UniMate: One Unified Model to Animate Diverse Skeletons

Forum topic · 小凯 · 2026-09-08

Summary

UniMate is a unified foundation model for character animation that synthesizes articulated motion for arbitrary rigged 3D skeletons from a text prompt, eliminating the topology constraints of existing learned animators. While automatic rigging can now produce animation-ready 3D assets at scale, generating the motion to drive them remains a bottleneck: prior methods rely on category-specific templates, per-skeleton fine-tuning, or reference motions at inference time. UniMate requires no test-time optimization or per-skeleton retraining. Its core contribution is a topology-aware diffusion transformer that integrates skeletal topology into attention through three mechanisms: (1) a graph-aware attention bias derived from pairwise joint relations and geodesic distances, (2) a spectral rotary position embedding that generalizes RoPE to skeletal structures, and (3) additional integration details truncated in the source. The paper (arXiv:2609.05415) is authored by Linzhan Mou, Jiahui Lei, Zhiyang Dou, Chenyue Cai, Chaoyue Song, Adam Finkelstein, and Szymon Rusinkiewicz, in the computer vision field.

Paper Overview

Field: Computer Vision (CV) Authors: Linzhan Mou, Jiahui Lei, Zhiyang Dou, Chenyue Cai, Chaoyue Song, Adam Finkelstein, Szymon Rusinkiewicz Published: 2026-09-04 arXiv: 2609.05415

Abstract (Original)

Recent advances in automatic rigging now deliver animation-ready 3D assets at scale, yet generating the motion to drive them remains a bottleneck. Existing learned animators are topology-constrained: they rely on category-specific templates or require per-skeleton fine-tuning and reference motions at inference. We present UniMate, a unified foundation model that synthesizes articulated motion for arbitrary skeletons from a rigged 3D asset and a text prompt, with no test-time optimization or per-skeleton retraining. UniMate introduces a topology-aware diffusion transformer, which integrates skeletal topology into attention via three mechanisms:

1. A graph-aware attention bias from pairwise joint relations and geodesic distances; 2. A spectral rotary position embedding generalizing RoPE to skeletal structures; 3. *(Third mechanism truncated in the source post.)*

Notes

The original abstract in the source post is truncated; see the arXiv page for the full abstract and details.
Chinese translation of the abstract was not available in the source post.

*Auto-collected on 2026-09-08.*