[MPlayer-users] Re: [MEncoder] Synchonization lost when using Vorbis in a OGM

[D Richard Felker III]

On Tue, Jun 17, 2003 at 11:01:12PM -0500, Jonathan Rogers wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> D Richard Felker III wrote:
> >Well OGM is not standard by any means, it's a hack from OggDS. And
> >since WMP also fails to play the file (using OggDS), I don't think we
> >can really call this a bug in mplayer -- it's mimicing what OggDS
> >does. :))
> >
> >My guess is that ogmmerge outputs something bogus when it gets a
> >0-length frame from the avi file. Probably a 0-length ogg packet. A
> >better approach, since ogm has timestamps, would be to just omit
> >output on 0-length input frames...
> 
> Yeah, in the absence of a standard it is difficult to define what is and 
> isn't a bug. I guess Mosu would be the one to ask. I've also thought it 
> would be nice if the framerate in a movie could vary, so there would 
> never have to be the duplicated or dropped frame nonsense. I know MPEG-4 
> video is supposed to be variable framerate. I think I also read that it 
> should be possible with Matroska.

It's a matter of the container format. ASF, QT, OGM, Matroska, and of
course our own NUT support variable framerate just fine. Actually AVI
can too if you set the framerate in the header to the least common
multiple (lcm) of the framerates you'll be using, and insert the
appropriate duplicate frames (which take essentially no space), but
it's not optimal.

When there's an mencoder with G2, it will keep timestamps all the way
to the end of the video filter chain (rather than doing timing just at
the beginning) so filters like inverse telecine can change timestamps,
and then do variable framerate output or frame dropping depending on
the muxer being used.

Rich