|
|
|
|
|
|
by ralferoo
54 days ago
|
|
|
The latter tests were all a bit pointless because they were just turquoise, and all looked much the same - a mix of blue and green, so I was pretty much answering based on whether it was bluer or greener than the previous image. The results said "Your boundary is at hue 179, bluer than 82% of the population. For you, turquoise is green." and definitely if I was judging the boundary on a gradient, I'd have placed the line a bit further to the right. |
|
|
It wants to do a binary search in the color space to find "The Boundary" between your personal concept of green and your personal concept of blue.
No such boundary exists. If you abandon the idea of binary search and just present a bunch of similar colors in no particular order, you'll get more realistic results, which look like "you called color x 'blue' x% of the time, and 'green' (1-x)% of the time". You could even display a grayscale histogram of the blue-green continuum according to the odds of any particular point on it being labeled "blue" or "green". That would be fun.