Wispy to Voluminous: Prior-free Multi-view Capture of Strand-level Facial Hair

2026-06-06

AI Summary

Existing methods for facial hair reconstruction in digital humans struggle to simultaneously achieve photorealism and editability, particularly in recovering sparse filamentous structures such as beards and eyelashes. This work proposes a prior-free, strand-level reconstruction approach from multi-view images, introducing for the first time an explicit representation of curved hair bundles based on 3D Gaussians. The method employs a four-stage pipeline—constrained Gaussian optimization, continuous hair tracing, physics-inspired surface anchoring, and opacity-driven density regulation—to accurately recover both the directional coherence and sparse distribution of facial hair. The resulting reconstructions are directly compatible with production-grade downstream applications, including animation, physical simulation, geometric grooming, and rendering.

Abstract

Facial hair is a defining trait of personal identity, yet remains a critical bottleneck for digital avatars. Recent volumetric methods achieve photorealism but bake hair into the underlying face geometry, preventing editability and failing to resolve sparse, strand-like structures. Meanwhile, scalp-hair reconstruction methods target dense hair volumes and do not transfer to the sparse, spatially-varying nature of facial hair. We present a pipeline that automatically reconstructs facial hair -- beard, mustache, lashes, and brows -- from multi-view images, converting an unstructured 3D Gaussian representation into an explicit curve-based strand representation. We resolve geometric ambiguities in four stages: (i) optimizing 3D Gaussians constrained by tracked head geometry to enforce early ray termination and suppress sub-surface noise; (ii) tracing continuous strands robust to frequent crossings and extreme curvature; (iii) grounding strands to the surface and resolving root-tip ambiguity via a physically-motivated prior; and (iv) refining the reconstruction through opacity-driven density control under photometric optimization. To our knowledge, this is the first method to reconstruct high-fidelity facial hair strands from a 3D Gaussian representation. The recovered strands faithfully preserve the orientation and sparsity patterns characteristic of facial hair, and yield assets immediately suitable for downstream production tasks, including facial animation and physical simulation, geometric grooming and transfer, appearance editing, and physics-based rendering.

Problem

• facial hair
• strand-level reconstruction
• multi-view capture
• 3D Gaussian representation
• digital avatars

Innovation

• strand-level reconstruction
• 3D Gaussian splatting
• facial hair modeling
• prior-free multi-view capture
• explicit curve representation