StreamAvatar: Streaming Diffusion Models for Real-Time Interactive Human Avatars

TL;DR

StreamAvatar introduces a real-time, streaming human avatar system using diffusion models, overcoming non-causal architecture and high computational costs for interactive applications.

Contribution

It presents a novel two-stage autoregressive framework with stability components, enabling high-fidelity, real-time, interactive human avatar generation with natural gestures.

Findings

Achieves state-of-the-art quality in avatar generation

Operates efficiently in real-time for interactive use

Produces natural talking and listening behaviors with coherent gestures

Abstract

Real-time, streaming interactive avatars represent a critical yet challenging goal in digital human research. Although diffusion-based human avatar generation methods achieve remarkable success, their non-causal architecture and high computational costs make them unsuitable for streaming. Moreover, existing interactive approaches are typically restricted to the head-and-shoulder region, limiting their ability to produce gestures and body motions. To address these challenges, we propose a two-stage autoregressive adaptation and acceleration framework that applies autoregressive distillation and adversarial refinement to adapt a high-fidelity human video diffusion model for real-time, interactive streaming. To ensure long-term stability and consistency, we introduce three key components: a Reference Sink, a Reference-Anchored Positional Re-encoding (RAPR) strategy, and a Consistency-Aware Discriminator. Building on this framework, we develop a one-shot, interactive, human avatar model capable of generating both natural talking and listening behaviors with coherent gestures. Extensive experiments demonstrate that our method achieves state-of-the-art performance, surpassing existing approaches in generation quality, real-time efficiency, and interaction naturalness. Project page: https://streamavatar.github.io .