SMooDi: Stylized Motion Diffusion Model

TL;DR

SMooDi is a new diffusion-based model that efficiently generates stylized motion from text and style references, outperforming previous methods in quality and diversity.

Contribution

The paper introduces SMooDi, a novel diffusion model tailored for stylized motion generation using content and style inputs, with a new style guidance mechanism and lightweight style adaptor.

Findings

Outperforms existing stylized motion generation methods

Generates diverse and realistic stylized motions

Effective across various applications

Abstract

We introduce a novel Stylized Motion Diffusion model, dubbed SMooDi, to generate stylized motion driven by content texts and style motion sequences. Unlike existing methods that either generate motion of various content or transfer style from one sequence to another, SMooDi can rapidly generate motion across a broad range of content and diverse styles. To this end, we tailor a pre-trained text-to-motion model for stylization. Specifically, we propose style guidance to ensure that the generated motion closely matches the reference style, alongside a lightweight style adaptor that directs the motion towards the desired style while ensuring realism. Experiments across various applications demonstrate that our proposed framework outperforms existing methods in stylized motion generation.