NECromancer: Breathing Life into Skeletons via BVH Animation

2026-02-06

AI Summary

This work addresses the limited generalizability of existing motion tokenization methods, which are constrained by species-specific skeletal structures and struggle to handle morphologically diverse skeletons. To overcome this, we propose NECromancer, a universal motion tokenizer that, for the first time, enables unified tokenization of arbitrary BVH skeletons. NECromancer integrates an embodied skeletal graph encoder (OwO), a topology-agnostic tokenizer (TAT), and a large-scale heterogeneous dataset (UvU) to achieve structure-aware encoding and discrete representation learning, effectively decoupling motion semantics from skeletal topology. The method achieves high-fidelity motion reconstruction at high compression rates, and its learned token space supports cross-species motion transfer, composition, denoising, and generation. Extensive experiments across diverse skeletal morphologies demonstrate NECromancer’s superior performance in cross-morphology motion modeling.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Deep Generative Models & Autoencoders
• Intelligent Robots: Embodied AI

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Motion tokenization is a key component of generalizable motion models, yet most existing approaches are restricted to species-specific skeletons, limiting their applicability across diverse morphologies. We propose NECromancer (NEC), a universal motion tokenizer that operates directly on arbitrary BVH skeletons. NEC consists of three components: (1) an Ontology-aware Skeletal Graph Encoder (OwO) that encodes structural priors from BVH files, including joint semantics, rest-pose offsets, and skeletal topology, into skeletal embeddings; (2) a Topology-Agnostic Tokenizer (TAT) that compresses motion sequences into a universal, topology-invariant discrete representation; and (3) the Unified BVH Universe (UvU), a large-scale dataset aggregating BVH motions across heterogeneous skeletons. Experiments show that NEC achieves high-fidelity reconstruction under substantial compression and effectively disentangles motion from skeletal structure. The resulting token space supports cross-species motion transfer, composition, denoising, generation with token-based models, and text-motion retrieval, establishing a unified framework for motion analysis and synthesis across diverse morphologies. Demo page: https://animotionlab.github.io/NECromancer/

Problem

• motion tokenization
• skeleton generalization
• BVH animation
• cross-species motion
• morphological diversity

Innovation

• universal motion tokenizer
• BVH animation
• topology-agnostic representation
• skeletal graph encoding
• cross-species motion transfer