In-Context Learning with Unpaired Clips for Instruction-based Video Editing
Published in arXiv preprint, 2025
We propose a cost-effective pretraining method that leverages unpaired video clips for instruction-based video editing. The approach yields a 12% improvement in editing instruction following and a 15% improvement in editing quality.
Recommended citation: Xinyao Liao, Xianfang Zeng, Ziye Song, Zhoujie Fu, Gang Yu, Guosheng Lin. (2025). "In-Context Learning with Unpaired Clips for Instruction-based Video Editing." arXiv:2510.14648.
Download Paper