Clips2Story：免训练、基于多模态检索嵌入生成的视频分镜与编辑
Clips2Story: Training-Free Video Storyboarding and Editing Using Multimodal Retrieval-Embedded Generation
AI 导读
研究者提出 Clips2Story，一个免训练、基于 LLM 的视频分镜与编辑流水线：对片段池提取多模态文本描述并按用户关键词检索相关片段，再由 LLM 排列成可编辑分镜并生成串联旁白，最终合成输出视频。评测数据集由五类题材共 100 个长视频切分打乱构建。人类评测显示其在叙事连贯性、关键词贴合度和整体真实感上优于强基线。
正文 · 原文
Abstract:Video editors often work with large collections of fragmented footage and must select and organize relevant clips into a coherent story. However, many existing video editing and summarization methods assume a temporally ordered source video and aim to preserve its original narrative and temporal structure. In this paper, we introduce Clips2Story, a training-free, LLM-based pipeline that enables users to create keyword-specified short videos from a pool of video clips. Given a pool of clips and a user-specified keyword, we first extract text-based multimodal descriptors for each clip and retrieve the clips most relevant to the keyword. We then provide the textual descriptors of these candidate clips to an LLM, which selects and arranges them into an editable storyboard with connective narrations. Finally, we assemble the selected clips and retrieve accompanying visuals for the narrations to compose the output video. To evaluate the proposed pipeline, we construct a synthetic evaluation dataset from 100 long-form videos across five genres by segmenting each video into short clips using a scene-change detection model and shuffling the resulting clips to remove their temporal order. Through objective evaluation and qualitative examples, we demonstrate that Clips2Story can effectively retrieve keyword-relevant clips and compose coherent storyboards based on user-specified keywords. Human evaluation further shows that Clips2Story outperforms a strong baseline in terms of narrative coherence, keyword alignment, and overall realism.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM) |
| Cite as: | arXiv:2611.05992 [cs.CV] |
| (or arXiv:2611.05992v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.05992 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Junnuo Wang [view email]
[v1]
Wed, 4 Nov 2026 05:37:16 UTC (4,756 KB)
来源：arXiv cs.AI · arxiv.org