[FFmpeg-trac] #4088(undetermined:new): Theora video duration different depending on scale

FFmpeg trac at avcodec.org
Fri Nov 7 08:45:24 CET 2014


#4088: Theora video duration different depending on scale
-------------------------------------+-------------------------------------
             Reporter:  mlinnosk     |                    Owner:
                 Type:  defect       |                   Status:  new
             Priority:  normal       |                Component:
              Version:  2.4.3        |  undetermined
             Keywords:  theora,pipe  |               Resolution:
             Blocking:               |               Blocked By:
Analyzed by developer:  0            |  Reproduced by developer:  0
-------------------------------------+-------------------------------------

Comment (by mlinnosk):

 Replying to [comment:16 Cigaes]:
 > Replying to [comment:11 mlinnosk]:
 > > Replying to [comment:10 Cigaes]:
 > > The ffmpeg is downloaded from: http://johnvansickle.com/ffmpeg/
 > > I know it's not the Git head version, but as I understand the latest
 stable. I'd rather not compile it my self!
 >
 > I was confused by the summary at verbose level, forget I said anything.
 >
 > > The duration can be seen in 'ffmpeg -i' output or 'ogginfo' that shows
 the same.
 >
 > You wrote in the initial description "and drops frames from the end",
 but neither {{{ffmpeg -i}}} nor {{{ogginfo}}} can show that. It could be
 just a matter of the duration being badly estimated.
 >
 > It would be of course much easier if you could provide a way to
 reproduce the bug. Otherwise, you need to run all the tests yourself.

 Indeed. The reason I observed this in the first place was that the video
 ended prematurely. The material has 24 frames of the same image at the end
 and it was cur short by the player (tested mplayer and vlc).

 The fact that puzzles me is how can the length estimate be dependant on
 the resolution of the video. Offcause this can be libtheora problem, not
 ffmpeg, but ffmpeg (and it's scaling library) is the component that feeds
 the raw frame data to libtheora, right? And the encoding of the raw frame
 should be quite deterministic.

--
Ticket URL: <https://trac.ffmpeg.org/ticket/4088#comment:17>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker


More information about the FFmpeg-trac mailing list