Reproduce “Gen4P: Video Generators are 4D Perceivers” 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

Yitong Xia, Zeren Jiang, Chuanxia Zheng

Gen4P turns a pretrained video generator into a single-pass system for recovering scene geometry and motion from monocular video. It encodes the clip once, then uses shared scene tokens to answer queries about points at locations and times. On Dynamic Replica’s dynamic points, Ours-II reaches 92.52% APD and 0.090 m EPE, versus OpenD4RT’s 82.06% and 0.185 m; results vary across benchmarks.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.