6

I do have a list of 10 teams, each team has its own color.

I do have to make the graphic that shows the teams with the respective color, and would need a way to find 10 colors that are "different enough" to be recognizable.

there are some conditions for the colors though:

  • no pink or similar is allowed. Red and violet is ok, but other "femminine" colors are not.
  • the saturation should be approximately the same for each color.
  • Dull/trist colors such as brown have to be avoided.

now, obviously the question is not about having you giving me 10 colors. I would rather like to understand what a good approach would be to find those colors in a way that they meet the above conditions.. Doing it by hand with color picker did not bring good results, so maybe there is a mathematical way or a tool to do so.

  • Humans can in general only memorize a few colors as a memonic. However they can memorize patterns – joojaa Apr 30 '18 at 16:23
  • Using only colour is a terrible way to distinguish between many otherwise similar items. After the basic seven, what would you call the modified ones? Is turquoise blue, blue green, cyan, or too "girlish?" – Stan Apr 30 '18 at 16:29
  • Purple and violet are mentioned as favoured by most women and are not considered masculine. Then, there's yellow which is the least-favoured hue. – Stan Apr 30 '18 at 16:47
4

With those restrictions You probably cant.

Here is a similar question, the user was asking about 20 but without the restrictions.

Here is a search of how metropolitan subway colors are used.

Now let's see a process and see the results.

You need to start with a color wheel, which is not a color wheel but a color solid. In this case a cone.

enter image description here

Set up your restrictions. But as you want the same saturation, you only need the upper slice of the cone.

The lower layers are for brightness, but they also are reduced in saturation.

enter image description here

And now let's see what color we can have mechanically (D) and later with some adjustments (E)

enter image description here

  • Grape?
  • Blue
  • "Light blue" (a lot of people do not know the term cyan)
  • Turquoise? Sometimes is hard to differentiate it from cyan.
  • Green (This fluorescent green is hard to differentiate from turquoise.
  • Lime? Is this green or yellow?
  • Yellow. But the apparent brightness is hard to combine with white text. Some subway systems fail to make this readable.
  • Orange
  • Red

We have like 6 colors and some 3 possible colors without taking into account color blind users.

Now let's see if we can expand this a bit removing the saturation (therefore brightness restriction)

enter image description here

We can move the green to a darker green, the Lime to a desaturated "grass" color, we can use brown, we can use gray and probably a marine blue, you can have a beige, some pastel purple...

As explained in the first link I provided, humans can not recognize that many colors, if they are not put together.

Also, there is a lot of differentiation per user.

Here is a funny image. http://www.thedoghousediaries.com/1406

enter image description here

  • A lot of interesting answers here, but this one striked me the most. thank you for the answer. – sharkyenergy Apr 30 '18 at 20:56
  • Proper Attribution for that image? (Well ... I already know where it came from. A quite famous – then – questionnaire on color names, with lots of funny answers such as "As I already told you twenty times, this is also called Green".) – usr2564301 Apr 30 '18 at 22:33
  • Atribution now updated. – Rafael May 1 '18 at 4:28
  • 1
    @Rafael the funny image is not true as such the actual ground truth is more like the male side for women also offcourse anybody who hasn't read the xkcd color survey and asks about color spreads shame on you – joojaa May 1 '18 at 6:43
  • Well, actually the rendered colors on the image are wrong, and have not a metodology on how they are choosen for the survey. It is just an anecdothic, not actual "ground truth". – Rafael May 1 '18 at 14:43
4

Because color is a science you can use a variety of methods to determine almost any aspect. There are a lot of papers on almost any aspect of color.

However when you start to look for distinct colors that will all be different you will run into issues.

Asking for no "feminine" colors is tough because there isn't a firm way to tell an algorithm what are considered "feminine" colors. Yes pinks-purples are considered "feminine" by our cultural norms but not so for color theory. And the more colors you pick the harder it will be to have the same level of variation between each.


Now there are several methods you can use to approach your problem. The first is a nice set of 20 distinct colors Sasha Trubetskoy adapted from Kelly’s 22 Colors of Maximum Contrast.

enter image description here

This is a good set to use as a standard "jumping point" because they are:

  • Easily distinguishable
  • Tastefully luminant
  • Given simple, intuitive names
  • Supplied with hex, RGB and CMYK values

All of that makes them excellent candidates for web and print use.


Second Gavin Kistne has taken this idea and made a wonderful site which allows you to set the hue, saturation and lightness parameters to generate a set of colors. Then you can further tweak the results by altering the threshold and interleaving.

enter image description here


Finally the real math, you can make unending use of chroma.js by Gregor Aisch.

