> Red green and blue form a triangle, which is essentially cyclic.
But the question then - why wavelength R=700, G=550 and B=450 form a triangle?
For your answer on Quora, it says "Hue is cyclical rather than linear" without any explanation, and this is part - in my understanding - can be only explained by "left hump" (red cells being sensitive to blue light as well) and nothing else.
The triangle comes from their being 3, nothing more. There are three types of sensors. If you try to visualize them geometrically, you get a triangle.
The gamut of all colors therefore is represented as a cube, with black, white, red, green, blue, magenta, yellow, cyan as corners.
None of this is dependent on the left hump... the left hump is just a minor quirk that causes the shortest wavelengths to also stimulate the "red cones" a bit (the cone cells whose peak sensitivity is longer wavelengths).
Its easiest to understand all this if you work with colors as RGB, as anyone who does things with colors on computers typically does (paint programs, html/css, etc). Note that when you see purple (or magenta, or violet) on a computer screen , it is by combination of red and blue, but by somehow making the wavelength of light coming off the screen actually be a shorter wavelength than blue.
When I try to visualize 3 types of sensors (without left hump), I get a straight line not a triangle - that is technically still a (zero area) triangle, but inherently NOT "cyclic".
Looks like you are mixing together magenta (red+blue) and violet (shorter wavelength than blue) - those are physically very different colors, but perceived by human eye as being the same / very similar. And the only explanation for that perception (= the reason why colors form a wheel) is that "left hump", nothing else
The triangle, and the “color gamut” of a monitor or color space - come from the projection of the boundary of a 3d color space diagram down to 2 dimensions.
What you’re talking about - getting a straight line out of 3 sensors - is because you’re plotting wavelength on the x axis. What Rob is talking about is a 2d plot of a 3d color space, where points are plotted not against wavelength, but where the axes of your space are the 3 primary colors you use to define it. This is where the RGB cube comes from - 3 separate values that range from [0,1].
If you look straight down the diagonal line x=y=z, then you are looking along the “brightness” axis, which sort-of factors it out. What’s left in RGB is a hexagon - the projection of the RGB cube down to 2d along the cube’s diagonal. The hue triangle Rob’s talking about is exactly this same idea - it’s a flattening of the color space into just hue, by removing the “brightness” dimension, and then looking at the boundary of whatever is left. This boundary has different shapes in different color spaces. (And some color spaces like HSB and HSV are defined using this linear transformation, so that one of the 3 axes lies along the RGB diagonal.) A hue triangle appears when you have only 3 specific primaries, so you can’t represent any colors outside the triangle, all colors are strictly linear interpolations of the 3 outermost colors.
In this case, Rob’s right, the left hump has nothing at all to do with the triangle shape.
Rob is right that in RGB cube 3 points always form a triangle. However, that color space does not include "true" purple (or violet), which is defined as light with shorter-than-blue wavelength, it only includes magenta = red + blue, and apparently Rob does not understand the difference.
The question is not why magenta is close to red - this is obvious, - but why magenta and violet are perceived similar by human eye, and that cannot be possibly explained by "cyclic nature of triangle" in RGB cube (simply because there is no violet in that colorspace at all).
Yeah, exactly. Spectral colors don’t exist in this RGB cube, only linear combinations of 3 specific primaries do.
I agree with you - the question of purple has not been adequately answered, and I’m also becoming convinced that the subtlety and deepness of the question is not being understood or appreciated in the answer offered above. Just because people plot colors on a wheel doesn’t explain why pure violet is perceived as containing red. I’d say it’s probably the other way around - that one likely reason that artists in the past put red-purple-blue on the color wheel is because purple was already ambiguous - there’s a strange perceived similarity to the colors on opposite ends of the rainbow. A perceived red response to short wavelength blue colors, aka the left hump, is better at explaining the color wheel than the color wheel is at explaining purple.
It doesn't seem like you are really trying to visualize the input from 3 sensors if you are putting it on a line, you are still trying to represent wavelengths. Otherwise you won't be treating each of the 3 sensors equally.... you'll be putting two of them at the ends, and one in the middle. Which one goes in the middle?
Maybe it would be better if you just think of it as a color cube, which includes black, white, and all shades of gray. It is a cube because that is the best way to geometrically represent 3 dimensions.
Here's another answer of mine on Quora that goes into more detail. https://www.quora.com/Why-do-we-use-RGB-to-represent-colors-...