Note: this has only been tested on Firefox browsers -- Chrome works, but does not always reach 240 Hz.
Supplemental website for the HPG 2019 short paper:
Temporally Dense Ray Tracing by Pontus Andersson, Jim Nilsson, Marco Salvi, Josef Spjut, and Tomas Akenine-Möller.
Hang on, measuring monitor refresh rate...
In these images, we show the effect of using the bowtie and hourglass sampling patterns.
Left: always using bowtie.
Middle-left: always using hourglass.
Middle-Right: choosing the either the hourglass and bowtie according to the rightmost image.
Right: selection mask for the middle-right image. Here, a blue pixel implies that the bowtie is used in its tile, and a yellow pixel implies that the hourglass pattern is used.
Note: We see notable reductions in aliasing in different part of the scene depending on which pattern is used. The bowtie pattern, for example, handles the beer taps and cash register a lot better
than the hourglass, but fails on the glasses, whereas the reverse is true for the hourglass pattern. Our method segments the scene properly and chooses the best pattern in both parts of the scene.