CRT Colour Gamuts and Human Vision

[AndyCunningham]

This has come up in some graphics work I'm doing and the answers I've heard so far aren't that great. Here's the question:

Advertising for CRTs often makes the claim of ability to display 16.7 million or 4 billion (24/32 bit) colours. Discuss this claim in view of the mechanisms of human colour perception.

I know there's two parts to this, and that the first bit is that, in mixing 3 colours, there is only a set gamut which can be formed, and this gamut can't cover the whole range of perceptible colours. What I don't know is the second major point about this. Anyone here know any more?

 

[CSoup]

Are you asking why three colors can't generate all human perceivable colors?

 

[CTho9305]

Why do you think you can't cover the whole range of perceptible colors? Do you have an example? (This may of course be something that only some people can notice - some people can't tell a 128kbps MP3 from CD, some can distinguish at up to 192kbps).

 

[CSoup]

<< Why do you think you can't cover the whole range of perceptible colors? Do you have an example? (This may of course be something that only some people can notice - some people can't tell a 128kbps MP3 from CD, some can distinguish at up to 192kbps). >>



Well, I don't have any links to illustrations, but a human visual gamut can be drawn as a something that looks like those circular color chosers. The RGB gamut is essentially a triangle with the points being each of the red, green, and blue phosphors. As you can see, this triangle does not cover the entire circle of human visible colors. Essentially, it is impossible to cover the entire human gamut with three phosphors unless you have one that subtracts light which is not physically possible. Same problem with the CYMK gamut. Also the RGB and CYMK gamut are different which leads to color matching problems as some colors might not show up on your monitor, but are printable, and vice versa.

Specifically, (recalling from memory here, so could be off) certain yellows and purples can not be displayed on screen.

Edit: ok found a link, the human gamut is not quite a circle as I tried to describe above, but look at the last picture on the following page. It gives an illustration of the human gamut compared to the RGB, and CYMK gamuts.

 

[CTho9305]

you learn something every day! 😉

 

[yomega]

<< you learn something every day! 😉 >>



That's just what I was thinking.

 

[LiquidCyanide]

the human retina can be compared to a field of various spikes. There are thousands of "rods and cones" scattered accross it. Each rod or cone sees one color, red, blue, or yellow, sort of like the inverse of a CRT screen. However, i would liek to see that document about why CRTs cant produce certain colors

 

[silverpig]

<< Why do you think you can't cover the whole range of perceptible colors? Do you have an example? (This may of course be something that only some people can notice - some people can't tell a 128kbps MP3 from CD, some can distinguish at up to 192kbps). >>


<brag>Hehe, I can tell between 320 kbps CBR with LAME and the original CD 🙂</brag>

As to the colour, well, cones are responsible for colour perception, rods for greyscale/intensity... (cones = day vision, rods = nightvision basically).

Perhaps the second part has to do with the perception scheme of the eye... Do all the cones contribute equally to colour perception, or do some register only red, some only green... etc... Also, it may have to do with our colour perception peaking at 5770 K (or something like that I think), and dropping of on each side... Perhaps monitors aren't calibrated to emit colour intensity on a curve similar to that which our eyes perceive...

 

[CSoup]

<< the human retina can be compared to a field of various spikes. There are thousands of "rods and cones" scattered accross it. Each rod or cone sees one color, red, blue, or yellow, sort of like the inverse of a CRT screen. However, i would liek to see that document about why CRTs cant produce certain colors >>



Ok, you will need to take a color science, graphics, or perception course to get a better explanation, but here is the idea. Rods basically are highly sensitive and detect luminance (essentially brighness). Cones come in three types that have there centers of sensitivity at three different wavelengths. These are S, M, and L (for short, medium, and long). Each of these cones have a response curve. There is a set of curves created from a set of people that is considered the standard mapping of human visual response. Basically the human gamut is the possible colors that are perceived under the constraints of these reponse curves. Now, the reason that three colors can be combined to caused many perceived colors is not that they cause the emisssion of the same wavelenght as some color, but that they cause the visual system to respond in the same way. Different "signals" that give the same visual response and thus color interpretation are called metamers. The reason that CRTs cannot produce all colors is that if you take their outpout and map it onto the response curve of the human eye, you can't find a mapping for all possibilities. Basically, the mapping function is not one-to-one. To be able to produce all human visible colors, you would need one phosphor that is able to subtract light, but that is a physical impossibility. In doing color conversion, you normally allow the possiblity of a negative light source in doing the linear transformation, but then it is mapped onto a nearest neighbor that is displayable. Normally you use a nearest neighbor method that preserves hue (basically the color component).