跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2609.36496 · v2

Reimagine Video Dynamics: A Unified Framework for Video Editing with Editable Spatiotemporal Tokens

首次提交 2026-09-29 09:51
版本更新 2026-11-18 07:42
雷达观测 2026-11-19 19:08
arXiv ↗PDF ↗

摘要

Most video editing methods can change how a scene looks but offer limited control over how it moves. We introduce Reimagine Video Dynamics (RVD), a unified framework for editing actions, motion pace, and appearance with reusable dynamics tokens and a shared renderer. The tokens capture motion from the source video and retain an explicit temporal axis. A language-guided editor transforms them to change an action or event. Resampling them changes the pace of motion, while keeping them fixed allows the renderer to reproduce the motion with a new appearance. We train the renderer to reconstruct source videos. We pretrain the editor on motion from real videos, then fine-tune it on counterfactual pairs. Across 450 dynamics edits, RVD achieves the highest instruction-following score and an 80.0\% pass rate. The same renderer supports retiming and motion reuse without additional task-specific training.

作者备注:Project Page: https://yuyuanspace.com/RVD/

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-18 07:42 · 最新观测版本

Reimagine Video Dynamics: A Unified Framework for Video Editing with Editable Spatiotemporal Tokens

Most video editing methods can change how a scene looks but offer limited control over how it moves. We introduce Reimagine Video Dynamics (RVD), a unified framework for editing actions, motion pace, and appearance with reusable dynamics tokens and a shared renderer. The tokens capture motion from the source video and retain an explicit temporal axis. A language-guided editor transforms them to change an action or event. Resampling them changes the pace of motion, while keeping them fixed allows the renderer to reproduce the motion with a new appearance. We train the renderer to reconstruct source videos. We pretrain the editor on motion from real videos, then fine-tune it on counterfactual pairs. Across 450 dynamics edits, RVD achieves the highest instruction-following score and an 80.0\% pass rate. The same renderer supports retiming and motion reuse without additional task-specific training.

查看此版本解读 →查看此版本原文 ↗