X-UniMotion: Animating Human Images with Expressive, Unified and Identity-Agnostic Motion Latents

2025-08-12

AI Summary

This work addresses the challenge of learning a unified implicit representation for full-body human motion—including facial expressions, pose, and gestures—without relying on explicit skeletal annotations or heuristic cross-identity adaptations. We propose an identity-agnostic, quadruply disentangled latent motion representation, learned within a self-supervised end-to-end framework. Motion is decomposed into semantically disentangled latent tokens; the architecture integrates a DiT-based video generative model, 2D spatial and color augmentations, and synthetically rendered cross-identity 3D motion data. An auxiliary decoder is introduced to explicitly supervise token learning. To our knowledge, this is the first approach enabling fine-grained, full-body, skeleton-free unified motion representation learning. Evaluated on large-scale motion datasets, it achieves state-of-the-art performance, significantly improving motion fidelity and identity consistency in cross-identity animation generation.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Representation Learning
• Natural Language Processing: Generation

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

We present X-UniMotion, a unified and expressive implicit latent representation for whole-body human motion, encompassing facial expressions, body poses, and hand gestures. Unlike prior motion transfer methods that rely on explicit skeletal poses and heuristic cross-identity adjustments, our approach encodes multi-granular motion directly from a single image into a compact set of four disentangled latent tokens -- one for facial expression, one for body pose, and one for each hand. These motion latents are both highly expressive and identity-agnostic, enabling high-fidelity, detailed cross-identity motion transfer across subjects with diverse identities, poses, and spatial configurations. To achieve this, we introduce a self-supervised, end-to-end framework that jointly learns the motion encoder and latent representation alongside a DiT-based video generative model, trained on large-scale, diverse human motion datasets. Motion--identity disentanglement is enforced via 2D spatial and color augmentations, as well as synthetic 3D renderings of cross-identity subject pairs under shared poses. Furthermore, we guide motion token learning with auxiliary decoders that promote fine-grained, semantically aligned, and depth-aware motion embeddings. Extensive experiments show that X-UniMotion outperforms state-of-the-art methods, producing highly expressive animations with superior motion fidelity and identity preservation.

Problem

• Develops unified latent representation for whole-body human motion
• Enables cross-identity motion transfer without skeletal poses
• Learns expressive identity-agnostic motion tokens via self-supervised framework

Innovation

• Unified latent tokens for multi-granular motion
• Self-supervised end-to-end DiT-based framework
• Motion-identity disentanglement via spatial augmentations