I understand that color is a 3D value, since all perceivable colors can be created my independently mixing 3 base colors. That naively gives us a color space shaped like a cube (like SRGB).
I also understand that every color space needs to remove colors from the ‘everything’ color space due to technical limitations. This is usually represented by ‘cutting out’ a triangle from the following representation of the ‘everything color space’.
BenRG and cmglee, CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0, via Wikimedia Commons
Even accounting for distortion of the color space (think Mercator projection), I believe there should only be three curves that cut the solid color space. Hover, in the image of the OKLAB color space there are 5. Why is that?
And why is the ‘amount of height’ they cut from this supposed triangular prism so different depending on hue?
The image is taken from the OKLCH Color Picker & Converter by Andrey Sitnik & Roman Shamin. https://oklch.com/#0.7344,0.1025,272,100
Each of the three types of human cone cells have their own nonlinear responses to the EM spectrum, including a notable multimodal response in the red “L” cells that also peaks locally around blue wavelengths. All of this captured in an admittedly quick-and-dirty approximation will lead to some unintuitively jagged curves; however, qualitative evaluation is best done on the colors themselves, rather than the appearance of one rendering of the bounds of the colorspace.
Think of it like the skin of a 3d shape sliced apart and stretched over a 2d plane, because that it what you are looking at.
The space is continuous in 3d.
The plot you show changes as you change the ratios of l,c,h
When you click the ‘show 3d’ button you can see the cut out areas correspond to colors that are not on the surface of the color volume.
