25. MINIMAL BRDF SAMPLING FOR TWO-SHOT NEAR-FIELD REFLECTANCE ACQUISITION

Department: Computer Science & Engineering
Research Institute Affiliation: Agile Center for Visual Computing
Faculty Advisor(s): Ravi Ramamoorthi

Primary Student
Name: Zexiang Xu
Email: zex014@ucsd.edu
Phone: 858-539-5700
Grad Year: 2021

Abstract
We develop a method to acquire the BRDF of a homogeneous flat sample from only two images, taken by a near-field perspective camera, and lit by a directional light source. Our method uses the MERL BRDF database to determine the optimal set of light-view pairs for data-driven reflectance acquisition. We improve on recent work on minimal BRDF sampling, which was based on optimizing the condition number. We develop a mathematical framework to predict error from a given set of measurements, and demonstrate better accuracy than using condition number alone; this result may also have broader applications. Crucially, our framework enables the use of multiple measurements in an image simultaneously, and we present improved error curves for acquisition from near-field setups, as opposed to conventional gonioreflectometric measurements. Finally, we demonstrate practical near-field acquisition of homogeneous BRDFs from only one or two input images.

Industry Application Area(s)
Materials | Software, Analytics

« Back to Posters or Search Results


Contact:   researchexpo@soe.ucsd.edu   (858) 534-6068