4

Ok, so I understand the eye has 3 different types of receptors and I've seen the process of converting from RGB to CMYK. However if in physics I can specify a color using a single number (its frequency) why do computers need 3 (RGB) or 4 (CMYK) numbers to specify the same color? Is there some kind of disagreement in the amount of degrees of freedom involved?

Qmechanic
  • 201,751
fuchini
  • 41

2 Answers2

2

Color is a perception that lives in the brain. Not all colors can be specified by frequency. For example, pink is not a single frequency. In fact, very few colors can be specified by frequency. Strictly speaking, none of the pure single frequency colors can be specified by a standard (in color and computer hardware science) RGB triple. This is due to choices that were made in defining standard red, blue, and green. Those choices were made based on available technologies for producing colors (in other words, what phosphors were available, etc).

The brain takes the three inputs from the receptors, and in some mysterious way converts that information into a perception.

garyp
  • 22,210
  • It may be worth noting that when light colors are modeled as mixtures of red, green, and blue, such a model works pretty well when combining them with each other or other lights which are not so modeled, but a CMYK model for transparent colors only works for modeling interactions with RGB-modeled light or other CMYK colors; it does not work well for modeling interactions with other colors. For example, a passing "red+green" yellow through a cyan filter will yield green, but passing sodium yellow through a cyan filter will yield dimmer sodium yellow. – supercat Jul 02 '14 at 21:58
1

The problem is "color" is an ambiguous term, depending on the context it can refer to

  • An individual wavelength
  • A spectrum of intensity verses wavelength
  • The response of the human eye to a spectrum

A laser produces light that is very close to being a single wavelength, plain LEDs are fairly narrow too*. Most other light sources produce a much broader spectrum. Most colors we see in day to day life are not individual wavelengths, they are the result of taking a light source with a very broad spectrum and filtering it down to a spectrum that is not quite so broad (but still much broader than that from a Laser or plain LED) and often has multiple peaks.

The human eye has three types of cone, so we perceive colour tri-chromatically. There are many many different combinations of wavelengths that trigger the same response.

By carefully choosing three primary colors of light (notionally "red","green" and "blue") we can "mix" most of the colors that people see in everyday life. It's not perfect, whatever combination of primary colors we choose, there will be some colors that the human eye can see, but we can't mix. Most of the time though it's "good enough".

In some crude sense, CMY mixing is the opposite of RGB mixing. We start with white light, and then cyan ink subtracts red, magenta ink subtracts green and yellow ink subtracts blue.

Unfortunately while with RGB, we could pick light sources with relatively narrow spectra, with CMY mixing we are stuck with the light we have. This means that a CMY red, green or blue has a much broader spectrum and hence a less-saturated color than a RGB red, green or blue.

Another problem with CMY mixing is that, even if you layer all three of the inks some light still leaks through, this leads to dark colors (particularly black) being poorly represented in CMY. To fix this a separate Black (called Key) is added to the printing process, giving CMYK.

* Note: Some LEDs (particularly those used for general illumination) are coated with a phosphor coating to give them a broad spectrum.

Peter Green
  • 1,171