4

It's not immediately obvious, but when you visit Adobe Color aka "Kuler", you'll notice that the hue of the color-circle is not evenly spaced:

hue circle comparison

As you can see in the image above, a regular hue spacing of 120° should result in Red, Green and Blue, when starting at 0°. Over at Adobe, there's a much bigger span between yellow and magenta, covering almost 180°. So if we point our "triad" at red, we won't get Green and Blue, but a Yellow and something closer to Teal instead.

That discrepancy is also reflected in the numerical-output. If you switch to the "Triad" mode and choose "HSV" as color-mode, you can easily observe this in the generated hue-values.

Point one "arm" of your triad to red (eg. Hue = 0) and you'll get hues of 59 and 203 for the other arms. So this is not even close to a 120° spacing. It gets more regular when you select a hue of 180, resulting in 316 and 50 for the other arms.

This phenomenon is also visible in the image, the hue-circle seems to be stretched at the orange colors and compressed at Green/Cyan/Blue

Is there a name for the effect that can be observed here? I've looked around for hue scaling but wasn't able to find anything. Since I was trying to replicate the color-harmonics in code, any mathematical explanation of what is going on here would be highly appreciated too.

1
  • Most likely your seing a normalized LCh variant (either LCh_ab or LCh_xy)
    – joojaa
    Mar 5, 2020 at 12:55

2 Answers 2

3

Color harmony should be a visual thing and Adobe has tried to implement a math rule for producing visually harmonic color combinations. That math rule isn't shown as calculation formulas, we can see only the result - tweak one and the rest move.

The RGB color system nor it's HSV math presentation aren't models for how we see colors, but how the colors in the computer screen are produced. The parameters H,S and V (=B in Photoshop) are not visually independent. Change for ex hue, the perceived colorfulness and lightness vary.

I guess Adobe's color harmony generator isn't trying to place colors symmetrically on the color wheel, but tries to find combinations which are visually in balance for presentation purposes. As I wrote, the used rule isn't shown as math.

Seemingly the triad strategy that you considered faulty because the colors aren't uniformly distributed on the color wheel lets us tweak the result infinitely, but the hue differences in HSV are not 120 degrees. I guess this is caused by the complexity of the rule under the hood. I guess actual calculations are made in some preceptionally uniform color mode and the results are converted to HSV.

But the color wheel in the harmony generator is matched with their rule. There the angles between the colors are 120 degrees.

Adobes harmony generator gives also CMYK combinations. Unfortunately they assume that computer screen RGB color range is printable. What color is actually printed with certain CMYK numbers depends on printing process. Also Lab mode allows you to input numbers which are not displayable in RGB screens, the shown result is clipped to displayable range, but the numbers are not. These things show that Adobes harmony generator should be treated with care. I guess it's useful in HSV and RGB.

ADD: As a related idea I tried to make an uniform color wheel where all colors have same lightness and chroma and all of them are originally displayable in sRGB. Try to make it more colorful or brighter then something will be clipped. It's this (or what's imgur has left)

enter image description here

The colors are taken from GIMP. Its LCH color system is a polar implementation of the same ideas that Lab is based on. It was a linear gradient with stops at 0, 60, 120, 180, 240, 300 and 360 degrees in LCH hues. It's bent to circular. L and C are constant and so low that sRGB can show everything.

The hues of HSV system are placed differently than in the usual HSV circle. The dots show where are HSV hues 0 (=red), 120 (=green) and 240 (=blue) degrees. Their spacing is far from 120 degrees in this wheel.

At first the placement of the dots surely seems to be a damned lie. My blue dot is placed on something which is quite reddish. But try it yourself. Open the color selector in Photoshop or GIMP and take for example pure full bright blue RGB=0,0,255. In HSV (=Photoshop's HSB) that's hue=240 degrees. Turn the saturation lower and see how the color creeps towards violet. Nobody would claim it's blue. This shows that in RGB system hue, saturation and brightness are not visually independent. Lab and LCH are attempts to fix it.

