Complementary confusion

Discussion in 'Digital Darkroom' started by thommy_andersen, Jul 5, 2014.

  1. Hi
    I'm trying to understand how to use complemetary colors in my photography or post-processing. In RGB for example we have red - cyan on opposite side of the color wheel and complementing each other, but often the complemetary color to red is said to be green as in RYB.
    Which one do I as a photographer have to consider? Do we use RGB table or RYB table?
    Thanks!

    Thommy
     
  2. What is RYB? Red, Yellow Blue? In terms of a color model what we deal with in photography and imaging, it's RGB.
     
  3. RYB = http://en.wikipedia.org/wiki/RYB_color_model
    - subtractive, mostly for painting, "predates modern scientific color theory, which argues that magenta, yellow, and cyan are the best set of three colorants to combine"​
    modern color modes in Photoshop include RGB, CMYK (for printing, mostly), and the ever-intriguing Lab.
     
  4. Hi, in photography we generally use red, green, and blue as our primary colors. Then we generally use cyan, magenta, and yellow as the complementary colors.
    In the artist's world, red, yellow, and blue are often considered to be the primaries. But there is a school of thought that the artist's "red" is closer to the photographer's magenta, the artist's "blue" is closer to the photographer's cyan, and yellow is yellow. In other words, the artist's primaries, irregardless of what they are named, may actually be more similar to the photographer's secondaries. Thus, if the artist's "red" has green as its complement, this sort of agrees with the photographer's magenta-green complementary pair.
    In real life it's not quite this simple. Many photographers seem quite smug that it's RGB, period, no two ways about it. But a problem shows up if you demand that they "define," for example, the color "red" in terms of the light spectrum. It turns out that there is no such strict definition, so "red" can largely mean whatever you choose it to mean.
     
  5. I'm not sure where a problem shows up there; the segmented RGBCMY wheel is a simplified mental model. There's no need to "define red" because a single definition of red is irrelevant.
    Is it reddish? Then it's red, and its complement is cyan. If it's more orange, then it's going to be more blue. That's the whole point of the model.
     
  6. Is it reddish? Then it's red, and its complement is cyan. That's the whole point of the model.​
    So "magenta" is also red (it has a reddish appearance)? And therefore the artist's red-green commplementary pairs are correct?
    So now, what is the "true" complementary color to red? Is the complement cyan, like you say, or is it green, like the artist says? Do you see the issue?
     
  7. So now, what is the "true" complementary color to red? Is the complement cyan, like you say, or is it green, like the artist says? Do you see the issue?​
    Cyan is the complementary color to red, at least as far as the color models we use within nearly all desktop applications.
    Red is just an English word we use to define a visual sensation. And is this red within human vision, red within a fixed RGB color space? In the end, at least in terms of our computer systems, there's no red, there is a set of numbers and numbers alone don't tell us what a color looks like.
     
  8. Complimentary colors are those that when mixed together result in a neutral gray, white, or black. In an additive
    system like RGB this is really simple to figure out because mixing colors (i.e. adding colors) is as simple a
    adding the individual components together.

    If you want to know the complimentary color of this RGB (150, 0, 0) you simply supply the missing components
    that add up to neutral. In this case RGB (0, 150, 150), i.e. cyan. For something a little more complicated like
    RGB(150, 223, 20) you'll end up with RGB(73, 0, 203). If you set your info palette to HSB you'll see that these
    two colors have hue angles 180º apart. In other words they are on opposite sides of the color wheel.

    Photoshop's invert command does things a little differently — rather than adding up the nearest neutral color, it
    finds the color that adds up to white. So the invert of RGB (150, 223, 20) is (105, 32, 235), which when added
    together results in white. This causes light colors to invert to dark ones, but the hue angle still changes by 180º.

    You can confirm the additive math by setting two colors are individual layers (with no background layer) and
    switching the top layer to Linear Dodge (add). The result will be the sum of the layers. Put [150, 0, 0] on top of
    [0, 150, 150] and you'll get [150, 150, 150]

    Just keep in mind that by tying the concept of complimentary colors to the math of an actual color space, your
    complimentary colors are essentially device dependent. If the space is not perceptually uniform, the
    mathematically 'correct' complimentary colors in that space may not be visually complimentary. But in practice,
    it's going to be close enough for well behaved working spaces.

    Also keep in mind that the lexical categories in which we place colors is cultural. Not all languages agree where
    blue stops and green starts. Some languages use one word for both colors. So it's silly to argue about the
    exact line between magenta and red. It's a continuum that is arbitrarily segmented by our language.
     
  9. Not all languages agree where blue stops and green starts. Some languages use one word for both colors.​
    Sort of. While cultures at different levels of technological complexity have different numbers of primary colors, the divisions at any given number are fairly consistent in terms of wave length. This understanding stems from the seminal work of Berlin and Kay (1969) discussed briefly at http://en.wikipedia.org/wiki/Color_term
     
  10. JDM, I'm pretty familiar with the Berlin and Kay work. I think most people interpret the subject in terms of
    Prototype Theory — i.e. groups with the words blue and green in their vocabulary will form some
    agreement on the ideal blue, but where blue becomes green is less certain.

