← ArchivePaper2017
A Data-Driven Approach to Four-View Image-Based Hair Modeling
Takes four-view hair photos and estimates a 3D direction field to grow dense strands; allows mixing views from different hairstyles for flexible modeling.
Abstract
We introduce a novel four-view image-based hair modeling method. Given four hair images taken from the front, back, left and right views as input, we first estimate the rough 3D shape of the hair observed in the input using a predefined database of 3D hair models, then synthesize a hair texture on the surface of the shape, from which the hair growing direction information is calculated and used to construct a 3D direction field in the hair volume. Finally, we grow hair strands from the scalp, following the direction field, to produce the 3D hair model, which closely resembles the hair in all input images. Our method does not require that all input images are from the same hair, enabling an effective way to create compelling hair models from images of considerably different hairstyles at different views. We demonstrate the efficacy of our method using a wide range of examples.
How to read this
- Category
- Method: image-based 3D hair modeling from multiple views
- Contributions
- A four-view (front, back, left, right) image-based hair modeling pipeline
- Estimates rough 3D hair shape from a predefined database, synthesizes a surface hair texture, and builds a 3D direction field used to grow dense strands from the scalp
- Allows mixing input views from different hairstyles, enabling hair models assembled from considerably different references
- Context
- Builds on single-image and database-driven hair modeling (Chai et al., AutoHair), extending capture from one image to a four-view setup with a direction-field strand-growth stage.Builds on: AutoHair: Fully Automatic Hair Modeling from a Single Image
- Correctness
- Demonstrated qualitatively across a range of examples; quality depends on the coverage of the 3D hair database and on consistent direction-field estimation from the photos, and the four-view input requirement limits fully casual capture.
- Clarity
- Pipeline is accessible; a first pass conveys the shape-then-texture-then-direction-field-then-grow flow, with stage details in later passes.
- How to read it
- First pass for the four-view pipeline and the cross-hairstyle mixing capability; second pass on direction-field construction and strand growth if you work on hair capture.
Built upon by
Nothing yet.
Related work
- Single-View Hair Modeling Using a Hairstyle Database 2015 / SIGGRAPH
- SMPLicit: Topology-aware Generative Model for Clothed People 2021 / CVPR
- SwinGar: Spectrum-Inspired Neural Dynamic Deformation for Free-Swinging Garments 2024 / TVCG
- SCANimate: Weakly Supervised Learning of Skinned Clothed Avatar Networks 2021 / CVPR
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 →