Uniform Color Spaces: 3D LUTs vs. Algorithms

Until recently I have always used L*a*b* as a very convenient tool to transform color data into a sensible isotropic color space. Since isotropic spaces are uniform in all directions, the space performed the difficult task of making sure that differences in hue, lightness and chroma were all taken into account and that any pair of equal distances between two colors, regardless of their location in color space appeared equally different to observers. How this happened was left as a mystery, because I needed this property for the next step in my work. I was willing to sweep under the rug the messy issues of color surround, the complexity of the image and multiple illuminants because I was in a hurry. I also knew of reports of departures of L*a*b* from perfect uniformity, as well as great many papers suggesting clever variations in formulas that moved this or that part of the color space closer to ideal color uniformity.