AAMDM: Accelerated Auto-regressive Motion Diffusion Model

TL;DR

AAMDM is a novel motion synthesis framework combining diffusion models and auto-regressive techniques to produce high-quality, diverse, and efficient animations in a lower-dimensional space, outperforming existing methods.

Contribution

The paper introduces AAMDM, integrating Denoising Diffusion GANs and auto-regressive diffusion in a lower-dimensional space for improved motion synthesis.

Findings

Outperforms existing methods in motion quality and diversity

Achieves faster runtime efficiency

Demonstrates effectiveness through ablation studies

Abstract

Interactive motion synthesis is essential in creating immersive experiences in entertainment applications, such as video games and virtual reality. However, generating animations that are both high-quality and contextually responsive remains a challenge. Traditional techniques in the game industry can produce high-fidelity animations but suffer from high computational costs and poor scalability. Trained neural network models alleviate the memory and speed issues, yet fall short on generating diverse motions. Diffusion models offer diverse motion synthesis with low memory usage, but require expensive reverse diffusion processes. This paper introduces the Accelerated Auto-regressive Motion Diffusion Model (AAMDM), a novel motion synthesis framework designed to achieve quality, diversity, and efficiency all together. AAMDM integrates Denoising Diffusion GANs as a fast Generation Module, and an Auto-regressive Diffusion Model as a Polishing Module. Furthermore, AAMDM operates in a lower-dimensional embedded space rather than the full-dimensional pose space, which reduces the training complexity as well as further improves the performance. We show that AAMDM outperforms existing methods in motion quality, diversity, and runtime efficiency, through comprehensive quantitative analyses and visual comparisons. We also demonstrate the effectiveness of each algorithmic component through ablation studies.