5

I could find numerous resources that specify how colour spaces such as RGB, sRGB and CMYK cover a part of the CIE XYZ Lab colour space.

I could find numerous graphics displaying the RGB and CMYK gamuts as an overlay to CIE XYZ (such as this). My suspicion is that HSV/HSL can be used to indicate the same colours as RGB, but is there any authoritative resource that says whether or not HSL/HSV can cover less colours than CIE XYZ, preferrably with an analogous graphic as the RGB ones I found?

I am not convinced this is the right SE for this question. Photography SE might be more appropriate, but not quite; maybe Math SE or MathOverflow? If you think I am wrong here, please move the question to the appropriate site, if possible.

  • Hsl and Hsv map to RGB in practice. But techically they do not have to map to RGB They can map to CMY or other spaces too... Anyway neither hsl/hsv/RGB/CMYK map to lab, devices do. The lab space is special here since its the only one that actually maps to defined physical measures by default. So how the other spaces map to lab depends on your image/device profile and gamut. So its a bit hard to compare to say RGB as there are different sized RGB spaces. I would say this belongs to some forum not present in stackexhange – joojaa Nov 24 '14 at 17:31
  • @joojaa: "I would say this belongs to some forum not present in stackexhange" - a ... "Visualization SE", or "Graphics Theory SE"? Would this be the point for a proposal in Area 51? – O. R. Mapper Nov 24 '14 at 18:16
  • well its kind of between physics.se and the proposed print.se – joojaa Nov 24 '14 at 18:33
3

CIE Lab is hard to wrap the wits around, because it defines a color space that includes such things as, for example, a theoretical fully-saturated yellow at 0 luminosity, which is perceptually indistinguishable from black, all the way to (e.g.) bright blue at 100% luminosity, which is indistinguishable from white. From that, once you get the (esoteric, to be sure) concept, it's easy to extrapolate that Lab permits entire ranges of color to be defined that can't exist in CIE XYZ, the original CIE RGB, or HSB/HSL/HSV, which are based on what the average human eye can actually distinguish. (CIE XYZ is simply a mathematical representation of the original research work on human RGB perception done in the late 1920s.)

The huge conceptual difference between CIE Lab and other color models is that Lab is, by design, entirely device-independent. It is a sort of meta-color-space that happily defines colors that cannot be seen. You could argue that since color is a visual perception, this is a contradiction in terms, but CIE Lab allows any perceivable or realizable color to be very precisely defined, regardless of how it is created.

The RGB and CMYK models and their related color spaces are not device-independent. You can't define on-screen color as CMYK, because monitor colors are additive. Stage or movie lighting is RGB for the same reason. RGB doesn't work for inks or paint, since these are subtractive.

All this makes Lab sound like a theoretician's delight and a designer's nightmare, but there are many cases (especially in Photoshop) in which the Lab color model allows operations that are impossible in RGB or CMYK, because Lab completely separates luminance from color, unlike the other models available in Photoshop. There is no way in RGB or CMYK to change, saturate or desaturate a color without simultaneously altering luminosity, and vice versa.

Pantone Solids are defined as Lab values in addition to their ink formulations, so that there is no ambiguity as to the intended perceptual effect of that formulation under standard lighting.

The best practical text on the subject for designers and photographers that I know of is Dan Margulis' "Photoshop LAB Color." The subtitle, "The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" gives a hint of the content. It's an important and unique book that's had a permanent place on my bookshelf since it was published. I highly recommend it for anyone who wants to get a practical, as opposed to theoretical, grasp on Lab.

If you want to dig into this further, Sareesh Sadhakaran has some excellent and lucid articles for the Wolfcrow.com blog, such as this one.

  • 1
    I would vote you up if i could figure out how this answers the question. Anyway you have the right points, color models are not device dependend. Except LAB which is a scientific scale. – joojaa Nov 25 '14 at 5:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.