Yu Yuan, Yawen Lu, Guoxian Song, Kevin Duarte, Ratheesh Kalarot, Di Chang, Xijun Wang, Stanley H. Chan
原文摘要
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.
Diffusion Transformers have demonstrated remarkable capabilities in visual synthesis, yet they often struggle with high-level semantic reasoning and long-horizon planning. This limitation frequently leads to visual hallucinations and mis-alignments with user instructions, especially in scenarios involving complex scene understanding, human-object interactions, multi-stage actions, and in-context motion reasoning. To address these challenges, we propose Plan-X, a framework that explicitly enforces high-level semantic planning to instruct video generation process. At its core lies a Semantic Planner, a learnable multimodal language model that reasons over the user's intent from both text prompts and visual context, and autoregressively generates a sequence of text-grounded spatio-temporal semantic tokens. These semantic tokens, complementary to high-level text prompt guidance, serve as structured "semantic sketches" over time for the video diffusion model, which has its strength at synthesizing high-fidelity visual details. Plan-X effectively integrates the strength of language models in multimodal in-context reasoning and planning, together with the strength of diffusion models in photorealistic video synthesis. Extensive experiments demonstrate that our framework substantially reduces visual hallucinations and enables fine-grained, instruction-aligned video generation consistent with multimodal context.