I really like the idea of iteration 7, but I feel like it would work a lot better with some minimum height on the y axis. Letting it peter out to zero loses the "spectrum-ness" of it, and it just looks like various random color blobs. Maybe could have a fixed minimum height, and somehow use saturation to distinguish "truly zero here" from "really low amount here".
This will doubtless be an unpopular comment, but the "Do not" imperative at the bottom of the page was super jarring to me. I really don't appreciate random websites trying to tell me what to do in such a direct imperative tone, and would never write like that even if it were my intention; not to mention that it doesn't actually prevent anything.
(Replying to myself because the above was posted over 2 hours ago)
I was thinking about how I'd go about it, and I think first I'd Hilbert sort the colours in something like OKLab space, then generate a PDF from this using Just Noticeable Diff weights, then sample that PDF using something either uniform regular or golden ratio samples (must be monotonically increasing).
Yes, the "Do not" imperative feels so bossy and unnecessary; the people who need to hear it don't care / won't heed it, and the rest of us were directly instructed not to do something (which we weren't even remotely interested in doing anyway) just because we happened to read a blog post all the way to the end.
I know it's a small thing but the tone really does bug me; consider if I would put a trailer to this Honestly Just Another Random Ass Comment On The Internet:
This is a very interesting project. As a design teacher, I recommend to my students that they do not employ color swatches for anything other than flat color designs. Certainly for an animation, a photo or a movie such simplified visualizations have little value as they do not convey the kind of ranges information this project is addressing.
That being said, I am certain that there is no 2D method of visualizing such fundamentally 3D information as color.
Visually the results are very compelling! It also gives an at-a-glance intuition about the image that the bar-style options fail to convey. I am a fan.
Pity, because it was a nice article.
I was thinking about how I'd go about it, and I think first I'd Hilbert sort the colours in something like OKLab space, then generate a PDF from this using Just Noticeable Diff weights, then sample that PDF using something either uniform regular or golden ratio samples (must be monotonically increasing).
I know it's a small thing but the tone really does bug me; consider if I would put a trailer to this Honestly Just Another Random Ass Comment On The Internet:
Do not reproduce this comment without permission.
That being said, I am certain that there is no 2D method of visualizing such fundamentally 3D information as color.