[FFmpeg-user] I found the bugs

[Mark Filipak]

On 19/06/2024 10.01, Rob Hallam wrote:
> On Wed, 19 Jun 2024 at 14:23, Mark Filipak <markfilipak.imdb at gmail.com> wrote:
> 
>> I think this may be a far ranging bug that affects '-ss' and that provokes many of the
>> "non-monotonous DTS" error messages that seem to appear out of nowhere when a trivial, non-timing,
>> non-timestamp change is made to a transcode or to a remux. I confess that I write "I think" rather
>> than "We think" because I've not yet gotten others on ffmpeg-user to look at Ticket_11055.m2ts and
>> to take this issue seriously enough to join me. Certainly, no one else on ffmpeg-user has
>> communicated that they've tested (or even watched) Ticket_11055.m2ts.
> 
> I downloaded and played Ticket_11055.m2ts with vlc and mpv (which I
> installed to perform this test).

Thank you, Rob.

Yes, VLC (and PowerDVD, too) play Ticket_11055.m2ts correctly. Note that neither of them get 
timestamps from FFmpeg.

MPV is more powerful and I think you'll like it. When it loads, quickly press the [L] key 
(unshifted) to loop, then press [Shift][O] to show running times. Note that running time pauses at 
0:00.825, then 0:00.950, then resumes around 0:02.952. MPV gets timestamps from FFmpeg [note1].

mpv Ticket_11055.m2ts --no-correct-pts --fps=24000/1001
produces different behavior that implies audio timestamps are also a problem.

[note1] mpv.exe contains (in whole or in part) sections of ffmpeg "libavutil libavcodec libavformat 
libswscale libavfilter libswresample". Some additional strings within mpv.exe are
"Seek failed (%s) Leaking %d nested connections (FFmpeg bug)."
"This format is marked by FFmpeg as having no timestamps! FFmpeg will likely make up its own broken 
timestamps. For video streams you can correct this with: --no-correct-pts --fps=VALUE, with VALUE 
being the real framerate of the stream. You can expect seeking and buffering estimation to be 
generally broken as well."
Note that "seek" is what ffmpeg calls '-ss'.

Based on the comment strings shown above, I'd say that FFmpeg has had the timestamp bug for a long 
time and the MPV developers have known it.

> Results appended below message, please let me know if you want this
> posted to the bug report on trac.

If this bug is important to you, Rob, it deserves more thought, investigation, supporting 
documentation -- and more ideas! I hope this issue will get more participation, especially from Paul.