Reproduce “LynnReal-Hand: Unlocking World Interaction with Hand 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
Shaohao Rui, Chaoyang Zhang, Xiaofeng Mao, Weijie Ma
LynnReal-Hand generates hand-controlled interaction videos from hand motion and camera movement. Its Interaction-Preserving Training gives hands dedicated supervision and separates hand-aware video guidance from background-only image guidance. On in-domain HandBench, Base reaches FVD 181.67 and MANO Dice 93.02, versus 294.32 and 90.60. The three-step Fast model generates 17-frame chunks at 540p, but uses a local two-chunk context without explicit long-term 3D memory.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.