SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning

2026-06-09

AI Summary

This work proposes the first end-to-end character animation framework that eliminates reliance on intermediate representations such as pose skeletons, which often introduce information loss. By directly incorporating driving videos into the input sequence and employing decoupled context-aware masking conditions, the method unifies multiple animation subtasks and achieves high-quality motion transfer without intermediate representations. Key innovations include the construction of MotionPair-60K, a large-scale synthetic dataset; the design of motion-modality-specific RoPE positional encoding; and the introduction of a Bias-Aware DPO preference optimization mechanism. Extensive experiments demonstrate that the proposed approach significantly outperforms state-of-the-art methods across various character animation benchmarks. The authors also release a portion of the dataset and model weights to facilitate future research.

Abstract

Controlled character animation requires transferring motion from a driving sequence to a reference character. Prior works heavily rely on intermediate representations, including pose skeletons to represent motion or masked background to represent environment, which inevitably leads to information loss. To address this, we present SCAIL-2, an framework that bypasses those intermediates and achieves \textbf{end-to-end} character animation. By directly concatenating driving videos to the sequence, the model can obtain all the required visual information from the input video. To address lack of end-to-end data, we unify sub-tasks of character animation with decoupled conditions and then curate a pipeline to synthesize MotionPair-60K, an end-to-end motion transfer dataset containing heterogeneous tasks of character animation. To archive the unification, we utilize in-context mask conditioning and mode-specific RoPE as soft guidance beyond textual instructions and raw visual information. To address synthetic discrepancy in detailed regions, we propose Bias-Aware DPO to construct preference items to mitigate the errors. Extensive experiments demonstrate that our method substantially outperforms existing state-of-the-art approaches in various character animation tasks. A large subset of synthetic data as well as model weights will be released at our project page: https://teal024.github.io/SCAIL-2/.

Problem

• controlled character animation
• end-to-end learning
• motion transfer
• intermediate representations
• visual information loss

Innovation

• end-to-end animation
• in-context conditioning
• motion transfer
• synthetic dataset
• Bias-Aware DPO