---
title: "Automated Movie Generation via Multi-Agent CoT Planning"
canonical_url: "https://www.modelscope.cn/papers/124513"
md_url: "https://www.modelscope.cn/papers/124513.md"
arxiv_id: 2503.07314
published: 2025-03-10
last_updated: 2025-03-10
authors:
  - "Weijia Wu"
  - "Zeyu Zhu"
  - "Mike Zheng Shou"
model_name: MovieAgent
model_developer: "National University of Singapore"
domain:
  - "计算机视觉"
  - "视频生成"
  - "多智能体系统"
  - "长视频生成"
  - "自然语言处理"
type:
  - "计算机视觉"
  - "视频生成"
  - "多智能体系统"
  - "长视频生成"
  - "自然语言处理"
  - "Computer Vision and Pattern Recognition"
arxiv_url: "https://arxiv.org/abs/2503.07314"
pdf_url: "https://arxiv.org/pdf/2503.07314"
code_link: "https://github.com/showlab/MovieAgent"
---

# Automated Movie Generation via Multi-Agent CoT Planning

> Existing long-form video generation frameworks lack automated planning, requiring manual input for storylines, scenes, cinematography, and character interactions, resulting in high costs and inefficiencies. To address these challenges, we present MovieAgent,…

「Automated Movie Generation via Multi-Agent CoT Planning」是 ModelScope 魔搭社区收录的论文，arXiv 2503.07314，作者为 Weijia Wu, Zeyu Zhu, Mike Zheng Shou，发表于 2025-03-10，属于 计算机视觉、视频生成、多智能体系统 领域。

- **ArXiv**: 2503.07314
- **Published**: 2025-03-10
- **Authors**: Weijia Wu, Zeyu Zhu, Mike Zheng Shou
- **Model**: MovieAgent
- **Developer**: National University of Singapore
- **Domain**: 计算机视觉, 视频生成, 多智能体系统, 长视频生成, 自然语言处理
- **ArXiv URL**: https://arxiv.org/abs/2503.07314
- **PDF**: https://arxiv.org/pdf/2503.07314
- **Code**: https://github.com/showlab/MovieAgent

Source: https://www.modelscope.cn/papers/124513

---

> MovieAgent：多智能体链式思维驱动的自动化电影生成新突破

## 摘要

本文提出了一种名为MovieAgent的自动化电影生成框架，通过多智能体链式思维（CoT）规划来实现长视频自动生成。现有的长视频生成框架缺乏自动化的规划过程，需要人工输入故事情节、场景、摄影和角色互动，导致高成本和低效率。MovieAgent通过引入多智能体系统模拟导演、编剧、分镜师和制片人的角色，实现了多层次的CoT推理过程，从而显著减少了人力投入，并确保了叙事连贯性、角色一致性、字幕同步和音频稳定。

具体而言，MovieAgent的主要优势包括：1) 首次探索并定义了自动化电影/长视频生成的范式；2) 引入了基于CoT的层次化推理过程，能够自动生成场景结构、摄像设置和摄影参数；3) 利用多个LLM智能体模拟不同电影制作专业人员的角色，简化了生产流程。实验表明，MovieAgent在剧本忠实度、角色一致性和叙事连贯性方面达到了新的最先进水平，为完全自动化的电影生成提供了新的见解。

该框架不仅大幅降低了电影制作的成本和时间，还为未来的自动化影视创作提供了新的思路和技术支持。

## Abstract

Existing long-form video generation frameworks lack automated planning, requiring manual input for storylines, scenes, cinematography, and character interactions, resulting in high costs and inefficiencies. To address these challenges, we present MovieAgent, an automated movie generation via multi-agent Chain of Thought (CoT) planning. MovieAgent offers two key advantages: 1) We firstly explore and define the paradigm of automated movie/long-video generation. Given a script and character bank, our MovieAgent can generates multi-scene, multi-shot long-form videos with a coherent narrative, while ensuring character consistency, synchronized subtitles, and stable audio throughout the film. 2) MovieAgent introduces a hierarchical CoT-based reasoning process to automatically structure scenes, camera settings, and cinematography, significantly reducing human effort. By employing multiple LLM agents to simulate the roles of a director, screenwriter, storyboard artist, and location manager, MovieAgent streamlines the production pipeline. Experiments demonstrate that MovieAgent achieves new state-of-the-art results in script faithfulness, character consistency, and narrative coherence. Our hierarchical framework takes a step forward and provides new insights into fully automated movie generation. The code and project website are available at: https://github.com/showlab/MovieAgent and https://weijiawu.github.io/MovieAgent.