To elaborate slightly: GENPTS is working within each segment, whose start and end times I have from NAV packets
On Wed, Jan 10, 2024 at 10:25 Marth64 <mart...@proxyid.net> wrote: > Thank you, Nicolas. I had tinkered with concat this way in the past as > PGCs are indeed just a bunch of segments mushed together, but I had run > into issues specifically with (1) concerns about forced subtitles that pop > up mid stream and (2) unable to figure out custom IO with it. Since dvdnav > is driving the navigation and reading of the disc, I do not have the luxury > of knowing VOB addresses for a perfect read of the title (without fully > playing back). So I don’t have a list of files and sectors to give concat. > As DVD also has cell commands that drive direction of the stream, this > complicates it further. The structure revealed by dvdread alone is not > enough to properly play back; the dvdnav VM plays a key role. Finally, I > had also worry about oddities when having demuxer->submuxer->subsubdemuxer. > > In my approach (using mpeg demux directly) there doesn’t seem to be any > issue with concatenation/handling the gaps, and I am flushing the > subdemuxer when they occur. Just the GENPTS sticks out as being a non-ideal > thing to do. > > That said, I will take another look and see if I misunderstood at the time. > > Best, > > On Wed, Jan 10, 2024 at 04:17 Nicolas George <geo...@nsup.org> wrote: > >> Marth64 (12024-01-10): >> > I will add that I still had to maintain a dependency on GENPTS for the >> > inner MPEG demuxer. This is because dvdnav_get_current_time() does not >> > provide exact stream-level PTS. However, the crashing/hanging is fixed >> (by >> > parsing NAV packets), and the output is produced smoothly. GENPTS fills >> in >> > the blanks. >> > >> > If anyone can think of a better way, please let me know. Right now, it >> > works well. >> >> You could imitate what tools/dvd2concat does internally: use >> libdvdread to get the structure of the DVD but let the concat demuxer >> handle all the actual demuxing work. >> >> Regards, >> >> -- >> Nicolas George >> > _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".