OK,
I spent a good part of the weekend working out a prototype to illustrate what
I've been thinking about.

First I needed a way to plot out RGB values so they were "flat", so that there
would be a dimension to extrude the histogram into.
I took a cue from plotting trinary HPLC solvent systems in a triangle, and after
much chasing my tail, somehow stumbled upon an RGB color hexagon.
That worked well enough, so then I needed to discretize it, so that there could
be as few or as many divisions between color value averages as desired.

(I thought about using HSV, but I was too far into it by that time, and I like
that the way this is done uses the "native" RGB values.)

After I got that worked out, it took me pretty much the rest of the day to work
out the processing of the image map RGB values and debug some other unrelated
coding stupidity.

At long last, I have something which, if not complete and as accurate as it
could be, demonstrates the idea, and serves as proof of concept.

I hope I have time to better work this out and make the code better and more
efficient (and faster!)

Post a reply to this message

Attachments:
Download 'hexagonalcolorhistogram.png' (312 KB)

Preview of image 'hexagonalcolorhistogram.png'