| --- |
| license: mit |
| pipeline_tag: image-to-video |
| tags: |
| - character-animation |
| - 3d-pose |
| - motion-transfer |
| --- |
| |
| # SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations |
|
|
|
|
| <div align="center"> |
| <a href="https://huggingface.co/papers/2512.05905"> |
| <img src="https://img.shields.io/badge/📖%20Paper-2512.05905-red"> |
| </a> |
| <a href="https://teal024.github.io/SCAIL/"> |
| <img src="https://img.shields.io/badge/🌐%20Project%20Page-green"> |
| </a> |
| <a href="https://github.com/zai-org/SCAIL"> |
| <img src="https://img.shields.io/badge/🐙%20GitHub-Repo-181717?logo=github&logoColor=white"> |
| </a> |
| </div> |
| |
|
|
| **SCAIL** (Studio-grade Character Animation via In-context Learning) is a framework designed to achieve high-fidelity character animation that meets studio standards. It addresses challenges in preserving structural fidelity and temporal consistency during motion transfer, especially in complex scenarios involving large motions and multi-character interactions. |
|
|
| Key features include: |
| - **3D-Consistent Pose Representations**: Provides a robust and flexible motion signal while preventing identity leakage. |
| - **Full-Context Pose Injection**: Enables effective spatio-temporal reasoning over entire motion sequences within a diffusion-transformer. |
| - **Studio-Grade Quality**: Trained on a curated data pipeline to ensure diversity and quality. |
|
|
| This repository contains the model weights for the **SCAIL-Preview (14B)** model. |
|
|
| ## 🔎 Project Page |
| Check the model architecture design, video demos, and comparisons against other baselines at the [official project page](https://teal024.github.io/SCAIL/). |
|
|
| ## 📋 Note |
| This repository contains the model weights for the SCAIL model. For model inference, environment setup, and the roadmap, please refer to the [official repository](https://github.com/teal024/SCAIL-Official). For pose extraction tools, refer to [SCAIL-Pose](https://github.com/teal024/SCAIL-Pose). |
|
|
| ## 📄 Citation |
|
|
| If you find this work useful in your research, please cite: |
|
|
| ```bibtex |
| @article{yan2025scail, |
| title={SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations}, |
| author={Yan, Wenhao and Ye, Sheng English and Yang, Zhuoyi and Teng, Jiayan and Dong, ZhenHui and Wen, Kairui and Gu, Xiaotao and Liu, Yong-Jin and Tang, Jie}, |
| journal={arXiv preprint arXiv:2512.05905}, |
| year={2025} |
| } |
| ``` |
|
|
| ## 👥 Authors |
| [Wenhao Yan](https://huggingface.co/wenhaoyan77), Sheng Ye, Zhuoyi Yang, [Jiayan Teng](https://huggingface.co/tengjiayan), ZhenHui Dong, [Kairui Wen](https://huggingface.co/SKearbvaanl), [Xiaotao Gu](https://huggingface.co/xgeric), Yong-Jin Liu, [Jie Tang](https://huggingface.co/jerytang). |
