Hue Shift on Tint

I’d like to build on what I’ve written in the previous blog posts about mixing colors more efficiently by identifying the hue of the intended color and then using an understanding of hue shift on neutralization of colors to trace the nearest tube color that can be used to match. There is a two-fold intent here:

  1. To establish a more systematic approach to understanding hue shift.
    • Is it true that neutralizing colors always shifts the hue towards the relatively cooler side of the color wheel? For example, neutralize yellow, and it looks greenish, not orangy.
  2. To provide an efficient color mixing process that artists can use to expedite painting, and spend less time mixing.
    • Painting time is precious, and unless the goal is just to mix colors, most artists will prefer to mix and lay down usable color quickly. Think here of the paintings of Sorolla or Sargent. Their color ‘matching’ may not be photographic or hyper-realist, but it is appealing and so effective that we consider them ideal. The way they were painted is the way they should have been painted. Too many of us get stuck on color matching and miss the dynamic action-response nature of painting that is a distinguishing characteristic of the medium. A robot can be programmed to observe, mix color and ‘print’ with paint according to the Munsell color system.

Here’s the ‘hue-shift’ hypothesis:

When a given color is mixed with a neutral grey of any value, the hue will shift towards the cooler neighboring color on the color wheel.

How do we test this hypothesis?

Plotting with computer scans

The simple way is to take a range of tube colors and tint/shade them to pre-determined values, and then read the colors of each with a digital color reader or scanner. The computer’s HSL reading will provide Hue numbers for each, and these can be plotted to show the hue shift. One person can do this, and I encourage readers to attempt it. All you need is patience, colors, white boards for the colors and a scanner. You can get the HSL values via the color picker. The end results might be plotted on a spreadsheet that, for each tube color, has the Value number (how light or dark it is) as rows, and the Hue number as columns. The Saturation number can be input into the cells for the colors measured. This will give the hue-shift profile for each tube color. Depending on how many values are recorded, we can potentially see a vector or curve of the hue-shift for that specific tube color.

One person can do this, and I encourage readers to attempt it. All you need is patience, colors, white boards for the colors and a scanner. You can get the HSL values via the color picker. The end results might be plotted on a spreadsheet that, for each tube color, has the Value number (how light or dark it is) as rows, and the Hue number as columns. The Saturation number can be input into the cells for the colors measured. This will give the hue-shift profile for each tube color. Depending on how many values are recorded, we can potentially see a vector or curve of the hue-shift for that specific tube color.

The end results might be plotted on a spreadsheet that, for each tube color, has the Value number (how light or dark it is) as rows, and the Hue number as columns. The Saturation number can be input into the cells for the colors measured. This will give the hue-shift profile for each tube color. Depending on how many values are recorded, we can potentially see a vector or curve of the hue-shift for that specific tube color. (In a future post I’ll submit an example chart.)

Note that while the same color swatches may scan differently in different scanners, as long as all the results are done with the same scanner we should get a reasonably accurate set of results. Monitor calibration does not factor in because we’re using the computer’s color picker to determine the HSL number for each. Note that due to pixelation, for any given swatch a few measures should be taken and the mean selected.

However, ‘objective’ computer color is not the same as perceived color. So here’s a subjective approach.

Plotting with user color matching

  • Preparation 
    • Take a set of tube colors that align with the primaries and secondaries on the color wheel.
    • For each, neutralize so that there is a light, medium and dark version. To avoid extreme desaturation, we don’t want to mix very light or very dark versions. On a scale of 1-10, perhaps a 7, 5 and 3 would suffice (but this can be modified during the experiment).
    • Put each of these individual colors on a color swatch, numbering them on the back with random numbers that are indexed to the tube colors on a chart the test users don’t have access to (so they can’t look at the numbers and correlate which swatches came from which tube color).
    • Set up a wall chart with the color spectrum in a horizontal grid, with chromatic hues in the middle, and lighter versions of the same hue above them, and darker versions below. Note that for each column the tints and shades of the same hue as the core chromatic color. These will have to be verified with a digital color reader to ensure that hue-shift isn’t skewing the chart itself.
    • Index each color on the chart with a number – it could be the Munsell number or something else that is easy to reference.
  • Validation
    • One at a time, give each user a set of colors across the spectrum (the count should be determined by how long the user can provide focused matching).
      • To get a good sample, ensure that in the end, all the swatches have been given at least three times to users.
    • Have the users match each of their swatches to a cell on the wall chart, and note the number of the cell matched.
    • Plot all user selections on a spreadsheet (see above) to determine hue shift for each tube color.

It would be interesting to have multiple instances of both types of experiments using a broad range of tube colors, and to bring the results together into a public knowledge base. The end result would be information of hue-shift for tube colors that artists can then use to mix colors faster. It can also help us rank tube colors in terms of how broad their hue mixing potentials are. How exactly would this work? Here’s a scenario.

Practical Application

  1. When starting a painting, the painter selects tube colors they want to use.
    • The artist might have preferred colors, or alternatively
    • A more sophisticated (future) color recognition software can propose tube colors based on the color profile of the image.
  2. The artist becomes familiar with the hue shift of each tube color.
  3. To match color:
    • For the beginner, the artist uses a Munsell color swatch to identify the target color, then uses the hue-shift chart to identify the tube color that will produce the closest match.
    • The advanced artist makes a mental note of the desired color, and from knowledge of the hue-shift charts for her palette, picks the corresponding tube color.

Conclusion

In publishing my Hue-shift based methodology for color matching, I’m ‘open-sourcing’ the idea so that the community can pick up on it and perhaps together we can be systematic about collecting this data. Software engineers may someday pick on this and create color-detection to tube color technology to expedite the process for the painter. Imagine you’re out in the field painting en Plein Air and you’re having trouble determining how to mix a color. Having the Zuniga hue-shift charts on hand can help. Alternatively, imagine a phone application lets you take a picture, zoom and select the problematic spot. Since you input the tube colors in your kit, it will tell you which color is the fastest path to obtaining the desired hue at the desired value.

— Roy Zuniga
Langley, WA

 

Copyright (c) 2017 Roy Zuniga

Neutralizing Colors Efficiently

I keep hearing art instructors tell students to neutralize saturated tube colors with their compliments. To neutralize Red, use the ‘compliment’ Green, for Orange use Blue, and Purple will tint Yellow. There are problems with this approach:

  • True compliments that neutralize colors are not what is popularly asserted. As I’ve written elsewhere (see Software Color Doesn’t Agree), what really neutralizes a red is not green, but a teal. Orange is neutralized somewhat greenish blue, and yellow by a somewhat purplish blue, as this chart illustrates (number correspond to hue numbers in the HSL computer color space):

neutral-blends-x

  • One of the reasons why the conventional advice evolved is that to the naked eye, it looks like the traditional compliments do neutralize. Take a Cadmium Red and mix it with Viridian and it will look neutral. So there is some truth to it, but it’s not a perfect neutralization. Keep this mind:
    • The blended color is not really a true compliment. It might not even be the hue of its label; it’s a tint of a warmer hue. The color you’re mixing with is not a tint of the hue you’re thinking of. That’s because hues shift towards perceptually cooler temperature as they are tinted. Therefore, a dark green like Viridian would look warmer if fully saturated and brighter.
    • Tinting with the ‘compliment’ means that a certain value (aka brightness) inherent in the added color is forced on the mixture, and it’s likely not the value you’re targeting. It’s very hard to mix traditional compliments for lighter value ranges, for example. If you want to neutralize a pink, you have to find a really light green. In practice ­one ends up mixing in white, and at that level of tinting, you can’t really discern the true hue anyway. So as long as it is close to the cool or warm we’re looking for, we’re satisfied.
    • Phenomena in perceived colors are projected into the color mixing process, and this doesn’t help. It’s true that with a cool light, the highlights will obviously be cool, and the shadows will look warm. We tend to think the light hitting the shadows is of a complimentary hue to the one hitting the lights, especially as the light is saturated. We get this phenomena in the extreme with sunsets – the very warm yellow setting sun will light up warm clouds where it hits, and the very same cloud will look blueish grey where the light does not hit. The color of the form is the same on the light and dark sides, but the complimentary highlight/shadow phenomena of light makes them look very different. Just because that’s how color is perceived doesn’t mean that’s how it has to be mixed!

Despite these tendencies, professional artists will arrive at good results, not because they are literally following the traditional compliments theory. Rather, through experience and a lot of mixing, they get what they want, and chalk up any deviations to the inherent ambiguity of artists’ tube colors. One manufacturer’s Brown Pink is a lot more orange than another’s. Not everyone can tell the temperature difference between a maroon and a scarlet.

In fact, we would benefit from an objective hue numbering standard for hue colors. For example, regardless of the traditional name, an additional hue number would be printed on the tube, along with some saturation and brightness rating. This could be mapped to a computer model, like Hue Saturation and Brightness (HSL). But that’s a topic for another day.

So how can we more efficiently mix the greys we’re after? The answer is actually straight forward, but it requires throwing out another rule that many painters who ‘tint with compliments’ swear by (including the Impressionists): don’t use black because it ‘muddies’ colors. Let’s now also dispel this scare tactic – a ‘muddy’ color is just another grey that finds itself in the wrong place. Any grey can be a muddy color in a given context. If you’re painting a sepia portrait, warm greys are wonderful. If you’re painting portrait in a cool light, the same grey will look muddy.

It’s ironic that artists who loathe black use white quite liberally. While both black and white are neutral colors and technically both have the capability of ‘muddying’ colors, artists don’t dump on white because at the lighter range, we’re much more forgiving of ‘wrong’ greys. We actually don’t see ‘muddy’ whites. Let’s call a spade a spade: all tints are greys, and ‘muddy’ colors are simply greys out of context.

So the problem is not about having black on your palette; the problem is with not understanding how to properly tint colors. This brings us back to two areas we touched on: hue shift, and target brightness. Hue shift is simply the phenomena that colors will look cooler when tinted. Take Vasari Ruby Red, for example, mix it with various shades of grey, and it will look pink in the lights and lavender in the darks (as shown here).

RubyRedGreys

Take a lemon yellow, mix it with black, and you’ll be convinced that there is a warm green in the mixture (but there isn’t).

YellowGreys

It’s valuable to know the expressive range of your colors on their own, and how they change with pure greys. This will simplify the thought process and let you focus on other challenges in the work. Develop a sense for that color’s hue shift on tinting. The best way to do this is to simply mix up a range of light and dark greys from black and white, and then mix your tube color into each. This will take some time, but if done on some canvass sheets, you can actually pin them to a cork board on your wall for reference. That way, instead of winging it every time guessing at the hue you’re looking at, you can instead compare it with your wall swatches and then pick the proper tube color and know which value of grey to mix it with.

What I’ve touched on here is a method of mixing that relies on using true greys – and this means use black! Let’s face it, if the traditional method of using compliments to neutralize is true, then exactly half way between the two colors is a perfectly neutral grey that is indistinguishable on its own from a mixture from black and white. As you know, black and white are cheaper than other pigment colors, so why waste all that money having a perfectly complimentary tube color for every color you intend to use? That’s anyway not possible because pigments don’t align neatly with the saturation and value of a given tube color’s perfect compliment. You’d also have to have many more colors that would otherwise be the case if you just mixed to grey. So take the mythology of compliments out of your mind for color mixing. Put that traditional color wheel in the bottom drawer. There is certainly a place for oscillating between warm and cool colors in a painting for interest. To achieve that level of visual excitement on the canvass, you can just mix each color directly, and apply it. Simulate the complimentary effect of light in the image, not in the mixing process. You’ll preserve visual interest while conserving on precious colors, and mix faster once you know your colors native potential.

Copyright © 2015 Roy Zuniga

A Tint-to-hue reference chart

To find the tint or shade you’re looking for, use a chart like the one below (available here in detail for download). Identify the tint in one of the middle colors, and then see to which saturated hue it corresponds by tracing to the color on the closest edge.

Image

This spectrum of blends is to provide a greater range of tints of the colors as they get blended with lighter or darker greys. Why is this important? For one, tinting actually causes a perceptual shift in hue. Even though a color sample in the computer will prove that the hue hasn’t changed, to the human eye, it seems that way. Lighter tints gravitate towards their neighbor, sometimes warmer (a red tinted lighter tends to look more like magenta, while a yellow tinted tends to look greenish, with the darker colors seeming a bit warner). This is exactly why we can’t rely on identifying the hue first in our minds. In the case of tints, the physical color will invariable be something else, albeit close by in hue.

The next challenge is of course mapping a given saturated color on the chart to the tube colors in your studio. We’ll address this mapping in later topics. For now, please use your best judgment visually mapping the chart colors to the closest tube color for mixing.

 

copyright (c) 2014 roy zuniga