StrandHead: Text to Hair-Disentangled 3D Head Avatars Using Human-Centric Priors

Xiaokun Sun, Zeyu Cai, Ying Tai, Jian Yang, Zhenyu Zhang*
Nanjing University *Corresponding Author
ICCV 2025
ArXiv Code Gallery
Teaser Image

Given (a) input prompts, StrandHead generates (b) realistic 3D head avatars featuring strand-level attributes and (c) 3D hair strands by utilizing human-specific 2D generative priors and 3D hair strand geometric priors. By precisely capturing the internal geometry of hair strands, our approach enables seamless and flexible (d) hairstyle transfer and (e) editing, as well as (f) physics-based simulation.

Abstract

While haircut indicates distinct personality, existing avatar generation methods fail to model practical hair due to the data limitation or entangled representation. We propose StrandHead, a novel text-driven method capable of generating 3D hair strands and disentangled head avatars with strand-level attributes. Instead of using large-scale hair-text paired data for supervision, we demonstrate that realistic hair strands can be generated from prompts by distilling 2D generative models pre-trained on human mesh data. To this end, we propose a meshing approach guided by strand geometry to guarantee the gradient flow from the distillation objective to the neural strand representation. The optimization is then regularized by statistically significant haircut features, leading to stable updating of strands against unreasonable drifting. These employed 2D/3D human-centric priors contribute to text-aligned and realistic 3D strand generation. Extensive experiments show that StrandHead achieves the state-of-the-art performance on text to strand generation and disentangled 3D head avatar modeling. The generated 3D hair can be applied on avatars for strand-level editing, as well as implemented in the graphics engine for physical simulation or other applications.

Method Highlights

pipeline

Compared to previous methods, StrandHead generate 3D heads with fine geometry and lifelike textures, as well as realistic textured strand-based hairstyles, and the entire framework does not require large-scale 3D-text paired data. .

Methodology

pipeline

StrandHead includes three stages: (a) We first create a FLAME-aligned 3D bald head using the improved HumanNorm. (b) Next, we introduce a differentiable prismatization algorithm to enable human-specific geometry-aware 2D diffusion models to supervise hair shape modeling. Additionally, two losses inspired by 3D hair geometric priors are applied to further regularize the hair geometry. (c) Finally, we use a human-specific normal-conditioned 2D diffusion model to generate lifelike hair textures.

Comparisons with Text-to-Head Methods

Comparisons with Text-to-Hair Methods

Haircut Transfer

Haircut Editing

Strand-Based Rendering

Physics-Based Simulation

BibTeX

@inproceedings{StrandHead,
  title={StrandHead: Text to Hair-Disentangled 3D Head Avatars Using Human-Centric Priors},
  author={Sun, Xiaokun and Cai, Zeyu and Tai, Ying and Yang, Jian and Zhang, Zhenyu},
  booktitle=ICCV,
  year={2025}
}