The main goal of this project is to identify relationships between human perception of real-world materials properties and corresponding
computational features. We will use Bidirectional Texture Functions as state-of-the-art digital representation of illumination and view
dependent real-world material appearance as an initial data for our analysis. We believe that rigorous analysis of human perception of
individual visual effects in different material samples will allow us not only understand principles of human perception of real material
surfaces but also to develop a more efficient acquisition, modelling, and editing methods of this type of massive but very accurate data.
The observed material properties are surface specularity, translucency, roughness, anisotropy, directionality, back-scattering, among
others. We will study the perceptual importance of these properties as well as their mutual dependency and dependency on type of
material and illumination and viewing conditions. We will use established models of low-level human perception, however, we will also
exploit psychophysical studies as a main source of knowledge about the way how human observers perceive more complex material
surface properties. We will use real-time renderings of visually rich view and illumination dependent material measurements as test stimuli
allowing the observer to freely control lighting and viewing conditions during the experiment. As an input data we are going to use publicly
available view and illumination dependent surface texture samples, however, our long-term goal is a development of flexible and efficient
sample measurement setup that does not sample the space of view and illumination directions uniformly as it is common in the previous
setups, but rather takes advantage of obtained knowledge about human perception of individual types of real-materials to produce their
very accurate but significantly reduced and thus easily applicable digital representation.