Intuitive editing of material appearance

Abstract

Many different techniques for measuring material appearance have been proposed in the last few years. These have produced large public datasets, which have been used for accurate, data-driven appearance modeling. However, although these datasets have allowed us to reach an unprecedented level of realism in visual appearance, editing the captured data remains a challenge. In this work, we develop a novel methodology for intuitive and predictable editing of captured BRDF data, which allows for artistic creation of plausible material appearances, bypassing the difficulty of acquiring novel samples. We synthesize novel materials, and extend the existing MERL dataset [Matusik et al. 2003] up to 400 mathematically valid BRDFs. We design a large-scale experiment with 400 participants, gathering 56000 ratings about the perceptual attributes that best describe our extended dataset of materials. Using these ratings, we build and train networks of radial basis functions to act as functionals that map the high-level perceptual attributes to an underlying PCA-based representation of BRDFs. We show how our approach allows for intuitive edits of a wide range of visual properties, and demonstrate through a user study that our functionals are excellent predictors of the perceived attributes of appearance, enabling predictable editing with our framework.