[This was my source for color names.](https://en.wikipedia.org/wiki/Lists_of_colors) What was supposed to be a simple art reference tool for myself turned into a rabbit hole…Turns out that color, as a 3d concept, comes with some MAJOR complications when trying to represent them all visually. This was as simple as I could get, and I wanted to see where Wikipedia’s named colors fell on a map of 16,777,216 digitally represented RGB colors, comprising approximately millions of visually distinguishable colors.
**Tools used:**
* **Python:** overall generation and layout
* **NumPy:** processing all 16.8 million RGB values
* **SciPy:** nearest-color matching, connected regions, and boundary regularization
* **Pillow:** rendering the atlas, borders, swatches, labels, and final PNG
* **OKLab/OKLCH:** the perceptual color space governing lightness, hue, chroma, and named-color proximity
Cha_Boy_C on
Did you discover a new favorite color through this work? It’s very cool
gavinlpicard on
For some reason this way of representing colors messes with my head. Feels like colors are missing until I zoom in and realize it’s present
samuelazers on
Do you think there are opportunities for new color names to be christened, unclaimed land
steinrrr on
Hexagons are the bestagons
Asger1231 on
What does the white borders mean, and how are they defined?
Me2910 on
This is cool but it’s a bit hard to understand
fhost344 on
It seems like about 6 million of those 16 million colors are some variation of yellow-green
OreoSpeedwaggon on
It’s definitely beaultiful-looking data, I’ll give you that, but it’s missing a big pop-up message at the top begging for donations to keep Wikipedia running.
How did you define the outlines / regions of a color? Did you use an algorithm to determine them? Or do these colors have a strictly defined region already amd that info is available?
I’m asking because I would have assumed that all the regions have the exact same shape and that there’s something mathematical and linear behind it
No_Obligation4496 on
Reminds me of compound insect eyes.
Hygro on
I wonder what color was furtherest away from being named. And which color had the most-named-most-adjacents.
manrata on
Fantasy world maps, now in rainbow colors.
DinosaurAlive on
Anyone mind if I name a couple of new ones?
Glemond
Bishblue
S’Plamina
Lorb
Gort
Plenmose
Sklub
That’s… that’s all I got from here until forever.
itsTyrion on
now redo it with wide color gamut like Display-P3 :b
Valeand on
In my opinion this could be improved by using a smaller amount of horizontal slices (=of equal lightness) through CAM16-UCS. That way the color boundaries are simply Voronoi cells, since Euclidean distance in CAM16-UCS is already one of the best perceptual distance metrics. The slices would all be different shapes and sizes, but still more legible overall.
And in general it would have made sense to repeat color names across slices, even if the color itself is located in a different one or not exactly on a slice at all.
Alternatively, exploding the 3D-cells in some interactive viewer would be the ultimate solution.
Leave A Reply
Du musst angemeldet sein, um einen Kommentar abzugeben.
17 Kommentare
[This was my source for color names.](https://en.wikipedia.org/wiki/Lists_of_colors) What was supposed to be a simple art reference tool for myself turned into a rabbit hole…Turns out that color, as a 3d concept, comes with some MAJOR complications when trying to represent them all visually. This was as simple as I could get, and I wanted to see where Wikipedia’s named colors fell on a map of 16,777,216 digitally represented RGB colors, comprising approximately millions of visually distinguishable colors.
**Tools used:**
* **Python:** overall generation and layout
* **NumPy:** processing all 16.8 million RGB values
* **SciPy:** nearest-color matching, connected regions, and boundary regularization
* **Pillow:** rendering the atlas, borders, swatches, labels, and final PNG
* **OKLab/OKLCH:** the perceptual color space governing lightness, hue, chroma, and named-color proximity
Did you discover a new favorite color through this work? It’s very cool
For some reason this way of representing colors messes with my head. Feels like colors are missing until I zoom in and realize it’s present
Do you think there are opportunities for new color names to be christened, unclaimed land
Hexagons are the bestagons
What does the white borders mean, and how are they defined?
This is cool but it’s a bit hard to understand
It seems like about 6 million of those 16 million colors are some variation of yellow-green
It’s definitely beaultiful-looking data, I’ll give you that, but it’s missing a big pop-up message at the top begging for donations to keep Wikipedia running.
i followed this same rabbit hole about a year ago!! can’t remember how i found it but i really enjoyed this guys video: https://youtu.be/HlDySNpGbyc?is=-tOHEr6QtKsUKbsF
How did you define the outlines / regions of a color? Did you use an algorithm to determine them? Or do these colors have a strictly defined region already amd that info is available?
I’m asking because I would have assumed that all the regions have the exact same shape and that there’s something mathematical and linear behind it
Reminds me of compound insect eyes.
I wonder what color was furtherest away from being named. And which color had the most-named-most-adjacents.
Fantasy world maps, now in rainbow colors.
Anyone mind if I name a couple of new ones?
Glemond
Bishblue
S’Plamina
Lorb
Gort
Plenmose
Sklub
That’s… that’s all I got from here until forever.
now redo it with wide color gamut like Display-P3 :b
In my opinion this could be improved by using a smaller amount of horizontal slices (=of equal lightness) through CAM16-UCS. That way the color boundaries are simply Voronoi cells, since Euclidean distance in CAM16-UCS is already one of the best perceptual distance metrics. The slices would all be different shapes and sizes, but still more legible overall.
And in general it would have made sense to repeat color names across slices, even if the color itself is located in a different one or not exactly on a slice at all.
Alternatively, exploding the 3D-cells in some interactive viewer would be the ultimate solution.