Summary
MoRight (arXiv 2504.06855) is a unified framework for controllable video generation that addresses two key limitations of existing methods. First, it provides decoupled motion control, allowing users to independently adjust object motion and camera viewpoint. Second, it achieves motion causality, ensuring that user-driven actions trigger coherent reactions from other objects in the generated scene. Prior approaches fall short on both fronts. The authors—Shaowei Liu, Xuanchi Ren, and Tianchang Shen—introduce decoupled motion modeling as the core mechanism, enabling fine-grained user control over generated video dynamics. Experiments across three benchmarks show that MoRight achieves state-of-the-art performance in generation quality, motion controllability, and interaction awareness. The work was published in April 2025 in the computer vision category.
Paper Overview
Research Area: cs.CV
Authors: Shaowei Liu, Xuanchi Ren, Tianchang Shen
Published: 2025-04-09
arXiv: 2504.06855
Abstract (Translated)
Generating motion-controlled videos requires two capabilities: (1) decoupled motion control, which allows users to separately control object motion and adjust the camera viewpoint; and (2) motion causality, which ensures that user-driven actions trigger coherent reactions from other objects in the scene. Existing methods fall short on both fronts.
We introduce MoRight, a unified framework that addresses both limitations through decoupled motion modeling. Experiments on three benchmarks demonstrate that the proposed method achieves state-of-the-art performance in generation quality, motion controllability, and interaction awareness.
Links
- arXiv paper: https://arxiv.org/abs/2504.06855
---
*Automatically collected on 2025-04-10.*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177169716