[Ffmpeg-devel] [BUG] no video transcoding
Sun Apr 15 16:28:28 CEST 2007
On Sun, Apr 15, 2007 at 01:44:06PM -0000, Wolfram Gloger wrote:
> > it doesnt look correct
> > 1. the cur_dts should be set to the first known dts - the previous pkt=20
> > durations (this can be tricky with I/P/B)
> I agree, it is just a workaround.. But it is better than before
> svn-8465, where cur_dts was set to some random packet's dts that
> happened to be parsed in av_find_stream_info().
> Do you think it is worth it to work more on this and backtrack the
> durations with all the ordering issues?
no, unless the resulting code is simple
actually it shouldnt be that hard
start at 0 continue until the first known timestamp then just
store the difference go back to the first packet and setcur_dtsto the
difference or something like that ...
> > 2. why does it fail at all, it shouldnt, even if the first few timestamps
> > are missing or wrong should ffmpeg.c not get stuck
> Agreed, too -- but what output timestamps should be generated in that
> case? I believe this is the same or at least a similar problem I
> reported two weeks ago with negative timestamps occurring..
i think ffmpeg.c should be changed so its able to deal with a tiny number of
totally wrong timestamps
the question what av_read_frame() should output, well i dont know
starting at a random timestamp shouldnt be differnt from a timestamp
discontinuity but it is it seems ...
rather outputting no timestamp before the first known timestamp is an
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
Concerning the gods, I have no means of knowing whether they exist or not
or of what sort they may be, because of the obscurity of the subject, and
the brevity of human life -- Protagoras
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
More information about the ffmpeg-devel