The Beam Radiance Estimate for Volumetric Photon Mapping
(Image Comparison)
Below you can find image comparisons for the four different scenes used in the paper "The Beam Radiance Estimate for Volumetric Photon Mapping". We compare our new beam radiance estimate (BRE) to traditional volumetric photon mapping (VPM) using ray marching. All images of the same scene are rendered using the same photon map.
◄ back to the publication web page
Jump to scene:
Stage
We compare the conventional volumetric photon mapping (VPM) and our beam radiance estimate (BRE) on the Stage scene with render times provided as (hours:minutes:seconds). Our method produces images with much less noise than an equal time rendering using the conventional approach for both a fixed radius and an adaptive radius approach
In the first comparison, we adjusted the ray marching parameters for the VPM renderings to obtain roughly equal render time to the BRE.
In the second set, we compare the beam radiance estimate (BRE) to a "ground truth" computed using traditional volumetric photon mapping (VPM) with a very small ray marching step size.
Cars
The Cars scene. For both the fixed and adaptive gathering approaches our method produces noise-free results while conventional photon mapping suffers from significant noise, especially around distant light sources. Render times are shown as (hours:minutes:seconds). In the first comparison, we adjusted the ray marching parameters for the VPM renderings to obtain roughly equal render time to the BRE.
Lighthouse
The Lighthouse scene. For both the fixed and adaptive gathering approaches our method produces noise-free results while conventional photon mapping suffers from significant noise, especially around distant light sources. Render times are shown as (hours:minutes:seconds). In the first comparison, we adjusted the ray marching parameters for the VPM renderings to obtain roughly equal render time to the BRE.
Cornell Box
The Lighthouse scene. For both the fixed and adaptive gathering approaches our method produces noise-free results while conventional photon mapping suffers from significant noise, especially around distant light sources. Render times are shown as (hours:minutes:seconds). In the first comparison, we adjusted the ray marching parameters for the VPM renderings to obtain roughly equal render time to the BRE.
