DAN MARGULIS APPLIED COLOR THEORY

Finding the Inverse Color

From: "Matthew"

Date: March 7, 2010 5:33:39 AM HST

Subject: [colortheory] inverse colors

I tried an experiment, which sort of failed, and I'd like to know why.

If red and cyan are complimentary, wouldn't it stand to reason that the inverse of red is cyan? I took a swatch of r255,g0,b0 and inverted it. It looked cyan, however, when checking its color in cmyk values, I got back c52,m0,y13,k0. Shouldn't it have been c100,m0,y0,k0?

It went like that for the compliments to green and blue as well. I also tried it the other way around, starting with c100,m0,y0,k0 in a CMYK document, inverting and came back with c0,m100,y100,k100. In that example it seemed like it just reversed all of the numbers, which after checking the Photoshop documents is what the inverse function will do.

Is there a way to properly invert a color in Photoshop, with respect to additive color?

Thanks,

Matt Paquin

_________________________________________________________________

From: Michael Jahn

Date: March 7, 2010 9:45:16 AM HST

Subject: Re: [colortheory] inverse colors

Hi Matthew,

Without getting too technical, since RGB is additive, and photoshop is

designed to convert RGB values into the closest reasonably CMYK that could

be expected to be printed, you will not get the round trip transformation

you are wishing for.

There are many colors that can be represented in a additive system such as

RGB color mode that can't really be represented identically in a subtractive

system such as CMYK color mode.

So, 255r inverted would actually be 255g 255b. Since RGB color space is

normally "viewed" using some backlit display, there are many colors that

really can't be represented properly using ink on paper with reflected

light.

In addition to this, the relationship between what you might have started

with in RGB - to what you will get in CMYK - is dependent on what you have

in your color settings (In Photoshop, under the Edit menu, select Color

Settings)

To help you visulize what is going on here, create a new RGB image - and

create two colors in it

one with a swatch of 255r 0g 0b with this inverted ( that is 0r 255g 255b)

open up the info window (under Window, select Info)

change the First Color Readout to "Actual Color"

change the Second Color Readout to "RGB Color"

now, when you convert that RGB mode to CMYK mode, look at the values

On my computer, I selected "North American Prepress 2" and as my color

settings for this experiment

the red square now 96% Magenta - 90% Yellow - but the RGB values - 205r 50g

51b - even if i go in and change the red while i am in CMYK mode to 100

Mageta 100 Yellow, the RGB only gets to 204r 34g 41b not the 255r 0g 0b you

were hoping for, and this is perfectly logical, as you can print something

that looks like 255r using Magenta and Yellow in on paper.

Hope that helps you understand.

If you want do a shallow dive;

http://en.wikipedia.org/wiki/CMYK_color_model

A little deeper dive requires you understanding color space in general

http://en.wikipedia.org/wiki/HSL_and_HSV

Michael Jahn

___________________________________________________________________________

From: Dan Margulis

Date: March 9, 2010 3:19:19 AM HST

Subject: [colortheory] Re: inverse colors

Matt writes,

I tried an experiment, which sort of failed, and I'd like to know why.

If red and cyan are complimentary, wouldn't it stand to reason that the inverse of red is cyan? I took a swatch of r255,g0,b0 and inverted it. It looked cyan, however, when checking its color in cmyk values, I got back c52,m0,y13,k0. Shouldn't it have been c100,m0,y0,k0?

No. This procedure doesn't work across colorspaces because of compensations for ink impurities, dot gain, gamma, and the gamuts of the two spaces. Here, 255r0g0b is outside of the CMYK gamut and 100c0m0y0k is outside of the RGB gamut. There's also a luminosity issue. In any RGB, 255r0g0b is vastly darker than 0r255g255b, its inverse. So converting it to CMYK is likely to get a value in the 50c range.

It went like that for the compliments to green and blue as well. I also tried it the other way around, starting with c100,m0,y0,k0 in a CMYK document, inverting and came back with c0,m100,y100,k100. In that example it seemed like it just reversed all of the numbers, which after checking the Photoshop documents is what the inverse function will do.

Is there a way to properly invert a color in Photoshop, with respect to additive color?

It's not clear what a "proper" inverse is, or what it would be used for if we did know what it was. Suppose that we say the definition is that when the inverse color is on a layer at 50% opacity above the original color, the result is 50% gray. Or, possibly better, that when the inverse color is on a layer at 50% opacity in Color mode, the result is a grayscale version of the document.

If that's the definition, and we're talking about photographs and not swatches of brilliant colors, it can only be achieved in LAB AFAIK, and even then it's of limited use. For example, what's the inverse color of bright yellow? Depending upon how you define the yellow it's going to have a value of 95L-98L. Invert that and it calls for a brilliantly intense blue that simultaneously is black, IOW an imaginary color. Invert it and put it on a Color layer and it calls for a brilliantly intense blue that's nearly as light as a white, which is also an imaginary color. So although LAB can theoretically create these "inverses", it's not obvious what the purpose would be.

Dan Margulis