Mismatch between Ray-based video pre-processing pipeline and final output

Dashboard, Monitoring & Debugging
1

Hi everyone,

I’ve been experimenting with a pipeline where I use Ray to preprocess video data (batch trimming, basic transformations, and some AI-based tagging), and then bring the processed clips into capcut APK for Android for final editing.

The pipeline works fine from a processing standpoint, but I’m noticing a mismatch at the final stage:

  • Clips processed via Ray look consistent when viewed standalone

  • After importing into CapCut, some clips feel slightly different in timing and pacing

  • In a few cases, transitions between clips don’t feel as seamless as expected

  • Overall flow of the video feels less “continuous” compared to the preprocessed sequence

I’m not doing heavy transformations—mostly batching and light preprocessing—so I wasn’t expecting this kind of difference.

:red_question_mark: Question:

Is this kind of mismatch normal when combining distributed preprocessing (Ray) with consumer editing tools like CapCut?

  • Could this be related to encoding differences during batch processing?

  • Any best practices for preparing video outputs in Ray so they behave more predictably in editors?

Trying to keep the pipeline efficient without losing consistency in the final edit.

Thanks!