Recent advancements in spatiotemporal reservoir resampling (ReSTIR) leverage sample r euse from neighbors to efficiently evaluate the path integral. Like rasterization, ReSTIR m ethods implicitly assume a pinhole camera and evaluate the light arriving at a pixel throug h a single predetermined subpixel location at a time (e.g., the pixel center). This prevents efficient path reuse in and near pixels with high-frequency details.

We introduce Area ReSTIR, extending ReSTIR reservoirs to also integrate each pixel's 4D ray space, including 2D areas on the film and lens. We design novel subpixel-tracking tem poral reuse and shift mappings that maximize resampling quality in such regions. This rob ustifies ReSTIR against high-frequency content, letting us importance sample subpixel an d lens coordinates and efficiently render antialiasing and depth of field.

Authors

Venue

Abstract