DATAM🟢SH

Compressed video uses I-frames (full reference pictures) and P-frames (motion predictions that say "move these pixels from the previous frame over there").

Datamoshing removes the I-frame at a scene change. Without the fresh reference, the decoder applies the new scene's motion vectors to the old scene's pixels — causing them to smear, melt, and warp in the shape of the new motion.

This tool simulates that process: it holds a frame buffer, estimates real motion vectors between consecutive frames using block-matching, then displaces the held pixels accordingly. Press H to manually lock the I-frame hold, or let auto-detection trigger on scene changes.