Preacher: Paper-to-Video Agentic System
Jingwei Liu1,2111Work done during an internship at DAMO Academy. Ling Yang6222Contributed equally. Ling Yang, Hao Luo2,3 Fan Wang2 Hongyan Li1,4,5 333Corresponding Authors: Hongyan Li, Mengdi Wang Mengdi Wang6 333Corresponding Authors: Hongyan Li, Mengdi Wang 1School of Intelligence Science and Technology, Peking University 2DAMO Academy, Alibaba group, 310023, Hangzhou, China 3Hupan Lab, 310023, Hangzhou, China 4 National Key Laboratory of General Artificial Intelligence, Peking University 5 PKU-Wuhan Institude of Artificial Intelligence 6 Department of Electrical and Computer Engineering, Princeton University
Abstract
The paper-to-video task converts a research paper into a structured video abstract, distilling key concepts, methods, and conclusions into an accessible, well-organized format. While state-of-the-art video generation models demonstrate potential, they are constrained by limited context windows, rigid video duration constraints, limited stylistic diversity, and an inability to represent domain-specific knowledge. To address these limitations, we introduce Preacher, the first paper-to-video agentic system. Preacher employs a top-down approach to decompose, summarize, and reformulate the paper, followed by bottom-up video generation, synthesizing diverse video segments into a coherent abstract. To align cross-modal representations, we define key scenes and introduce a Progressive Chain of Thought (P-CoT) for granular, iterative planning. Preacher successfully generates high-quality video abstracts across five research fields, demonstrating expertise beyond current video generation models. Code will be released at: https://github.com/Gen-Verse/Paper2Video
中文速览
每年全球新发表超过三百万篇科学论文，把论文自动转化成视频摘要可以大幅降低传播成本、提升引用率，但现有视频生成模型受限于上下文窗口短、风格单一、无法处理学术图表等问题，难以直接胜任这项任务。为此，研究者提出了 Preacher——首个将学术论文自动转换为视频摘要的智能体系统（paper-to-video agentic system），其核心思路是"自顶向下"先将论文分解、提炼为结构化的"关键场景（key scenes）"作为跨模态桥梁，再"自底向上"将各场景交由风格各异的生成工具（包括文本转视频、Python可视化、说话人头像等）合成片段并拼装成完整视频，同时引入渐进式思维链（Progressive Chain of Thought，P-CoT）来解决大模型在长上下文和细粒度规划中的性能退化问题。在五个研究领域的论文上进行测试，Preacher 在准确性、专业性、美观度和内容对齐等多个维度均超越现有最强基线。这项工作的意义在于，它提供了一套端到端的低成本替代方案，让研究人员无需专业视频制作技能也能快速生成高质量视频摘要，有望显著加速科学知识的大众化传播。
原文 arXiv:2508.09632;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2508.09632v6