[FFmpeg-devel] [Ffmpeg-devel] [PATCH] Support raw mpegts streams from dvr

Michael Niedermayer michaelni
Wed Jun 6 15:28:23 CEST 2007


On Tue, Jun 05, 2007 at 09:47:37PM -0000, Wolfram Gloger wrote:
> Hi,
> > > error, non monotone timestamps 297911 >=3D 9911
> > 
> > true, but old revissions fail the same way with -ss and stream copy
> Old revisions usually do not fail here, only sometimes.
> Anyway that's very probably the famous problem with initial timestamps;
> I'll get to it eventually.
> > seeking with mpeg ts doesnt work very well, it didnt in the past either
> Yes it did work well until a few days ago!
> > ive tried it with ffplay and ffmpeg and couldnt find any difference
> > between svn head and a few old binaries ive had laying around
> > 
> > unless the new version choose a different stream but that can hardly
> > be considered a bug, its just a 50:50 chance that it might work slightly
> > better (or worse) due to the stream selected
> The problem is: by eliminating the admittedly ugly "goto do_pcr" you
> have effectively disabled detection of the pcr pid unless there is a
> valid SDT.  And with pcr_pid==0 all seeking fails.
> Enabling the "mpegtsraw" demuxer in allformats.{c,h} (should I send a
> patch for that?) and specifying "-f mpegtsraw" does _not_ help,


> because then we get the pcr_pid but don't have auto_guess on.
> What I would suggest is for the auto_guess case to auto-detect the
> pcr pid as well as new streams.  Patch appended, lightly tested.

your patch is a quick hack, not a proper solution

first lets take a look at what the demuxer does, the current pcr_pid code
is completely broken and simply wont work with multi program TS streams
multi program TS streams didnt work at all until recently ...
what needs to be done, first is to change the demuxer so it does keep track
of the pcr_pid per program, then the stream used for seeking has to be mapped
to the correct pcr_pid and if that is unknown (-1 or whatever) then the code
should just use any packet to get the pcr hoping that this is not a multi
program stream

do you want to implement that? or do you have some other suggestion?

Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070606/08fff04a/attachment.pgp>

More information about the ffmpeg-devel mailing list