Reproduce “Towards Practical 20 kbps Generative Video Compression via Adaptive Tokenization and Entropy Modeling” 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
Jiancheng Zhao, Jixiang Luo, Haibin Huang, Chi Zhang, Xuelong Li
H3VC turns a pretrained video model’s continuous representations into discrete symbols and entropy-codes them, while adapting token assignments to reduce coding cost. Across UVG, MCL-JCV, and HEVC Class B, it reports average bitrate savings versus GLC-Video of 84.1% at matched LPIPS and 82.3% at matched DISTS. These are perceptual rate–distortion results on the tested sequences, not a universal quality or speed claim.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.