3 ms·
Great work. I've already been more fulfilled with this than I have with the multimillion dollar Color startup... It'd be cool to see the actual deviation in nu
by danso 15y ago
Great work. I've already been more fulfilled with this than I have with the multimillion dollar Color startup...
It'd be cool to see the actual deviation in numbers...I feel that some of my "perfect" matches were just shots in the dark.
- duopixel 15y agoThe formula for finding the difference in color is CIEDE2000, http://en.wikipedia.org/wiki/Color_difference#CIEDE2000 http://en.wikipedia.org/wiki/Color_difference#CIEDE2000, the actual number is quite meaningless, but in case you're interested: score = "perfect" if distance <= 2 score = "very good" if distance > 2 and distance <= 6 score = "good" if distance > 6 and distance <= 12 score = "average" if distance > 12 and distance <= 18 score = "poor" if distance > 18 The score is more strict on the first exercise (Hue only)
- cr4zy 15y agoThanks for linking that. I recently contrived my own computer vision color difference formula by using distance within an HSL bicone (http://dl.dropbox.com/u/9632169/HTM/Hcl-hcv_models.svg http://dl.dropbox.com/u/9632169/HTM/Hcl-hcv_models.svg). This is much better in that it takes into account differences in human sensitivity to certain colors.
- duopixel 15y agoCheck out this fork of D3 that features Lab conversion and different distance formulas https://github.com/jheer/d3/blob/master/d3.color.js https://github.com/jheer/d3/blob/master/d3.color.js