Hacker News new | past | comments | ask | show | jobs | submit login

Looks like applying the motion vectors from one video to the starting conditions from another, which sometimes happens when fast forwarding with a buggy video decoder or a glitchy video.

I wonder, could you do this just by chopping I frames from one video with B and P frames from another?

That (removing I-frames) is exactly what they were doing - the technique's called datamoshing.

Applications are open for YC Summer 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact