Software color doesn’t agree

Artists face challenges mixing colors, especially in portraits where it is critical to get the subtle color of cheeks, lips and hair just right in relation to each other as the form of the head turns. Because of the immense variation in lighting and coloration from person to person, useful skin colors do not come straight from a tube of paint. Skin tones are never a fully saturated red or orange. Rather, they are lighter tints or darker shades of colors, meaning that each tone is a lighter or darker version of the well-known hue, albeit at a lower saturation and different value.

The skilled artist needs to be able to mix any color at will, and to do so we have been taught to identify the hue, find the corresponding tube color, and mix a tint. I call this the Hue-to-Tint method. It depends on understanding the traditional artist’s color wheel because when you know a colors compliment, you can generate a tint by mixing the compliment (or lighter and darker tints of that compliment as necessary). For example, if you identify a light pink cheek color, you go for a cool red, which will be darker, and both neutralize and raise the value with a light green. 

That’s the theory anyway. In practice it’s much harder because the tube colors don’t match the chromatic color wheel on either side, and there ends up being a lot more experimentation and happy or unhappy accidents along the way.  Being a technologist I turned to my computer to see if I could generate more helpful color charts to aid in color mixing. What I ended up discovering was a better method of mixing I call Tint-to-Hue method, which I’ll explain a subsequent topic. First some more context.

Having been trained as an artist in the traditional color wheel model that uses the three ‘primary’ colors, i.e. Red, Yellow and Blue to mix the ‘secondary’ colors, i.e. Violet, Orange and Green, as well as other ‘tertiary’ colors, I decided to blend these colors in my computer to get neutrals. The traditional theory states that when you mix compliments together, they neutralize each other, resulting is a hue-neutral grey in the middle (i.e. zero hue saturation grey). We learn this in art class and take it for gospel.

When mixing complimentary tube colors, say a red with a green, we therefore expect to get a neutral. In practice, I would get a tint since any mixture is more neutral than the original colors. I assumed the color theory was at work here, not realizing that mixing tints is not the same thing as getting a pure neutral grey. Because tube colors rarely come as pure and saturated as what we find in color wheel charts, it’s hard to discern a problem with the theory. If we don’t get a true neutral, we just assume it’s a function of the offset of the tube color from the theoretical color.

I took the traditional theory for granted in my painting and would often pull the compliment of a color to neutralize it. By not using black or a grey as the neutral to mix with, the resulting tints are more interesting than just adding black and/or white to the mix anyway. You get a richer effects, analogous to chords in music where you can ‘hear’ several tones at once. Thus the lack of true neutrals in mixing traditional compliments actually can be artistically interesting – in some cases a happy finding. So if the results are good, never mind the means. So we shrug off any theoretical differences when looking at a successful work of art. The problem is that accidents are very repeatable, and the next round may be less successful.

Any artist who has tried to mix colors know that by trial and error, and sometimes with great frustration, we eventually get something believable. It is an axiom among seasoned artists to say ‘the color doesn’t have to be 100% accurate; it just has to be believable.’ In fact, take a look at many of the master works and you’re see that while the colors are believable within the world of the work, they are not accurate. So the professional learns to mix progressions of colors that are believable, and then the traditional color wheel and its theoretical neutrals are just guidance; it’s not critical they be true. In other words, the seasoned artist creates a personal mixing methodology that works, and then the correspondence to a color wheel theory is irrelevant (unless of course, they have to teach it).

Artists respond and compensate on the fly. We are not scientists. We live with paradox, compensate with contrasts, all of which contributes to artistic impact when done skillfully. We are masters of compensating for inaccurate color. Said another way, accuracy is not our measure; artistic impact is. And you can have great impact with inaccurate color. The few times I’ve watched masters at work, the really interesting things they do is ineffable. I learn more by watching them paint that listening to what they say.

So when I got scientific with color on my computer’s color picker, expecting the traditional color wheel to be validated, I was surprised. The traditional color wheel’s compliments – like green and red, orange and blue, yellow and purple – did not in fact neutralize.

As I played around with a graphics program (Adobe Illustrator), I created two colors and had the software blend them from left to right. For example, orange and blue are compliments in the traditional model. So mixing them should produce a few less saturated tints and a neutral in the middle, right?

Wrong. The color theory taught for generations was not being validated by the software:


To my chagrin, the middle colors (from left to right) were not the neutral I expected. The tints were interesting enough, but they were not really something I felt I could predict logically. For example, in the blue-orange blend, the middle color looks like a dark violet, which we should be getting by mixing red and blue instead. That was confusing. The red-green blend looked a little more predictable, i.e. a warm green. But yellow and violet were unexpected – we should be looking at something more greenish, as you do when you mix yellow with black?

‘Hmmm,’ I thought, ‘maybe paint mixing artists work in another color space?’ What color space does my computer have?  Knowing a few things about graphics programs from my previous life as a graphic designer, I used the Hue-Saturation-Brightness (HSB, also known as HSL) model in order to explore the numbers of the hues. I soon realized that increments of 30 gave me all the primary, secondary and tertiary colors in the traditional model. Since the hue numbers go from zero to 360, it became obvious that the computer also has a color circle (zero and 360 are equivalent, i.e. red). Logically it would follow that the colors that are 180 degrees from each other would be the compliments. For example:


Is this true? Let’s blend these colors and see if they result in neutrals:


To my delight, the middle tones were all a perfect zero-saturation grey, meaning there is no saturation in them. The computers compliments did in fact perfectly neutralize each other. However, the HSB compliments are decidedly not the traditional compliments. At this point it was clear that the software model confirmed the theory that compliments neutralize each other. However, the definition of the complimentary colors themselves was decidedly different than the traditional model. Why are the two models so at odds with each other? Are they inherently different realms, or is one of them wrong? What did the software engineers know that the artists didn’t pick up on?

The answer to the last question will make for some interesting research. Who came up with the first color wheel decades or even centuries ago, before the time of software? Was it driven by the mathematics of light spectrum, or by experimenting with mixing pigments and eye-balling the results? I have a hunch that hue-shift on tint may have had something to do with the state of the traditional model.

If the old model is broken, why aren’t our modern masters complaining about it? I have a hunch about this one: masters get so good at color mixing it becomes instinctive. They reach for one color or another based on what they are seeing, not on what they are thinking about seeing. Like athletes who just act to make the crucial shots, instead of thinking about them.

In other words, color wheels are for newbies, training wheels that have little to do with the actual experience of creating master works. The masters ‘just do it’ until it looks right. They get to know each color and how they blend and tint. Theory is too much mental work when you know how to get the desired result! It also helps to note that the old theory isn’t completely broken. As illustrated above, you can get some very interesting tints by mixing colors that are not mathematically speaking true compliments.

Okay, so the computer thinks differently. When has that ever influence how artists think? Should it? Moreover, can it be proven with actual color mixing? (We’ll experiment with this question later.) Well, being an artist myself, I can attest that it matters a lot because a solid methodology can greatly simplify the color mixing process. It can also save us time and money. We can drive practical lessons from the computer model, which, after all, is based on the same light spectrum that governs our pigments. Fewer colors to mix means more time painting and fewer materials used, as we shall see with the Tint-to-Hue method. A consistent methodology also contributes to sanity because no, dear student, it’s not that you don’t get it. It’s possible you were taught an inaccurate color wheel in the art books.


Copyright © 2014 Roy Zuniga

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s