CGHair: Compact Gaussian Hair Reconstruction with Card Clustering

TL;DR

This paper introduces CGHair, a pipeline that reconstructs high-fidelity hair efficiently by clustering strands into cards and codebooks, reducing storage and computation while maintaining visual quality.

Contribution

It presents a novel clustering and grouping approach integrated with 3D Gaussian Splatting to significantly reduce hair reconstruction time and memory usage.

Findings

4-fold reduction in strand reconstruction time

Over 200x lower memory footprint

Comparable visual quality to existing methods

Abstract

We present a compact pipeline for high-fidelity hair reconstruction from multi-view images. While recent 3D Gaussian Splatting (3DGS) methods achieve realistic results, they often require millions of primitives, leading to high storage and rendering costs. Observing that hair exhibits structural and visual similarities across a hairstyle, we cluster strands into representative hair cards and group these into shared texture codebooks. Our approach integrates this structure with 3DGS rendering, significantly reducing reconstruction time and storage while maintaining comparable visual quality. In addition, we propose a generative prior accelerated method to reconstruct the initial strand geometry from a set of images. Our experiments demonstrate a 4-fold reduction in strand reconstruction time and achieve comparable rendering performance with over 200x lower memory footprint.