[FFmpeg-devel] Matroska and output_ts_offset

[Andreas Rheinhardt]

Peter Zebühr:
>> On 29 Aug 2022, at 16:37, Andreas Rheinhardt <andreas.rheinhardt at outlook.com> wrote:
>>
>> The Matroska muxer is buggy wrt. to the ts_offset relating to codec
>> delay. You can see it in lines 1839-1841 which are commented out.
>> Commenting them out happened in commit
>> 82e4f39883932c1b1e5c7792a1be12dec6ab603d, merging the libav commit that
>> implemented it (namely
>> https://github.com/FFmpeg/FFmpeg/commit/a1aa37dd0b96710d4a17718198a3f56aea2040c1 <https://github.com/FFmpeg/FFmpeg/commit/a1aa37dd0b96710d4a17718198a3f56aea2040c1>).
>> It mentions "assertion failures and av sync errors". I can only think of
>> one way that it could have led to assertion failures, but this should
>> have been fixed in 4ebeab15b037a21f195696cef1f7522daf42f3ee (and since
>> then I wondered whether it can't be enabled).
>>
>> - Andreas
> 
> Interesting,
> 
> It does look like re-enabling that commented out section would indeed shift my teimstamps forward by the encoder dalay and work with my ts_offset setting. 
> I do wonder though, would that not also mean that for the cases where the stream should start at 0 it would result in streams starting 7ms in due to the handle_avoid_negative_ts already having shifted the first packets PTS from -7 to 0? And that would end up being shifted once again for the inital_padding?
> 

Yes, the shifting code in libavformat/mux.c does not yet know about the
Matroska muxer's ability to shift by initial_padding samples. But at
least it will shift all packets, thereby preserving sync in case there
are multiple streams. This is not done currently.

- Andreas