Mode Seeking meets Mean Seeking for Fast Long Video Generation
2026-02-27 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors developed a new way to generate long videos by combining two approaches: one that focuses on small, detailed parts of the video and another that looks at the overall story and flow. They train a model to keep short segments realistic while also maintaining the bigger picture using limited long video examples. This method helps create longer videos that are both clear and consistent in motion and story, even when there isn’t much long video data available.
video generationDiffusion Transformerflow matchingdistribution matchingmode-seekingreverse-KL divergencelong-range coherencelocal fidelitysupervised learningsliding window
Authors
Shengqu Cai, Weili Nie, Chao Liu, Julius Berner, Lvmin Zhang, Nanye Ma, Hansheng Chen, Maneesh Agrawala, Leonidas Guibas, Gordon Wetzstein, Arash Vahdat
Abstract
Scaling video generation from seconds to minutes faces a critical bottleneck: while short-video data is abundant and high-fidelity, coherent long-form data is scarce and limited to narrow domains. To address this, we propose a training paradigm where Mode Seeking meets Mean Seeking, decoupling local fidelity from long-term coherence based on a unified representation via a Decoupled Diffusion Transformer. Our approach utilizes a global Flow Matching head trained via supervised learning on long videos to capture narrative structure, while simultaneously employing a local Distribution Matching head that aligns sliding windows to a frozen short-video teacher via a mode-seeking reverse-KL divergence. This strategy enables the synthesis of minute-scale videos that learns long-range coherence and motions from limited long videos via supervised flow matching, while inheriting local realism by aligning every sliding-window segment of the student to a frozen short-video teacher, resulting in a few-step fast long video generator. Evaluations show that our method effectively closes the fidelity-horizon gap by jointly improving local sharpness, motion and long-range consistency. Project website: https://primecai.github.io/mmm/.