Motion Generalist: Multimodal Motion Generation

Zeyu Zhang, Yiran Wang, Wei Mao, Danning Li, Akira Zhao, Biao Wu, Zirui Song, Bohan Zhuang, Ian Reid, Richard Hartley

Keywords: Text-to-Motion, Music-to-Dance, Human Motion Generation

Abstract: Conditional motion generation has been extensively studied in computer vision, yet two critical challenges remain. First, while masked autoregressive methods have recently outperformed diffusion-based approaches, existing masking models lack a mechanism to prioritize dynamic frames and body parts based on given conditions. Second, existing methods for different conditioning modalities often fail to integrate multiple modalities effectively, limiting control and coherence in generated motion. To address these challenges, we propose Motion Generalist, a multimodal motion generation framework that introduces an Attention-based Mask Modeling approach, enabling fine-grained spatial and temporal control over key frames and actions. Our model adaptively encodes multimodal conditions, including text and music, improving controllability. Additionally, we introduce Text-Music-Dance (TMD), a new motion dataset consisting of 2,153 pairs of text, music, and dance, making it twice the size of AIST++, thereby filling a critical gap in the community. Extensive experiments demonstrate that Motion Generalist surpasses state-of-the-art methods across multiple benchmarks, achieving a 15% improvement in FID on HumanML3D and showing consistent performance gains on AIST++ and TMD.

Primary Area: generative models

Submission Number: 23136