Reproduce “PISCO: Precise Video Instance Insertion with Sparse Control” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiangbo Gao, Renjie Li, Xinghao Chen, Yuheng Wu, Suofei Feng, Qing Yin, Zhengzhong Tu
PISCO performs video instance insertion by propagating object cutouts and masks supplied at selected frames through a moving shot. It uses a video diffusion model with depth and availability conditioning. On PISCO-Bench, PISCO-14B with first-and-last-frame guidance reports whole-video FVD 204 and LPIPS 0.097, versus 371 and 0.103 for VACE 14B, which receives full mask video; comparison uses different inputs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.