PALUM: Part-based Attention Learning for Unified Motion Retargeting

TL;DR

PALUM introduces a novel part-based attention framework that effectively retargets motion across diverse skeleton structures by learning shared representations and ensuring semantic consistency, advancing animation quality.

Contribution

The paper proposes a new method that learns skeleton-agnostic motion representations using attention mechanisms and cycle consistency, enabling more accurate motion retargeting across varied skeletons.

Findings

Outperforms existing methods in diverse skeletal retargeting tasks

Maintains motion realism and semantic fidelity across unseen skeletons

Demonstrates robustness in handling large structural differences

Abstract

Retargeting motion between characters with different skeleton structures is a fundamental challenge in computer animation. When source and target characters have vastly different bone arrangements, maintaining the original motion's semantics and quality becomes increasingly difficult. We present PALUM, a novel approach that learns common motion representations across diverse skeleton topologies by partitioning joints into semantic body parts and applying attention mechanisms to capture spatio-temporal relationships. Our method transfers motion to target skeletons by leveraging these skeleton-agnostic representations alongside target-specific structural information. To ensure robust learning and preserve motion fidelity, we introduce a cycle consistency mechanism that maintains semantic coherence throughout the retargeting process. Extensive experiments demonstrate superior performance in handling diverse skeletal structures while maintaining motion realism and semantic fidelity, even when generalizing to previously unseen skeleton-motion combinations. We will make our implementation publicly available to support future research.