Detangled: A Framework for Creating, Editing, and Inferencing Feature Rich Hair Strands

2026-07-09

AI Summary

This work addresses the challenge of entanglement between local hair strand textures and global hairstyle orientation in hair modeling. It introduces the first interpretable five-dimensional parameter space that accurately captures intrinsic strand-level textures—such as curl, kink, and twist—and decouples them from overall hairstyle style through a geometric mapping of strand centerlines into a normalized texture space. Leveraging this representation, the method integrates neural-network-driven texture annotation with generative AI models to enable high-fidelity hair strand synthesis, texture transfer, and user-controllable editing. The approach demonstrates robust generalization across diverse hair types, significantly enhancing both controllability and photorealism in hair synthesis.

Abstract

We present a framework for understanding and generating feature rich hair strands. Drawing upon both scientific and cultural expertise, we define strand texture as the various distinctive patterns (curling, switchbacks, twist, etc.) that are formed by forces internal to a hair strand. We begin by proposing a novel five-dimensional parameter space, intended to be a bijection with naturally occurring hair strand textures. This encoding is both qualitatively accessible, allowing users to readily locate their own hair in the parameter space, and quantitatively precise, allowing the generation of individual strands from texture inputs. Importantly, strand texture should be independent from the overall strand direction. In order to disentangle strand texture from the overall strand direction, we identify centerline geometry and use it to map strands into a canonical space (a strand texture space). We construct centerlines using a novel method that cleanly distills complex hair grooms, separating the strand texture from the overall style (parameterized by style guides). We enable the creation of new strands conforming to our parametric description of texture via a generative artificial intelligence approach supervised by a separate neural network trained to label candidate strands according to our five-parameter description. The ability to create new strands conforming to any desired texture enables groom editing using either texture transfer or user-provided inputs. We demonstrate results on a variety of hair types.

Problem

• hair strands
• strand texture
• disentanglement
• centerline geometry
• texture representation

Innovation

• strand texture
• disentanglement
• centerline geometry
• generative AI
• parameter space