Measurement-based Modeling for Computer Graphics

Abstract

Modeling and visualizing scenes that are indistinguishable from real photographs is one of the long standing goals of computer graphics. This talk will focus on the former aspect (modeling), and in particular on appearance modeling. Appearance modeling roughly refers to the design and creation of digital representations of materials for the use in virtual scenes, and all other aspects that impact appearance during visualization. When reproducing the appearance of physical scenes, it is often more convenient to use measurements as a basis for modeling than to rely solely on artistry and manual modeling tools, due to the overwhelming amount of details present in these physical scenes.

In this talk I will provide an overview of my contributions to various aspects of measurement-based modeling in computer graphics. In particular I will discuss contributions to appearance acquisition, specialized compression techniques, editing of measured appearance data, and finally the physical (re-)synthesis of appearance. A reoccurring theme in the presented research is the use of a combination of active illumination and mathematical techniques to improve specific aspects of measurement-based appearance modeling.