genBRDF: Discovering New Analytic BRDFs with Genetic Programming

ACM Transactions on Graphics (Proc. SIGGRAPH), August 2014

Adam Brady, Jason Lawrence, Pieter Peers, Westley Weimer.



Abstract

We present a framework for learning new analytic BRDF mod- els through Genetic Programming that we call genBRDF. This ap- proach to reflectance modeling can be seen as an extension of tradi- tional methods that rely either on a phenomenological or empirical process. Our technique augments the human effort involved in de- riving mathematical expressions that accurately characterize com- plex high-dimensional reflectance functions through a large-scale optimization. We present a number of analysis tools and data vi- sualization techniques that are crucial to sifting through the large result sets produced by genBRDF in order to identify fruitful ex- pressions. Additionally, we highlight several new models found by genBRDF that have not previously appeared in the BRDF litera- ture. These new BRDF models are compact, and more accurate than current state-of-the-art alternatives.

Paper

Talk Slides

Supplemental Results