_second
Audio sync method. "Stretches/squeezes" the audio stream to match the timestamps, the parameter is the maximum samples per second by
which the audio is changed. -async 1 is a special case where only the start of the audio stream is corrected without any later correction.

I don't totally understand that. More info:
[Ffmpeg-user] How does -vsync work ?
[Ffmpeg-user] problem with seek to given time position

Has anyone got any suggestions?

1. Try a simpler command:

ffmpeg -i VTS_01_3.VOB -vcodec libxvid -f avi -s 320x128 -b 300k -acodec libmp3lame out.avi

2. Try FFmpeg's mpeg4 encoder:

ffmpeg -i VTS_01_3.VOB -vcodec mpeg4 -vtag XVID -f avi -s 320x128 -b 300k -acodec libmp3lame out.avi

3. Use the libx264 presets as mentioned in the tutorial. Example:

ffmpeg -i VTS_01_3.VOB -vcodec libx264 -s 320x128 -crf 30 -acodec libfaac -ab 96k -threads 0 -vpre hq out.mp4

4. You could try to compile the latest Xvid yourself since the version in the repo is outdated.

[FakeOutdoorsman]

I hope I don't break the discussion. Just a quick question, aren't there meta-packages for this anywhere?

There are some unofficial deb files available:

Up to date FFmpeg xvid and x264 builds ppa

Might be more convenient for some users, but you will be at the mercy of someone else configuring FFmpeg for you. They are untested by me. Use at your own risk.

[andrew.46]

Hi FakeOutdoorsman,

Woooo hoooooooooo FFmpeg release is out!! Will this figure in your guide or will you stick with svn?

Andrew

[FakeOutdoorsman]

Hi FakeOutdoorsman,

Woooo hoooooooooo FFmpeg release is out!! Will this figure in your guide or will you stick with svn?

Andrew

Hi Andrew,

I think the most recent development code will always be the way to go for FFmpeg, but I may add a little section mentioning FFmpeg 0.5 for those who insist on "stable" releases. I want to keep the guide as simple as I can, but I, and Linux users in general, like to have many choices.

[cor2y]

With this modern stable release, do you think it will make it into the repos for intrepid etc?
Hopefully it will be jaunty but what about current earlier distros of ubuntu, do you think it will be added?

[FakeOutdoorsman]

With this modern stable release, do you think it will make it into the repos for intrepid etc?
Hopefully it will be jaunty but what about current earlier distros of ubuntu, do you think it will be added?

I don't know much about how Ubuntu updates or backports earlier releases. It also may not make it into Jaunty because it would require a feature freeze exception. Here's a bug report with more info:

Bug #340303: Please sync with upstream release of ffmpeg .5

[qyot27]

qyot27,

Thanks for the input. Did you try installing zlib1g-dev from the repository instead of compiling it? The package from the repository is up to date. I used to have gpac as a dependency for this guide, but I removed it since I didn't think it was necessary.

I may add a sentence about the patches, but I want to keep things simple.

My memory is rather sketchy on it, but I do seem to remember that GPAC wouldn't build against zlib1g-dev and that I had to use the package from zlib.net (basically I only make mental notes on what works rather than what doesn't, except in outstanding cases, so I could be wrong here). In general though I only include GPAC in my builds to be as holistic as possible (and because, way back - as in, two or three years - in x264's development, I remember there were problems with the MKV output, so to be able to pass correct fps and res info directly from the input file to the output, MP4 was necessary; I'm sure that it's fixed now, but old habits die hard). If there was an OS-native way to build in AVIS support on Linux I'd probably do that too. At the very least, though, piping in the video from mplayer secures that the info is retained, even with MKV. It eliminates complaints that the video isn't in sync with the audio, since MP4Box defaults to a framerate of 25fps if no value is explicitly declared.

On the patches, I'm honestly not sure what the win_zone_parse_fix patch does, but I know that the HRD/Pulldown patch is necessary for creating streams that're compliant with Blu-ray (and possibly HD-DVD, although I'm not sure). It'll be important if anyone wants to correctly author standard discs for those players.


Also, concerning checkinstall, is there a way to read the revision number from the x264 sources/binary and include that in the created package's statistics?