AnimateAnything: Consistent and Controllable Animation for Video Generation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912122519158784 |
|---|---|
| author | Lei, Guojun Wang, Chi Li, Hong Zhang, Rong Wang, Yikai Xu, Weiwei |
| author_facet | Lei, Guojun Wang, Chi Li, Hong Zhang, Rong Wang, Yikai Xu, Weiwei |
| contents | We present a unified controllable video generation approach AnimateAnything that facilitates precise and consistent video manipulation across various conditions, including camera trajectories, text prompts, and user motion annotations. Specifically, we carefully design a multi-scale control feature fusion network to construct a common motion representation for different conditions. It explicitly converts all control information into frame-by-frame optical flows. Then we incorporate the optical flows as motion priors to guide final video generation. In addition, to reduce the flickering issues caused by large-scale motion, we propose a frequency-based stabilization module. It can enhance temporal coherence by ensuring the video's frequency domain consistency. Experiments demonstrate that our method outperforms the state-of-the-art approaches. For more details and videos, please refer to the webpage: https://yu-shaonian.github.io/Animate_Anything/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_10836 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | AnimateAnything: Consistent and Controllable Animation for Video Generation Lei, Guojun Wang, Chi Li, Hong Zhang, Rong Wang, Yikai Xu, Weiwei Computer Vision and Pattern Recognition We present a unified controllable video generation approach AnimateAnything that facilitates precise and consistent video manipulation across various conditions, including camera trajectories, text prompts, and user motion annotations. Specifically, we carefully design a multi-scale control feature fusion network to construct a common motion representation for different conditions. It explicitly converts all control information into frame-by-frame optical flows. Then we incorporate the optical flows as motion priors to guide final video generation. In addition, to reduce the flickering issues caused by large-scale motion, we propose a frequency-based stabilization module. It can enhance temporal coherence by ensuring the video's frequency domain consistency. Experiments demonstrate that our method outperforms the state-of-the-art approaches. For more details and videos, please refer to the webpage: https://yu-shaonian.github.io/Animate_Anything/. |
| title | AnimateAnything: Consistent and Controllable Animation for Video Generation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2411.10836 |