Captain Cinema: Towards Short Movie Generation
Junfei Xiao, Ceyuan Yang, Lvmin Zhang, Shengqu Cai, Yang Zhao, Yuwei Guo, Gordon Wetzstein, Maneesh Agrawala, Alan Yuille, Lu Jiang
2025-07-25
Summary
This paper talks about Captain Cinema, a new AI system that can create short movies from detailed text descriptions by planning key scenes first and then generating the video between those scenes.
What's the problem?
Creating long and coherent movies using AI is hard because it needs to keep the story and visuals consistent across many scenes, which is tough for existing methods.
What's the solution?
The researchers designed Captain Cinema to first generate a series of keyframes that outline the main moments of the movie, then use these keyframes to guide a video generation model that fills in the movements and transitions while keeping everything coherent and visually consistent.
Why it matters?
This matters because it brings us closer to automating movie production with AI, making it easier and faster to generate creative video content without the need for human filmmakers.
Abstract
Captain Cinema generates coherent short movies from textual descriptions using top-down keyframe planning and bottom-up video synthesis with Multimodal Diffusion Transformers.