Paper 2512.08765
Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance
- Published
- Dec 2025
- Research lab
- Independent
- Citations
- 37
- GitHub
- 650 stars
01 In brief
Summary
Wan-Move is a framework for motion-controllable video generation that enhances existing image-to-video (I2V) models without adding auxiliary modules.
It represents object motion using dense point trajectories, which are transferred into latent space and used to replicate first-frame features along each trajectory, creating a motion-aware condition feature map.
This map is integrated directly into the base I2V model (e.g., Wan-I2V-14B) via concatenation, enabling scalable fine-tuning.
The authors also introduce MoveBench, a benchmark with 1,018 videos (5 seconds, 480p) across 54 categories, featuring hybrid human+SAM annotations for point trajectories and masks.
Experiments show Wan-Move outperforms existing methods (e.g., Tora, MagicMotion) on MoveBench and DAVIS, achieving lower EPE and higher visual quality, and matches commercial tools like Kling 1.5 Pro in user studies.
Ablations confirm the effectiveness of latent feature replication over pixel replication or random embeddings, and show robustness to trajectory numbers and large-motion scenarios.
Limitations include control loss under prolonged occlusion and potential misuse for misleading content.
The code, models, and benchmark are open-sourced.
02 From the paper
Abstract
We present Wan-Move, a simple and scalable framework that brings motion control to video generative models. Existing motion-controllable methods typically suffer from coarse control granularity and limited scalability, leaving their outputs insufficient for practical use. We narrow this gap by achieving precise and high-quality motion control. Our core idea is to directly make the original condition features motion-aware for guiding video synthesis. To this end, we first represent object motions with dense point trajectories, allowing fine-grained control over the scene. We then project these trajectories into latent space and propagate the first frame's features along each trajectory, producing an aligned spatiotemporal feature map that tells how each scene element should move. This feature map serves as the updated latent condition, which is naturally integrated into the off-the-shelf image-to-video model, e.g., Wan-I2V-14B, as motion guidance without any architecture change. It removes the need for auxiliary motion encoders and makes fine-tuning base models easily scalable. Through scaled training, Wan-Move generates 5-second, 480p videos whose motion controllability rivals Kling 1.5 Pro's commercial Motion Brush, as indicated by user studies. To support comprehensive evaluation, we further design MoveBench, a rigorously curated benchmark featuring diverse content categories and hybrid-verified annotations. It is distinguished by larger data volume, longer video durations, and high-quality motion annotations. Extensive experiments on MoveBench and the public dataset consistently show Wan-Move's superior motion quality. Code, models, and benchmark data are made publicly available.