[vlc-devel] issue arising from OggDirac muxing

[Laurent Aimar]

On Wed, Oct 29, 2008, David Flynn wrote:
> On 2008-10-28, Laurent Aimar <fenrir at via.ecp.fr> wrote:
> >  That is an error of the ogg muxer, it should not be done independantly
> >  The proper fix would be to use the lowest DTS over all ES IF ogg allows it.
> 
> Actually, i take that back; it does do it correctly, (I was running it without
> audio, and the rest of this post describes what that is)
> 
> The actual problem is if there is no audio, and if it were to be muxed
> in with a command line tool later (or even vlc, later).  So vlc will
> generate the following oggdirac file:
> 
> VideoDT: [0, 1, 2, 3, 4, 5 ...]
> VideoPT: [6, 1, 2, 3, 4, 5 ...]
> 
> (Note the absence of any video at time PT==0)
> 
> If you were to try to mux it with an audio file (which was supposed to
> be coincident in time):
> 
> AudioPT: [0, 1, 2, 3, 4, 5 ...]
> 
> You end up with:
> 
> VideoDT: [0, 1, 2, 3, 4, 5 ...]
> VideoPT: [6, 1, 2, 3, 4, 5 ...]
> AudioPT: [0, 1, 2, 3, 4, 5 ...]
> 
> Note that when AudioPT = 0, the corresponding correct videoPT=1.  (Since
> the -ve offset has been lost).
> 
> It might be possible to solve with OggSkeleton, but thats far too vague
> to even consider.
> 
> So, is this something VLC should care about?
 Well I don't see a solution that will not add a lot of latency in the
packetizer except by fixing all timestamps once you have them all (ie once
the file is written).

-- 
fenrir