3DGH: 3D Head Generation with Composable Hair and Face

2025-06-25

AI Summary

This work addresses the problem of entangled hair and facial geometry in 3D head generation, which severely hinders controllable editing. We propose an unconditional, composable full-head generative model. Our method introduces: (1) a template-based deformable 3D Gaussian splatting representation that explicitly decouples hair and facial geometry; (2) a dual-generator architecture with cross-attention mechanisms to jointly model their semantic correlations, enabling structural-appearance co-generation; and (3) a training framework integrating template registration, synthetic rendering supervision, and attention-based feature fusion, supporting high-fidelity unconditional generation and independent hairstyle or facial editing. Extensive evaluations on multiple benchmarks demonstrate that our approach significantly outperforms state-of-the-art methods in image quality, geometric fidelity, and editing controllability.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We present 3DGH, an unconditional generative model for 3D human heads with composable hair and face components. Unlike previous work that entangles the modeling of hair and face, we propose to separate them using a novel data representation with template-based 3D Gaussian Splatting, in which deformable hair geometry is introduced to capture the geometric variations across different hairstyles. Based on this data representation, we design a 3D GAN-based architecture with dual generators and employ a cross-attention mechanism to model the inherent correlation between hair and face. The model is trained on synthetic renderings using carefully designed objectives to stabilize training and facilitate hair-face separation. We conduct extensive experiments to validate the design choice of 3DGH, and evaluate it both qualitatively and quantitatively by comparing with several state-of-the-art 3D GAN methods, demonstrating its effectiveness in unconditional full-head image synthesis and composable 3D hairstyle editing. More details will be available on our project page: https://c-he.github.io/projects/3dgh/.

Problem

• Generates 3D human heads with separate hair and face components
• Uses deformable hair geometry for diverse hairstyle variations
• Enables composable 3D hairstyle editing via dual-generator GAN

Innovation

• Template-based 3D Gaussian Splatting representation
• Dual generators with cross-attention mechanism
• Synthetic renderings with stabilized training objectives