1

I've selected a list of colors (for months of the year). This initial set seems (logically) good to me, however I certainly need to adjust some color tones in brightness, saturation or so.

Is there a systematic way to make these colors more balanced and matching?

PS: It's for data plots so the color should be distinguishable.

  • 1
    You might want to read up on colour theory a bit. The go-to resource for things like this are a series of three articles on Smashing Magazine, the first can be found at smashingmagazine.com/2010/01/28/…. Hope that helps! – Vincent Sep 16 '14 at 9:35
  • Oh, is it complex? :/ I was hoping something like "just use equal saturation/brightness and or interpolate" would do. Or just any maths function I can optimize. OK, I try to read more about it. – Gerenuk Sep 16 '14 at 9:49
  • 1
    No, it is not simple.. :D we have colour receptors that do not have equal intensity, and therefore we can distinguish more colours in one range than in another.. these might be useful: graphicdesign.stackexchange.com/questions/26154/… and this graphicdesign.stackexchange.com/questions/16551/… – benteh Sep 16 '14 at 9:55
2

This is only related to your question as-stated, and more to your intended use of the colours:

You might want to ask yourself if you're ever going to find twelve colours distinctive enough for a single data plot. And, if you do, whether that data plot is going to be legible. Twelve lines in a single plot is crowded no matter how you bring it, no matter how distinguished your colours are from each other.

It might be an idea to split the plot in two: six lines per plot is way easier to read. Or, you might want to just plot the months that are remarkable, the data you are trying to talk about. Another potion might be to not plot months, but quarters.

  • What's a sensible limit for number of line colors in a plot? I was always wondering why packages have so few different colors. In my case it's easier since it's a bar plot and the above bar is actually an example. The order is even fixed - just sometimes bar pieces are missing. – Gerenuk Sep 16 '14 at 10:09
  • There's people around this site who have a way more fact-supported opinion on that question than I do ( @Benteh, joojaa) but I'd say about six to eight in the bar graph example you give. – Vincent Sep 16 '14 at 10:14
1

I think you should instead pick your initial colors using color theory, that would make it a ton easier.

Easiest way to do this without any studying is to pick all the 12 colors from this image http://www.motocms.com/wp-content/uploads/2014/08/color-theory-infographic-paper-leaf.jpg

If the colours in the wheel arent the ones you are looking for then just adjust the hue and brightness of image until you get what you like. Anyway by using this image as base you will generate colors that are balanced and follow color theory.

  • It's a nice idea. However, these particular 12 colors are not distinguishable enough for data plots :/ – Gerenuk Sep 16 '14 at 9:51
0

You can make them feel more cohesive by picking one of them and applying that as a 20% tint across all of the others. The result isn't perfect (the colour theory answers are a better choice for that) but it's quick, and a better result than you have at present.

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.