Why the RGB Color Model is so Popular

[niggawut]

Why is the Red, Green and Blue color model so commonly used to produce a broad array of colors?

According to Wikipedia:

The choice of primary colors is related to the physiology of the human eye; good primaries are stimuli that maximize the difference between the responses of the cone cells of the human retina to light of different wavelengths, and that thereby make a large color triangle.

The normal three kinds of photoreceptor cells in the human eye (cone cells) have peak responses of light wavelengths near 570 nm, 540 nm and 440 nm).

Wait a minute? Isn't red light generally defined at 650nm instead of 570nm?

So shouldn't we be using a Yellow, Blue and Green model?

 

[niggawut]

The point is, why would Red 650nm be chosen when the human eye's peak is far off at 570nm?

 

[xts]

There are two main reasons:
1. physiology: eye receptors has very wide spectral characteristics. The goal was not to fit to their maxima, but to find wavelengths exciting only one of them, leaving other two as little excited as possible. Consider yourself watching the red rose. Red light excites our yellow receptor, but the green one remains almost not excited. Now you watch yellow sunflowers. Yellow light excites yellow receptor and green one too. We perceive excitement of yellow receptor in absence of excitement of green one as red colour, and excitement of both of them as yellow. If we use yellow/green/blue we wouldn't be ever able to produce the impression of red, as yellow light excites green receptor too.

Our goal is to find such scheme that:
R - excites our "yellow" receptor, leaving the remaining two unexcited (we maximalise ratio of excitement of yellow/green receptors)
B - excites our "blue" receptor, leaving the remaining two unexcited (we maximalise ratio of excitement of blue/green receptors)
G - excites our "green" receptor, leaving the remaining two unexcited (we maximalise ratio of excitement of green/max(yellow,blue) receptors)
Those physiological optima are different than sensitivity maxima of the receptors.

The best scheme is red - as long wave, as yet visible, violet/blue - as short as yet visible, and something in the middle, differentiating between green/blue receptors to the same level as between yellow/green.

2. Historical technology: 60 years ago, when first colour TV had been built, RGB were chosen a little bit shifted from physiologically optimal, but as possible to realize by luminophores for early CRT tubes. We had no good violet luminophores those times, but - fortunately - the impression of violet may be made by simultaneous excitement of our blue and yellow receptors with green one unexcited. Green was also chosen as a bit shorter wave than yellowish-green which would be more optimal, as we had good green luminophore, but no good one for yellowish-green.

 

[Drakkith]

Realize that our color receptors are only called "red" "green" and "blue" because it is precisely those individual colors that we can combine to give us almost any color that we see. You could easily call the Red cone a Yellow cone instead. It makes no difference expect that it would probably confuse people.

As point 2 in XTS's post above says, the colors might be shifted a little bit from "perfect" but that only points out that we don't need to have a perfect system in order for it to work. Each color could be shifted slightly and there would be a negligible effect.

 

[niggawut]

interesting.

what do you mean by "maximalising" the ratio though?

 

[Drakkith]

interesting.

what do you mean by "maximalising" the ratio though?

Take the L receptor for example. (The red receptor) Using 600 nm light instead of 650 nm would result in much more activation of the L receptor than currently. However, that would also vastly increase the activation of the M receptor. So instead of that pixel activating just 1 receptor it would activate 2 and the pixel would appear as a reddish orange. This would throw off the color balance and also mean that the device would be unable to show the color red, as its lowest frequency would show red-orange instead of just red.

 

[xts]

what do you mean by "maximalising" the ratio though?

Let us describe the excitement level of the receptor numerically. For photodiodes it would be a current we measure, for neural receptor it is (disclaimer: I am not a physiologist) a combination of voltage and frequency of discharges on neurons in vision nerve. So just assume it has some numerical value such, that higher value is perceived as lighter light, and lower value as dimmer one. BTW: that's the number shown on y-axis of your plots of receptor characteristics.

Now we peek some wavelength.
Then we adjust light intensity such, that the signal from "yellow" receptor has some assumed value.
We measure signals from "green" and "blue" receptors.

We play with different wavelengths and the above procedure to find the lowest posible signals from "green" and "blue".

 

[Andy Resnick]

Why is the Red, Green and Blue color model so commonly used to produce a broad array of colors?

The RGB color model is but one of many- the RGB model is based on human color vision (as xts points out), but requires negative values to produce certain colors (blue-green region). So, color television instead used the "XYZ" model (which may be also called the YUV model, I'm not sure) based on luminance and chrominance to reproduce most of the visible spectrum without any negative values. This was an FCC standard, and I don't know how the standard has carried over to digital transmission.

There are many color models: CMYK (cyan, magenta, yellow, black) used for printing, HSV (hue, saturation, value) which is a cylindrical coordinate representation of the RGB model, etc. etc.

 

[Drakkith]

Exactly Andy. Well put.