Estimating Specular Roughness and Anisotropy from Second Order Spherical Gradient Illumination
Eurographics Symposium on Rendering 2009
Abhijeet Ghosh    Tongbo Chen    Pieter Peers    Cyrus A. Wilson    Paul Debevec   
USC Institute for Creative Technologies

Figure 1: Specular reflectance properties ((a)-(c)) of a plastic orange estimated using polarized second order spherical gradient illumination conditions. The estimated specular roughness map (c) is used as the per-pixel distribution for a Torrance-Sparrow BRDF to create a rendering (d) that closely matches the validation photograph (e).
Abstract

We present a novel method for estimating specular roughness and tangent vectors, per surface point, from polarized second order spherical gradient illumination patterns. We demonstrate that for isotropic BRDFs, only three second order spherical gradients are sufficient to robustly estimate spatially varying specular roughness. For anisotropic BRDFs, an additional two measurements yield specular roughness and tangent vectors per surface point. We verify our approach with different illumination configurations which project both discrete and continuous fields of gradient illumination. Our technique provides a direct estimate of the per-pixel specular roughness and thus does not require off-line numerical optimization that is typical for the measure-and-fit approach to classical BRDF modeling.

Introduction

In this work, we estimate spatially varying appearance properties of materials by directly acquiring second order statistics of per surface point specular reflectance. This includes an estimate of per-pixel specular roughness as a measure of variance about a mean (e.g., reflection vector), as well as an estimate of the tangent vectors for anisotropic materials. We take a computational illumination approach towards efficient computation of these statistics by employing polarized second order spherical gradient illumination for our measurements. With this approach, we are able to obtain a more complete spatially varying BRDF information for arbitrary objects from observation under just 7 to 9 lighting conditions for isotropic and anisotropic materials respectively. Furthermore, since the proposed method relies on only up to nine distinct illumination conditions with minimal capture time, it is amenable to capturing per-surface point roughness parameters of human subjects.

Specular Reflectance Properties

Figure 2: Specular reflectance properties a flat spatially varying BRDF sample ((a)-(c)) estimated using second order spherical gradient illumination using an LCD monitor as emitter. The estimated specular roughness map (c) is used as the per pixel distribution for a Torrance-Sparrow BRDF to create renderings (d) that closely matches the validation photograph (e) with illumination from the side.
Results


    Anisotropic BRDF     Isotropic BRDF

Downloads

Eurographics Symposium on Rendering 2009 Paper
egsr2009-final.pdf, (15.3MB)

SIGGRAPH 2009 Talk
SIG2009-Talk.pdf, (1.92MB)

Related Projects

High-Resolution Face Scanning, Eurographics Symposium on Rendering 2007
Linear Light Source Reflectometry, SIGGRAPH 2003 Conference Proceedings

Light Stage 1
Acquiring the Reflectance Field of a Human Face, SIGGRAPH 2000
Facial Reflectance Field Demo, SIGGRAPH 2000 Creative Applications Laboratory
Realistic Human Face Scanning and Rendering, ICT Graphics Lab 2001

Light Stage 2
A Photometric Approach to Digitizing Cultural Artifacts, VAST 2001
Animatable Facial Reflectance Fields, EGSR 2004
Reflectance Field Rendering of Human Faces for "Spider-Man 2", SIGGRAPH 2004 Sketch

Light Stage 5
Postproduction Re-Illumination of Live Action Using Interleaved Lighting, SIGGRAPH 2004 Poster
Performance Geometry Capture for Spatially Varying Relighting, SIGGRAPH 2005 Sketch

Footer With Address And Phones