← ArchivePaper2022
Neural Strands: Learning Hair Geometry and Appearance from Multi-View Images
Radu Alexandru Rosu, Shunsuke Saito, Ziyan Wang, Chenglei Wu, Sven Behnke, Giljoo Nam
Joint learning of explicit hair geometry and view-dependent appearance using a neural scalp texture encoding individual strand properties at each texel.
Abstract
We present Neural Strands, a novel learning framework for modeling accurate hair geometry and appearance from multi-view image inputs. The learned hair model can be rendered in real-time from any viewpoint with high-fidelity view-dependent effects. Our model achieves intuitive shape and style control unlike volumetric counterparts. To enable these properties, we propose a novel hair representation based on a neural scalp texture that encodes the geometry and appearance of individual strands at each texel location. Furthermore, we introduce a novel neural rendering framework based on rasterization of the learned hair strands. Our neural rendering is strand-accurate and anti-aliased, making the rendering view-consistent and photorealistic. Combining appearance with a multi-view geometric prior, we enable, for the first time, the joint learning of appearance and explicit hair geometry from a multi-view setup. We demonstrate the efficacy of our approach in terms of fidelity and efficiency for various hairstyles.
How to read this
- Category
- Method: neural hair geometry and appearance from multi-view images
- Contributions
- A neural scalp texture representation encoding the geometry and appearance of individual strands at each texel
- A strand-accurate, anti-aliased neural rendering framework based on rasterizing the learned strands, giving view-consistent photorealistic results in real time
- Enables, for the first time, joint learning of explicit hair geometry and appearance from a multi-view setup, with intuitive shape and style control
- Context
- Builds on multi-view strand-level hair capture such as Nam et al. Strand-Accurate Multi-View Hair Capture (2019), adding learned appearance and neural rendering atop explicit geometry.Builds on: Strand-Accurate Multi-View Hair Capture
- Correctness
- Demonstrated across various hairstyles for fidelity and efficiency and contrasted with volumetric approaches for controllability; it depends on a multi-view capture setup and a geometric prior, so results are bounded by capture coverage and the prior's accuracy.
- Clarity
- Fairly technical; a first pass conveys the neural-scalp-texture plus strand-rasterization idea, a second pass covers the rendering and joint-learning formulation.
- How to read it
- First pass for the explicit-strand-plus-neural-appearance framing and why it beats volumetric for control; second pass on the scalp-texture encoding and strand-accurate rasterization if implementing hair capture.
Builds on
Built upon by
Related work
- Neural Haircut: Prior-Guided Strand-Based Hair Reconstruction 2023 / ICCV
- FLARE: Fast Learning of Animatable and Relightable Mesh Avatars 2023 / SIGGRAPH Asia
- 3D Hair Synthesis Using Volumetric Variational Autoencoders 2018 / SIGGRAPH Asia
- HAAR: Text-Conditioned Generative Model of 3D Strand-Based Human Hairstyles 2023 / arXiv
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →