Please note: This PhD seminar will take place in DC 2314 and online.
Weijie Zhou, PhD candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Toshiya Hachisuka
As increasingly sophisticated correlation-based techniques emerge, efficiently leveraging correlations between pixel estimates becomes critically important. Back et al. [BHHM20] introduced the combiner theory to effectively fuse independent and correlated pixel estimates; however, their approach focuses solely on spatial correlations.
In this paper, we propose the Residual Combiner, a generalized extension of the Deep Combiner framework, explicitly designed to exploit correlations across spatial, temporal, and multiscale domains. Our method enables robust cross-domain fusion, effectively reducing systematic artifacts and significantly enhancing temporal coherence, especially important in animation scenarios. We demonstrate the effectiveness of our proposed method through several practical applications, showcasing substantial improvements in temporal stability, visual fidelity, and reduction of residual errors across diverse rendering scenarios.
To join this PhD seminar in person, please go to DC 2314. You can also attend virtually on Zoom.