Capturing Reflectance - From Theory to Practice

Abstract

One important problem in photorealistic or predictive rendering nowadays is to realistically model the light interaction with objects. Measurements can capture the reflection properties of real world surface, i.e., they are one way of obtaining realistic reflection properties. For arbitrary (non-fluorescent, non-phosphorescent) materials, the reflection properties can be described by the 8D reflectance field of the surface, also called BSSRDF. Since densely sampling an 8D function is currently not practical various acquisition methods have been proposed which reduce the number of dimensions by restricting the viewing or relighting capabilities of the captured data sets. In this tutorial we will mainly focus on three different approaches, the first allowing to reconstruct opaque surfaces from a very small set of input images, the second allows for arbitrary surfaces but under the assumption of distant light sources and the last which allows for relighting an arbitrary scene with arbitrary spatially varying light patterns. After a short introduction explaining some fundamental concepts regarding measuring and representing reflection properties, the basics of data acquisition with photographs will be addressed. The tutorial present the set of current state-of-the art algorithms for acquiring and modeling 3D objects. The tutorial investigates the strengths and limitations of each technique and sorts them by their complexity with regard to acquisition costs. Besides describing the theoretical contributions we will furthermore point out the practical issues when acquiring reflectance fields in order to help interested users to build and implement their own acquisition setup.