Konversatorium on Friday, June 30, 2017 - 10:30
Advances in materials modeling and rendering
Modeling and rendering complex materials is among the central concerns
in Computer Graphics, and the Entertainment Industry is always eager to
feature photorealistic and attractive emulations of our everyday life.
However, accurate modeling and rendering of materials with complex
mesostructure (for instance, porous materials) has not received as much
attention as other topics in the literature. In this presentation we
will review some proposals in modeling and rendering this kind of
As a modeling case, bread turns out to be a structurally complex
material, for which the eye is in particular very sensitive in spotting
improper models, making adequate bread modeling a difficult task. We
developed an accurate computational bread baking model that allows to
faithfully represent the geometrical mesostructure and the appearance of
bread through its making process. This is achieved by a careful
simulation of the conditions during proving and baking to get a
realistically looking result. Some of the generative steps in the
process can be easily adapted to model other kinds of porous materials
(f.e., stones or sponges).
Regarding rendering, a remarkable property of porous materials is how
water or other liquids in their surface alter significantly their BRDFs,
which in turn determines subtle or overt changes in their visual
features. For this reason, rendering materials that change their
appearance when wet continues to be challenging. We are developing a
principled and comprehensive technique to model and render the changes
in appearance of absorbent materials under humidity conditions. This
includes a method to solve the interaction between the fluid and the
solid model, the fluid diffusion within the solid porous media, and a
physically based rendering model that adequately simulates the light
transfer behavior under these conditions. Additional features of this
model are: geometry easy to represent and to interact with, and
reasonable rendering times using off-the-shelf GPUs.
Claudio Delrieux is BS in Electric Engineering, and PhD in Computer Science. He is currently full professor and PI at the Electric and Computer Engineering Department in the Universidad Nacional del Sur (Argentina), fellow of the National Council of Science and Technology of Argentina (CONICET), and Chair of the Imaging Sciences Laboratory. His current interests are Image and Video Processing, Computer Graphics, Scientific Visualization, and Artificial Intelligence. He is author of more than 45 refereed journal papers, and more than 100 refereed international conference papers. Supervisor of 10 PhD theses (currently another 10) and 5 MSc theses (3 ongoing). PI of 3 ongoing research projects, and 5 ongoing applied research and technology transfer agreements.