Figure 6 shows results for a scene rendered using three different light sources. While this hue correction method overshoots for the image with the red light source (because the gray world assumption doesn’t hold for this image), on the whole it does a credible job on removing the light source’s influence. Changing the gray world assumption implies the chromatic α and β aver- ages should be moved to a location other than (0, 0). By making that a 2D choice problem, it’s easy to browse possibilities starting at (0, 0), and Figure 6’s third col- umn shows the results of this exercise. We hand cali- brated the images in this column to resemble the results of Pattanaik et al.’s chromatic adaption algorithm,7 which for the purposes of demonstrating our approach to hue correction we took as ground truth. Finally, we showcase a different use of the lαβspace in the sidebar “Automatically Constructing NPR Shaders.”