This is the meat and potatoes of color manipulation. When you want to get into the super nitty-gritty details of using HSV, LAB, and bezier interpolation to achieve smooth color schemes this is your one-stop-shop!

There is an amazing interactive documentation you can start to use right away to start playing around with.

Taking the 20 (22) colors from the first method you can further extrapolate and experiment. You can also take colors, compute their deltaE and work from there by mathematically using the deltaE to map out colors with your desired difference levels.

Additionally there is a website that takes chroma.js and wraps it in a nice interface to select and export colors.

  • 1
    Nice one. from your first image, I'd say the middle row has already quite distinct colors, minus magenta plus black and white that will fit the question perfectly. – Luciano May 1 '18 at 12:17
2

This is an exploratory method, not a "final result" method.

I generally start with roughly 50% of the necessary colors, or the "odd maximum half" if it's not an even 50%. Then blend between colors to create half-steps of color variations. It's rare you'll end up with overly unsaturated colors this way or pinks unless you use white/black for one of the base colors. Browns can happen based upon the starting colors. So, merely discard those.

5 base colors, then blended between steps to get 4 more.....

enter image description here

What is or is not "feminine" is really in the eye of the beholder.

And you can always continue blending between colors to get more....

enter image description here

  • Dull/trist colors such as brown have to be avoided. Your method doesn't account for visibility and differentiation. – Ovaryraptor Apr 30 '18 at 16:08
  • 1
    @Ovaryraptor merely because a blend results in an undesired color that does not mean that color must be used. This is an exploratory method, not a "final result" method. You can then further select a color and adjust saturation/vibrance in many instances. Art.. not science.... ya know.... :) – Scott Apr 30 '18 at 16:18
  • I completely disagree. Color is 100% a science. – Ovaryraptor Apr 30 '18 at 17:19
  • 4
    Color is a science when implemented but not when determining what will or will not work. It's exploratory in design.. it's science in production. – Scott Apr 30 '18 at 18:25
  • @Ovaryraptor: there is no scientific formula to avoid "too feminine colors", so a pure scientific approach will not work for the OP. I think this, and some of the other answers, serve to give a large enough set to be able to manually pick Mainly Manly Colors. – usr2564301 Apr 30 '18 at 22:38
1

From a perceptual point of view, you're limited to about five different distinct "nameable" hues. If you push the criteria, you can squeeze seven. Also, some hues are better identifiable at a distance than others.

Colour is the last of the visually identifying object characteristics (after position, movement, size, shape, and finally colour.)

Small areas of some hues make identification difficult. (deep Blue on Black, deep red on black, White with Yellow, Yellow with Orange, etc.) Sufficient area helps avoid confusing visual similarity.

What to do…
Consider colour combinations with a dominant and accent colours and neutrals which can be reversed for home and away use. Use White names and trim on Blue jerseys for away games/Blue names and trim on White jerseys for home games, etc. Then, even same colours can compete. example: White with red trim (home team - the 'Whites' versus the Red with white trim (away team - the 'Reds.') There is also the Black, Grey, and White as dominant 'colours' added to the normal spectral hues.

1

The property you are working for is distinction, 10 sports colors which are most distinct from each other and easily recognizable.

You have the 6 colors of the rainbow to start, and you must make 10, so the additional 4 will need to be variations of one of the 6 rainbow colors.

4 of the 6 colors can have a dark version and a light version to make 10.

To stay in conventional athletic color schemes and get the most distinction I recommend you create dark and light versions of blue and green.

The next two will take more consideration as light versions of red, orange, yellow or violet may look too feminine. What I recommend is you experiment with creating a Dark and Medium version of these colors, not light.

Create some possibilities of these final colors, maybe dark red and bright red, or dark orange and medium orange. Present them to the league or interested parties to see which are best.

Equal saturation is not possible with this approach but everyone would agree that light blue and dark blue are both acceptable distinct teams colors, even with differing saturation.

1

Finding the optimal subset of a wide range of colors in order to maximise the minimum distance between each color of the subset is a (NP) hard problem but it can be solved easily if your original set of colors is not too big.

I would suggest starting with X11 colors so (155 colors). You said you don't want pink or some other colors, the best is you first select in the X11 colors every colors that are OK for you.

Then, using Lab distance between colors (the more distant colors are, the more visually distinct they should be, in theory) you can "randomly" build subsets of 10 colors and keep the best one.

I've coded a such tool for myself (https://mokole.com/palette.html) and you can adapt the code for yourself.

Here is what I get after a few shots, using all the X11 colors:

10 distinct colors, sorry for the fushia

0

I like to use Adobe's colour wheel as a starting place if I need to find a range of colours. It is obviously limited to a certain extent, but I think it could be worth a look. You can remove the restricted choices from the colour wheel as you play with the range.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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