Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations (CVPR 2026 Findings)
| Date | Stars |
|---|---|
| 2026-07-24 | 1024 |
| 2026-07-25 | 1024 |
| 2026-07-28 | 1024 |
| 2026-07-30 | 1024 |
| 2026-08-06 | 1024 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<h1>SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations</h1>
<div align="center">
<a href='https://arxiv.org/abs/2512.05905'><img src='https://img.shields.io/badge/📖 arXiv-2512.05905-red'></a>
<a href='https://huggingface.co/zai-org/SCAIL-Preview/tree/main'><img src='https://img.shields.io/badge/🤗 HuggingFace-yellow'></a>
<a href='https://modelscope.cn/models/ZhipuAI/SCAIL-Preview'><img src='https://img.shields.io/badge/🤖 ModelScope-blue'></a>
<a href='https://teal024.github.io/SCAIL/'><img src='https://img.shields.io/badge/🌐 Project Page-green'></a>
</div>
This repository contains the official implementation code of our paper accepted by *CVPR 2026 Findings Track*: **SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations**. The code is for the inference of SCAIL-Preview Model, a 14B DiT that enables challenging character animation by **In-Context Learning** of **3D-Consistent Pose Representation**.
<p align="center">
<img src='resources/teaser.png' alt='Teaser' width='90%'>
</p>
## 🔎 Motivation and Results
SCAIL identifies the key bottlenecks that hinder character animation towards production level: limited generalization towards characters and incoherent motion under complex scenarios (e.g. common failures in basic motions like flipping and turning). We revisit the core components of character animation -- how to represent the pose condition and how to inject the pose condition.
The first contribution of this paper is **3D-Consistent Pose Representation**, an identity agnostic representation that can both be aware of depth and preserve rich motion information.
<p align="center">
<img src='resources/pose_comp.png' alt='Teaser' width='60%'>
</p>
The second contribution and the core of this paper lies in **how we inject the pose condition**. Common injection methods (e.g. channel concat, pose-guider, residual layers) adds feature instead of showing the full context, which yield decent results under the setting of *controllable generation*, but the performance is limited by the pretrained backbone under wild scenarios. Taking channel concat injection under a 1.3B model as an example, it fail to maintain correct body rotation due to limited model capability in complex human motion. Instead, we show the model **full context**, not only *telling what to follow*, but also *teaching how to do*. As shown below, revealing the the full turning context with the **3D-Consistent Pose Representation** help the less capable 1.3B model learn how to generate plausible turning motion.
<p align="center">
<img src='resources/explore.png' alt='Teaser' width='50%'>
</p>
Check detailed methods, results gallery, as well as comparisons against other baselines at our [project page](https://teal024.github.io/SCAIL/).
<p align="center">
<img src='resources/1.gif' width='66%'>
<img src='resources/2.gif' width='66%'>
<img src='resources/3.gif' width='66%'>
</p>
## 🌱 Community Works
❤️ A heartfelt thanks to friends in the community for their creativity! All results below are shared with their gracious consent. We were surprised to see the emergent abilities our model exhibited — understanding the 3D spatial relationships of 2D characters, driving hand-drawn artwork, and even controlling quadrupeds despite having no animal training data at all. We believe such results work as a compelling demonstration of how **In-Context Learning** can push the upper bound of the model's capabilities.
<table align="center" border="0" cellspacing="0" cellpadding="6">
<!-- 第一行 -->
<tr>
<td align="center">
<img src="resources/community1.gif" width="220"><br>
<em>Chibi Gotham Battle</em>
</td>
<td align="center">
<img src="resources/community2.gif" width="250"><br>
<em>Homer Bullet Time (w/ Uni3c)</em>
</td>
<td align="center" rowspan="2">
<img src="resources/community4.gif" wExcerpt of 12,529 characters
Read on GitHub
94
Yuxuan Zhang · Z.ai · China
4
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:cfc22bf6cf34a8a6, topic:video-generation