EchoMimicV2: Towards Striking, Simplified, and Semi-Body Human Animation

TL;DR

EchoMimicV2 tackles the challenge of high-quality half-body animation with simplified conditioning. It introduces Audio-Pose Dynamic Harmonization to balance audio and pose cues, employs Pose Sampling and Audio Diffusion, and uses Head Partial Attention to enrich facial expressiveness from headshot data. A Phase-specific Denoising Loss guides training across pose, detail, and low-level quality, while a new EMTD benchmark enables standardized evaluation. Extensive experiments demonstrate state-of-the-art performance and practical potential for audio-driven half-body animation with reduced control complexity.

Abstract

Recent work on human animation usually involves audio, pose, or movement maps conditions, thereby achieves vivid animation quality. However, these methods often face practical challenges due to extra control conditions, cumbersome condition injection modules, or limitation to head region driving. Hence, we ask if it is possible to achieve striking half-body human animation while simplifying unnecessary conditions. To this end, we propose a half-body human animation method, dubbed EchoMimicV2, that leverages a novel Audio-Pose Dynamic Harmonization strategy, including Pose Sampling and Audio Diffusion, to enhance half-body details, facial and gestural expressiveness, and meanwhile reduce conditions redundancy. To compensate for the scarcity of half-body data, we utilize Head Partial Attention to seamlessly accommodate headshot data into our training framework, which can be omitted during inference, providing a free lunch for animation. Furthermore, we design the Phase-specific Denoising Loss to guide motion, detail, and low-level quality for animation in specific phases, respectively. Besides, we also present a novel benchmark for evaluating the effectiveness of half-body human animation. Extensive experiments and analyses demonstrate that EchoMimicV2 surpasses existing methods in both quantitative and qualitative evaluations.

Figures (6)

• Figure 1: EchoMimicV2 utilizes a reference image, an audio clip, and a sequence of hand pose to generate a high-quality animation video, ensuring coherence between audio content and half-body movements.
• Figure 2: The overall pipeline of our proposed EchoMimicV2.
• Figure 3: The results of EchoMimicV2 given different reference images, hand pose and audios.
• Figure 4: The results of EchoMimicV2 compared to pose-driven half-body human animation baselines.
• Figure 5: The results of EchoMimicV2 compared to audio-driven half-body human animation baselines.