TALK    [MERL Seminar Series 2025] David Lindell presents talk titled Imaging Dynamic Scenes from Seconds to Picoseconds

Date & Time:

Wednesday, January 29, 2025; 1:00 PM

Abstract:

The observed timescales of the universe span from the exasecond scale (~1e18 seconds) down to the zeptosecond scale (~1e-21 seconds). While specialized imaging systems can capture narrow slices of this temporal spectrum in the ultra-fast regime (e.g., nanoseconds to picoseconds; 1e-9 to 1e-12 s), they cannot simultaneously capture both slow (> 1 second) and ultra-fast events (< 1 nanosecond). Further, ultra-fast imaging systems are conventionally limited to single-viewpoint capture, hindering 3D visualization at ultra-fast timescales. In this talk, I discuss (1) new computational algorithms that turn a single-photon detector into an "ultra-wideband" imaging system that captures events from seconds to picoseconds; and (2) a method for neural rendering using multi-viewpoint, ultra-fast videos captured using single-photon detectors. The latter approach enables rendering videos of propagating light from novel viewpoints, observation of viewpoint-dependent changes in light transport predicted by Einstein, recovery of material properties, and accurate 3D reconstruction from multiply scattered light. Finally, I discuss future directions in ultra-wideband imaging.

Speaker:

David Lindell
University of Toronto

David Lindell is an Assistant Professor in the Department of Computer Science at the University of Toronto. Prior to joining the University of Toronto, he received his Ph.D. from Stanford University. His research combines applied optics, emerging sensor platforms, machine learning, and physics-based algorithms to enable new capabilities in visual computing. He is a recipient of the ACM SIGGRAPH Outstanding Dissertation Award Honorable Mention, a Google Research Scholar Award, a Sony Faculty Innovation Award, and the Marr Prize.

MERL Host:

Joshua Rapp

Research Areas:

Computational Sensing, Computer Vision, Signal Processing