Length-varying Neural Motion Stitching via Cluster Transition Graph

2026-09-29

AI Summary

This study addresses the limitation of existing motion stitching methods, which rely on manual configuration or fixed-length transitions and struggle to achieve natural connections between significantly disparate motions. To overcome this, we propose a variable-length neural motion stitching framework based on clustered transition graphs. Specifically, our approach constructs a directed transition graph through motion clustering and searches for optimal paths to adaptively determine transition durations and guiding sequences, subsequently employing a Transformer encoder to generate smooth transitional motions. This work breaks through conventional fixed-length constraints, enabling high-fidelity, natural connections between heterogeneous actions such as crawling and basketball shooting, thereby significantly enhancing the diversity and adaptability of motion synthesis.

Abstract

Motion stitching aims to create new character animations by seamlessly combining existing motion sequences. Existing approaches often require manual selection of transition range or assume fixed transition length, restricting the types of motions that can be connected. To broaden the diversity of motions that can be synthesized, it is essential to generate transitions of varying lengths, allowing the character sufficient time to adapt its pose when the input motions differ significantly. To this end, we propose a length-varying neural motion stitching method based on a cluster transition graph, which produces naturally connected motion sequences given two distinct input motions. Our framework consists of three stages: motion clustering, cluster pathfinding, and motion generation. First, motion clustering maps input motions to discrete clusters. Next, we identify the corresponding clusters in the cluster transition graph and search for a connecting path. In this graph, nodes represent motion clusters, and directed edges indicate valid transitions between them. The resulting path determines both the transition length and a guide sequence that informs motion generation. Finally, the path and input motions are provided to a Transformer encoder-based motion generator to produce the final transition poses. Experimental results demonstrate that our method adaptively adjusts the motion length and successfully generates plausible transitions between distinct motions, such as crawling, basketball shooting, and slow locomotion. We also show that using a graph structure effectively estimates transition durations and produces high-fidelity results compared to methods that assume a fixed transition length, or directly compute the time.

Problem

• motion stitching
• variable-length transition
• character animation
• motion synthesis

Innovation

• Motion Stitching
• Cluster Transition Graph
• Variable-length Transition
• Transformer Encoder
• Motion Clustering