Skip to content

Conversation

@mjh1
Copy link
Member

@mjh1 mjh1 commented Nov 4, 2025

If we find a source segment that doesn't start with a keyframe then we redo the segmenting with a re-encode to forcefully insert keyframes.

This ensures that we have a keyframe at the beginning of every segment
@mjh1 mjh1 requested a review from emranemran November 4, 2025 15:17
@mjh1
Copy link
Member Author

mjh1 commented Nov 4, 2025

I'm considering whether to only apply this re-encode if the requested segment duration is <5s or something as we only seem to have hit this issue when the customer wanted to go to 3s.

@mjh1 mjh1 requested a review from thomshutt November 24, 2025 13:32
@mjh1
Copy link
Member Author

mjh1 commented Nov 24, 2025

@thomshutt I've changed this to only re-encode when needed as discussed. We check that the source segments start with a key-frame after the normal non-encode segmentation and retry with a re-encode if needed.

@mjh1 mjh1 changed the title Switch to an encode command for segmenting Re-encode during segmenting when needed Nov 24, 2025
Copy link
Contributor

@thomshutt thomshutt left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM - we should monitor both how often this runs and also how often we see issues because there were segments with the issue that weren't in the first 4 or last 2 segments

@mjh1 mjh1 force-pushed the mh/segmenting branch 2 times, most recently from 075223c to aab4025 Compare November 25, 2025 14:49
@mjh1 mjh1 merged commit f1738b4 into main Nov 25, 2025
14 of 15 checks passed
@mjh1 mjh1 deleted the mh/segmenting branch November 25, 2025 15:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants