Reproduce “MovieRoll: A Systematic Framework for Rolling Out Long-Form Videos” 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
Yunyang Ge, Weijie Kong, Ryan Xu, Yue Wu, Qi Tian, Jiangfeng Xiong, Bin Lin, Mingzhe Zheng and 4 more
MovieRoll generates continuous videos across prompted events by jointly denoising neighboring chunks and selectively retaining historical attention states. It trains a multi-step base with windowed training, then distills it into a five-step generator. On the 60-second, six-event HeliosBench task, MovieRoll-Distilled scores 6.492 versus 5.880 for Helios-Base; MovieRoll-Base was not evaluated at this duration, and generation remains below real-time playback.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.