    There's a good overview here: http://www.blutner.de/color/Color_Words.pdf

    and a fun wikipedia article on cross-cultural blue-green terms here:
    http://en.wikipedia.org/wiki/Distinction_of_blue_and_green_in_various_languages
     
  11. Thanks for all great feedback!
    Still not 100% convinced what to use and how in my photography. Here are some articles and videos about how to use and make benefit of the complementary colors.
    And all stated that red and green are complementary colors :

    http://www.digital-photo-secrets.com/tip/1128/make-your-photos-more-colorful-with-complimentary-colors/

    http://www.digitalcameraworld.com/2013/03/08/color-photography-understanding-complementary-colors/

    https://www.youtube.com/watch?v=4163nHq_KkI
    I would say that 90% of all photo related articles, tutorials, videos etc I have seen claims red and green to be complementary.

    Why is that?
     
  12. I would say that 90% of all photo related articles, tutorials, videos etc I have seen claims red and green to be complementary.

    Why is that?​
    They are confused?
     
  13. To see it visually, if you have a Mac, download ColorSchemer software. If you have a PC, download TigerColor's ColorImpact. You can demo the software and see the colors in real time.
     
  14. # Andrew - then I'm not alone :)
    # Rich - great software tips - do you know any freeware of same caliber?

    I'm using Adobe Kuler and even their RGB wheel using color rule "complementary" with value 255 0 0 gives green as complementary.

    Even Adobe are confused?

    Maybe someone here is Santa Claus? Or is everyone lying about him as well?
    BTW - is his clothes red?
    If yes - why do we often see green complementing it as accessories around him?

    Thommy
     
  15. do you know any freeware of same caliber?​
    Nope. ColorImpact is only $40. Color Schemer is $50, but it does so much more. It's what I use. Your not stuck with just complements. You can see split complements, triads, tetrads, analogous and mono. You can create your own schemes, export them as .ase files to use with PS, IND and IL. You can download premade schemes, pull the colors out of a photo to create a new palette, take one color and build a scheme, create schemes for the color blind and a whole lot more. It's worth fifty bucks. I use it almost every day. It pays for itself in time saved.
     
  16. "I would say that 90% of all photo related articles, tutorials, videos etc I have seen claims red and green to be complementary"

    There are many ways to relate colours to each other. The one a bit more technical than the other. None of them 'correct'.
    Red and green are often said to be complementary because a tiny addition of the one in something having the other as main colour intensifies the perception of that other colour. It complements, not in being an numerical opposite or something like that, but because a bit of red adds to the sensation of greens, and vice versa.
    Is that correct? Perhaps. Perhaps not. The entire colour scheme/wheel thing, no matter what version, is a construct that is propped up by theory more than by evidence. It works anyway (as long as you don't go looking for the real truth about 'three plus three' colour). So it doesn't matter.
     
  17. I'm using Adobe Kuler and even their RGB wheel using color rule "complementary" with value 255 0 0 gives green as complementary.
    Even Adobe are confused?

    Adobe seems to have made a conscious choice of using the painter's traditional primary scheme of RYB for Kuler. There's a discussion here: https://forums.adobe.com/thread/75007.
    This choice leads to some weirdness, for example the hue angles reported don't correspond with the actual angle on the circle and the reported "complimentary" colors when mixed result in a third color rather than canceling each other out.
     
  18. #JDM:
    I'm intrigued! Please help me understand this phrase: "the ever-intriguing Lab."
    I'm working on a profile for b/w printing (using QuadTone RIP if you're familiar with that program) I'm supposed to enter the Lab values of various shades of grey. I'm using Elements, which does not have that mode, unfortunately. So, just what is "Lab"?
    Many thanks, and
    Intrigued,
    Paul
     
  19. So, just what is "Lab"?​
    It's a device independent color space perceptually based (well that's the idea, it isn't fully) on how we humans see color. It's useful for some tasks, not others and it was designed many years before Photoshop, image editing and color management existed. When QuadTone RIP asks for Lab values, it's doing so because the numbers are not based on RGB or CMYK which are device dependant color spaces and the numbers would differ greatly depending on the 'flavor' of RGB or CMYK and that would be a big problem. Lab numbers are not ambiguous. R45/G98/B129 is ambiguous as I've as yet not told you the color space (sRGB, ProPhoto RGB, Adobe RGB (1998), Epson 3880 Luster RGB). Each RGB color space above will produce a different color with the same set of RGB triples. With Lab, L89/a0/b0 is one color description, no ambiguity.
    See: http://www.ppmag.com/reviews/200504_rodneycm.pdf
     
  20. Thanks, Andrew, very helpful. Your article covered more than LAB... to tell the truth I never understood those other color spaces either - until now!
    Sounds like you're familiar with QuadTone RIP also...
    Regards,
    Paul
     
  21. They got there first. It's been called something like the "desert colorspace" because it works especially well with canyons and desert images. Dan Margulis has written a whole book about it called
    2006, Photoshop LAB Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace. Peachpit Press.
    Here is a Lab-adjusted Lab
     

Share This Page