Minimal BRDF Sampling for Two-Shot Near-Field Reflectance Acquisition

摘要

We develop a method to acquire the BRDF of a homogeneous flatsample from only two images, taken by a near-field perspectivecamera, and lit by a directional light source. Our method usesthe MERL BRDF database to determine the optimal set of lightviewpairs for data-driven reflectance acquisition. We develop amathematical framework to estimate error from a given set of measurements,including the use of multiple measurements in an imagesimultaneously, as needed for acquisition from near-field setups.The novel error metric is essential in the near-field case, where weshow that using the condition-number alone performs poorly. Wedemonstrate practical near-field acquisition of BRDFs from onlyone or two input images. Our framework generalizes to configurationslike a fixed camera setup, where we also develop a simpleextension to spatially-varying BRDFs by clustering the materials.