← back to paper
arxiv: 2606.20233 · 2 revisions
Cinematic Compositing Using Character-Environment-Harmonized Video Generation Models