CineEvolve improves cinematic multi-shot video creation with feedback loop
CineEvolve: An Agentic Self-Evolution Framework for Cinematic Multi-Shot Video Generation
Computer Vision and Pattern Recognition
Summary
Making movies with multiple scenes needs plans and careful editing to tell a good story. The authors created CineEvolve, a system that acts like a director and a critic working together by giving and receiving feedback to improve videos step by step. This system learns by trying out changes and checking what works better, much like filmmakers do. They also made a large dataset showing how videos are made, reviewed, and fixed. Their tests show CineEvolve can make better videos than previous methods.
What this means in practice
- •For video production teams: Improve automated editing systems by integrating iterative critique and revision processes to enhance multi-scene video quality.
- •For game development studios: Create more controlled and narrative-driven cinematic cutscenes using AI that evolves scenes through feedback cycles.
Authors
Jingzhe Zhang, Peixuan Zhang, Genying Li, Bing Ma, Shuchen Weng, Si Li, Boxin Shi
Abstract
Cinematic multi-shot video generation requires both narrative planning and iterative refinement. Existing methods either lack narrative reasoning or rely on feed-forward workflows without policy updates. In this paper, we propose CineEvolve, an agentic self-evolution framework for cinematic multi-shot video generation that unifies a Director and a Critic within a single vision-language policy. Emulating real-world filmmaking that couples iterative direction with critical feedback, we train CineEvolve via alternating reinforcement learning, where the Director surfaces flaws to sharpen the Critic, and the updated Critic guides the Director's targeted revisions. To stabilize co-evolution, we introduce Evolutionary GRPO (E-GRPO), leveraging historical anchoring and revision safeguarding to mitigate advantage inflation. Additionally, we contribute CineTraj, a 40K cinematic trajectory dataset spanning creation, critique, and revision. Extensive experiments demonstrate that CineEvolve outperforms existing approaches. Our code will be available at https://github.com/JingZhe-8/CineEvolve.