PanoHair: Detailed Hair Strand Synthesis on Volumetric Heads

TL;DR

PanoHair is a novel model that efficiently synthesizes detailed, diverse hair strands on volumetric head models using a generative approach with rapid mesh generation, bypassing complex data acquisition.

Contribution

It introduces a generative, fast, and data-efficient method for detailed hair strand synthesis on volumetric heads, leveraging knowledge distillation and latent space manipulation.

Findings

Generates hair strands in under 5 seconds.

Produces diverse hairstyles with latent space control.

Outperforms existing methods in speed and quality.

Abstract

Achieving realistic hair strand synthesis is essential for creating lifelike digital humans, but producing high-fidelity hair strand geometry remains a significant challenge. Existing methods require a complex setup for data acquisition, involving multi-view images captured in constrained studio environments. Additionally, these methods have longer hair volume estimation and strand synthesis times, which hinder efficiency. We introduce PanoHair, a model that estimates head geometry as signed distance fields using knowledge distillation from a pre-trained generative teacher model for head synthesis. Our approach enables the prediction of semantic segmentation masks and 3D orientations specifically for the hair region of the estimated geometry. Our method is generative and can generate diverse hairstyles with latent space manipulations. For real images, our approach involves an inversion process to infer latent codes and produces visually appealing hair strands, offering a streamlined alternative to complex multi-view data acquisition setups. Given the latent code, PanoHair generates a clean manifold mesh for the hair region in under 5 seconds, along with semantic and orientation maps, marking a significant improvement over existing methods, as demonstrated in our experiments.