Motion2Motion: Cross-topology Motion Transfer with Sparse Correspondence

2025-08-18

AI Summary

This work addresses the challenging problem of motion transfer across skeletons with inconsistent source/target bone topologies, where establishing one-to-one correspondences is infeasible. We propose Motion2Motion, a training-free cross-topology motion transfer framework requiring only sparse skeletal correspondences and a small set of example motions from the target skeleton. Our method integrates geometric constraints with kinematic modeling, achieving robust motion reconstruction via sparse correspondence matching and motion retargeting. Unlike existing approaches that rely on large-scale paired data or pre-trained models, ours is the first to enable generalizable, zero-shot motion transfer across species and structural classes—without assuming topological alignment. Experiments demonstrate high-fidelity animation generation on heterogeneous skeletons—including human, quadrupedal, and robotic arm configurations—and the framework has been successfully integrated into an industrial-grade animation production system.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Motion and Path Planning
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

This work studies the challenge of transfer animations between characters whose skeletal topologies differ substantially. While many techniques have advanced retargeting techniques in decades, transfer motions across diverse topologies remains less-explored. The primary obstacle lies in the inherent topological inconsistency between source and target skeletons, which restricts the establishment of straightforward one-to-one bone correspondences. Besides, the current lack of large-scale paired motion datasets spanning different topological structures severely constrains the development of data-driven approaches. To address these limitations, we introduce Motion2Motion, a novel, training-free framework. Simply yet effectively, Motion2Motion works with only one or a few example motions on the target skeleton, by accessing a sparse set of bone correspondences between the source and target skeletons. Through comprehensive qualitative and quantitative evaluations, we demonstrate that Motion2Motion achieves efficient and reliable performance in both similar-skeleton and cross-species skeleton transfer scenarios. The practical utility of our approach is further evidenced by its successful integration in downstream applications and user interfaces, highlighting its potential for industrial applications. Code and data are available at https://lhchen.top/Motion2Motion.

Problem

• Transfer animations between characters with different skeletal topologies
• Overcome topological inconsistency in source and target skeletons
• Address lack of paired motion datasets for diverse topologies

Innovation

• Training-free framework for motion transfer
• Sparse bone correspondences between skeletons
• Works with minimal example motions