File size: 2,747 Bytes
32e74cc 9cf7b6f 1882243 32e74cc 1882243 32e74cc e5fb5ed 1882243 e5fb5ed 32e74cc 1882243 32e74cc 1882243 32e74cc 1882243 32e74cc 1882243 32e74cc 1882243 32e74cc 1882243 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 | ---
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). |