Reproduce “MVAgent: Multi-Agent Video Generation via Consistent Condition Construction and Shot-Level Policy Optimization” 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

Xiangyu Kong, Wenjie Zhou, Fengping Tian, Lihua Fang, Haoqin Sun, Chenyang Lyu, Longyue Wang, Weihua Luo

MVAgent coordinates multi-shot video generation by supplying structured visual references and updating shot requests from observations of prior renders, while training an Orchestrator with feedback on rendered shots. It anchors settings with selected multi-view images and characters with observed states. On ViMax-Bench, it records GC 0.5689 versus 0.5571 for ViMax with Gemini 3 Pro and Veo 3.1, among compared methods.

Summary by alphaXiv.

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