Reproduce “LoT-Pass: Long-term-robust Image Watermarking for Image to Video Generation” 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
Guanjie Wang, Zehua Ma, Han Fang, Weiming Zhang
LoT-Pass trains an image watermark against simulated video-like changes, then uses optical flow to align video frames toward a reference before decoding and voting. On CogVideoX, it reached VACC 0.966 and RDD 49, compared with TrustMark’s 0.808 and 11. This improvement is bounded: on Stable Video Diffusion, LoT-Pass had VACC 0.823 and RDD 1.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.