On 2023-01-10 04:32 pm, Anton Khirnov wrote:
Quoting Gyan Doshi (2023-01-03 11:22:17)
At present, the offset for discontinuity adjustment is applied per-file but
the check for discontinuity is intra-stream so the same discontinuity when
seen in multiple streams with copyts, leads to compounded adjustment of the
discontinuity offset. This introduces gaps in streams, leading to loss of sync
or even de facto loss of stream.
The ts_offset_discont parameter is transferred to InputStream and adjusted
based on intra-stream gaps.
I never had much use for this feature, but I wonder if adding different
offsets to different streams isn't going against the whole point of
timestamps, which is synchronizing them.
A robust method would have to look at the inter-stream delta among first
packets in each stream after the jump,
which neither the existing code nor my changes do.
The existing code compares dts against next_dts for the stream where the
jump is first seen, and derives the offset from that.
That offset adjustment applied to all other streams may not preserve the
inter-stream deltas for the packets after the jump
since inter-frame intervals in each stream will be different. In
practice, the discrepancy is going to be minor, except for pathological
inputs.
But this patch avoids the compounded adjustment currently being applied,
which makes the output defunct, or worse frozen and
a CPU/memory hog if video encoding is CFR.
Regards,
Gyan
_______________________________________________
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".
