A single representation to render anything you ever saw?
Wojciech Jarosz - Dartmouth College
March 13, 2026, 12:30 p.m. - March 13, 2026, 1:30 p.m.
ENGMD 280
Hosted by: Paul Kry
Computer graphics has seen significant progress over the past few decades, with advances in light transport, material fidelity, and scene complexity greatly expanding the types of images that can be rendered. Scene representations, on the other hand, have evolved only slowly since their inception. Today, there is still a dichotomy that permeates through computer graphics and vision where we treat scene components either as volumetric media or as hard surfaces. We define their shape using different representations (e.g., volume densities vs. triangles) and even use different equations and appearance models to govern how they interact with light. Unfortunately, many scenarios create effects that lie somewhere between surfaces and volumes: While we might use a surface to model a blade of grass viewed up close, this representation becomes impractical when viewing a field of grass from afar. Transitioning to a lighter-weight volumetric representation is enticing, but often fails to produce the same appearance. Handling such continuous level of detail is a long-standing grand challenge in graphics and many other domains. In this talk, I will explain how treating geometry itself as random can unify these two extremes while enabling a new appearance continuum in between. I'll discuss some of our recent progress, as well as the ample opportunities for future research -- a journey that will touch on stochastic processes, machine learning, remote sensing, and beyond.
Wojciech Jarosz is an Associate Professor of Computer Science at Dartmouth College and co-founder of the Visual Computing Lab. Before joining Dartmouth, he led the rendering group at Disney Research Zurich (2009-2015) and was an adjunct lecturer at ETH Zurich. He obtained a Ph.D. (2008) and M.S. (2005) in computer graphics from UC San Diego, and a B.S. (2003) from the University of Illinois at Urbana-Champaign. His research is concerned with capturing, simulating, manipulating, and physically realizing how light interacts with its environment. In effect, he strives to understand why things look the way they do, how we can simulate their interaction with light efficiently, how we can intuitively author or edit that appearance, and how we can create physical objects with control over their appearance. His work in these areas has been incorporated into production rendering systems and has been used in the making of feature films, including Disney's Tangled, Frozen, and Big Hero 6. In 2013, he received the Eurographics Young Researcher Award and in 2019 the NSF CAREER Award.
