Linear Models for Material BTFs and Possible Applications

Abstract

Due to the richness of real-world materials, arguably one of the biggest challenges in rendering is to come up with models that describe their appearance well. The image-based bidirectional texture function (BTF) is known to be able to model many effects that are hard or impossible to reproduce with analytical reflectance models. This advantage comes at the price of demanding storage and acquisition requirements. In previous work, we have demonstrated that these requirements can be lifted to some extent by means of data-driven linear models. We give a more in-depth overview on our research on such models and summarize the applications we investigated so far, followed by an outlook on what might yet be achievable.