0

I'm working with the RGB cube, trying to create a maximal palette of strongly distinct colors. So I'm taking slices of the RGB cube perpendicular to the diagonal between the black and white vertices and applying the CIELab color distance formula. It works great at the edges where the most saturated colors are, i.e. most interesting colors, but as you get closer to the center where the less interesting colors are the density increases dramatically instead of decreasing like it should. I want the density to go to 0 at the center.

I'm not sure where to begin with inverting a density distribution so that the center has minimum density rather than maximum density? My only thought is to create my own probability field and use the Monte Carlo method with it. I know how to do that in 1D, but not 2D.

Mike Reynolds
  • 29
  • 5
  • are you talking about simplex in the unit cube? You need uniform sampling on the simplex? – Severin Pappadeux Dec 28 '15 at 00:59
  • It's not clear to me what you're trying to do, but I'll hazard a guess. How about constructing a density which is a function of the distance from the center of the cube, with density that is strictly increasing with distance? E.g. a power function. To sample from such a density, choose a direction at random and find the point at which a ray in that direction intersects with the surface of the cube. Then sample from the segment of that ray from the center to the surface of the cube. We can make this more concrete if it seems helpful. – Robert Dodier Dec 28 '15 at 20:00

0 Answers0