1
  • Thanks for your answer. I was also suspecting something along these lines, that the values are skewed towards our color-perception instead of an evenly spaced hue distribution. I would be very much interested in a name/formulae for this skewing though. Maybe there's some research on it?
    – bummzack
    Mar 5, 2020 at 9:57
2

I'm currently investigating this issue.

To state my problem, it is that, with a different colour wheel, you get different suggested colour harmonies.

In the Adobe wheel, red and green are 'complimentary'. However, with an RGB wheel. Red and Cyan are complimentary. I personally find this more believable to the eye - they calm each other whereas green and red provoke each other. And also, in reality, red and cyan paint will create a grey, whereas green and red mix to create a brown (That is, a darkened colour with a yellow hue).

See this helpful article. https://www.wtamu.edu/~cbaird/sq/2015/01/22/why-are-red-yellow-and-blue-the-primary-colors-in-painting-but-computer-screens-use-red-green-and-blue/

What I have found is that a significant number of colour wheels are very misleading - using RGY as primary colours, rather than CMY and RGB. The answer above seems to be a little wrong to state that RGB does not replicate our vision, because our eyes literally have conicles that detect either R,G, or B light. I suppose it is fair to say that the RGB colour models used for computer screens don't cover all the colours that are visible to us. Unfortunately the reason for this is beyond my current knowledge.

I'm sharing this answer to try and help the conversation forward. I think it is correct to say that 'colour harmony should be a visual thing', though clearly this is verging on a meaningless statement. Ultimately, you're saying something like 'go with your gut'. But, it is incredibly helpful to have some way of understanding why the tools which purport to give us 'colour harmonies' produce such different results.

For now, I find that a CMY/RGB colour wheel, with even spacing gives me a more concrete basis for determining colour harmonies than the Adobe wheel. (See paint comment above for basis in reality - and this demonstration - https://www.youtube.com/watch?v=jQqxN8LpGzw&t=29s) Frankly, the colour harmonies presented by the Adobe wheel sometimes feel extremely un-harmonic. However, my next step on the journey is to understand this underlying 'formula' that the Adobe wheel uses and the reasons for it. It may be that my mind is changed.

So, unfortunately, I have no direct answer to the question in terms of a mathematical formula. But, appreciate others noticing the discrepancy.

6
  • The normal rgb color wheel is almost certainly wrong. And color harmonies people use have allmost no reason or rhyme.
    – joojaa
    Sep 24, 2022 at 11:57
  • See true eye sortof has red yellow and green. But the colors are tied in pairs reading positive and negative signal leading to a system that does not replicate standard rgb wheel. See blog.asmartbear.com/color-wheels.html its pretty accurate as to how the eye works. But thing is color wheels are really tools for utilizing your mix space well not really good for finding the best colors possible at all. So just vecquwe its often misused eoes not mean its right. In fact there is little justification to most color harmony picking.
    – joojaa
    Sep 24, 2022 at 12:02
  • ...!? "The normal rgb color wheel is almost certainly wrong" ...!? It is one interpretation, one model for some applications. It is not "almost certainly wrong". Is that how physiology works? No. Is it the best model? Depends. Is it almost certainly wrong" I'm repetitive here... No.
    – Rafael
    Sep 24, 2022 at 13:58
  • @joojaa Thanks for the comments. Wrong is a little hard. My hypotheses is currently that it is like music. The musical scale does not have a scientifically even distance between each note. But, if it did it would sound out of tune. The LCh wheel used by color.adobe and illustrator has a scientifically equal distance between each hue. But, to my eye, it is out of tune. Interestingly coloors.io uses and RGB wheel, whereas Adobe uses LCh. Go check out the harmony results they give.
    – Andelad
    Sep 26, 2022 at 10:16
  • @Rafael Wrong as in something is missing in the explanation. Its certainly possible that the entire rationale for color harmonies is just snake oil. It does not mean it does not work, but any other system might work too. So like sound tuning it might be that we are just so used to the way your doing it that it feels right. But there might not be any justification for it. After all the RGB wheel wasnt used untill very recently and thats just the simplest possible polar transform of rgb done by people whis didnt take a lot of deep color theory classes.
    – joojaa
    Sep 26, 2022 at 11:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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