twelvelabs-io / tl-jockey

Jockey is a conversational video agent.
51 stars 13 forks source link

Enhance stability when combining clips #96

Open DmitriiTsy opened 1 week ago

DmitriiTsy commented 1 week ago

When running a large portion of clips in a request (like combining 7 clips into one), or sometimes even with simpler requests, an error can occur with the video editing functionality due to the usage of certain video and audio codecs. I believe using simpler codecs like vcodec="libx264" and acodec="aac" can improve stability. However, we should aim for a 90-100% stability rate for all user requests, so extra work on the entire clip-combining process is needed

DmitriiTsy commented 2 days ago

Previously, when trying to make a regular requests (server-side Jockey), it resulted in an ffprobe issue in most cases (around 80%) when combining clips, due to incompatible codecs or compatibility problems

Screenshot 2024-11-11 at 8 20 07 PM

My suggestion is to use these codecs for now: vcodec="libx264", acodec="aac",

And continue contributing to this PR until we achieve 99%–100% stability in most cases, making sure that Jockey is ready for any demos

Screenshot 2024-11-11 at 8 32 22 PM