This question is not TeX specific. When creating scientific documents with graphs and figures, what colors are good to use such that they help the reader understand the figure but will also print well on a black and white printer? For example, I would assume that yellow is a poor choice since it is difficult to read on the screen and it prints very light on the page.

  • I would think that that would depend on the colors in the figure/graph? I tend to use thick lines and so far have been using a white background so will be interesting to see what others think. – Peter Grill Jan 24 '12 at 0:03
  • 1
    This might depend on the printer, I think. By the way, if you're using something like TikZ, with a little work you should be able to set up two color schemes, one for black-and-white printing and one for color printing, that you can just switch between by changing a line in the preamble. – JohnJamesSmith Jan 24 '12 at 1:29
  • Unfortunately the scientific journals require a single PDF. – Nathan Farrington Jan 24 '12 at 2:13
  • Do you have a specific kind of graph in mind (bar chart with three different series, line graph with a single series, chloropleth map)? – Jake Jan 24 '12 at 3:56
  • Here is an example paper. – Nathan Farrington Jan 24 '12 at 7:17

In general using patterns is preferable than colour, especially for scientific papers. If you need to use colours though, I suggest you use a combination of high contrast colour values and limit them to two or three colours. For example in the figure below, I wanted to emphasize delayed activities for a Project. I used orange to highlight these values and shades of gray for the rest.

enter image description here

Shades of gray, come out well in print as well as photocopying. For screen viewing, black colours don't look attractive. If you expect people to view your paper on the screen (if for example you posting it on a web page) rather use colours and control the screen version via a boolean, but best option of course is to choose colors that look well both in print as well as on a screen.

  • I hope this is not a new question, but how can I get the contrast of a specific color, such as the orange color in your graph, to find out what gray level it will print with? Photoshop tells me that your dark gray is RGB 128,128,128, your medium gray is RGB 159,159,159, your light gray is 191,191,191, and your orange is 255,140,25. – Nathan Farrington Jan 24 '12 at 17:03
  • @NathanFarrington If you convert to grayscale you will see immediately, either with photoshop or gimp. The gimp manual is a good read gimp-savvy.com/BOOK/index.html?node54.html for this sort of problem. Personally I just experiment and when I see something good I tend to save it and try it later on. The highest contrast is black and white, the further the delta between the channel values the better. – Yiannis Lazarides Jan 24 '12 at 17:19
  • Hmmm, is there a "color scheme" I could use for my papers? For example, product literature such as data sheets often use the colors of the corporation, possibly as a branding mechanism since the primary interface to these companies is through their data sheets. It would be nice to write down some where the RGB values of say up to 5 different colors to use for every figure. This way I wouldn't have to hunt and experiment each time. Perhaps I could choose an "orange" color scheme and write down 5 shades of orange that also convert well to grayscale? – Nathan Farrington Jan 24 '12 at 21:30
  • @NathanFarrington You can check the colourlovers.com for some inspiration. Note that the xcolor package provides a number of testing routines and environments and is a good read. – Yiannis Lazarides Jan 25 '12 at 3:01
  • @NathanFarrington: have a look at the ConTeXt default colorgroups and palettes – morbusg Mar 12 '12 at 10:17

I'd take a look at http://colorbrewer2.org/. It is a site designed to help pick color schemes for cartography, but I've used it to pick schemes for regular scientific plots as well.

There is an optional button that will only show color schemes that are photocopy-able. If you click on learn more it says:

Photocopy Friendly: This indicates that a given color scheme will withstand black and white photocopying. Diverging schemes can not be photocopied successfully. Differences in lightness should be preserved with sequential schemes.

  • I am using the three color qualitative photo-copy-friendly palette, but two of these colors look identical when printed in b/w. – Joachim Breitner Mar 16 '18 at 23:03

An easy way to check whether the black-and-white version of a coloured document will be usable is:

  • 2
    I don't expect this will do anything to figures that aren't drawn with TikZ or such? – Christian Jul 2 '12 at 15:28
  • @Christian: Yes, external graphics included with \includegraphics are not affected. – mhp Jul 2 '12 at 15:41

Some folks at Matplotlib have spent a lot of time thinking about this. There is an excellent talk here: https://www.youtube.com/watch?v=xAoljeRJ3lU

Summary: Nathaniel Smith and Stéfan van der Walt designed a colourmap called Viridis that, in an optimal sense, differentiates sequential data in a visually proportional way, projects well to grayscale and is readable to the colourblind. It's very cool research. Don't use Jet!

There are several others, namely magma, plasma and inferno, which are also close to optimal, and have some blues, reds and pinks if you're aesthetically partial to Jet.

If you would like the colours as a discrete palette for making diagrams and such, I have converted them to Gimp color palettes for my use in Inkscape.

Here is a comparison of those color maps to jet:

comparison of those color maps to jet

  • Which versions of matplotlib are these available in? Also, what are the names of the six color scales you are showing? – user545424 Mar 19 '18 at 15:09
  • @user545424 Any version from the last few years. They are called Viridis, Plasma, Inferno and Magma. See matplotlib.org/users/colormaps.html – Timtro Mar 21 '18 at 15:39

This is from my experience in writing research papers, personally I find the easiest to read is not by color because sometimes we print papers ourselves to read and due to printer differences it's hard to read sometimes. I would recommend using different patters in your graphics (dotted lines, dots, dashes, stripes, etc...)

Again this is just my opinion but I find that patterns are easier to recognize than different color shades sometimes.

  • 1
    Agreed. I try never to make color be the only differentiating factor in a graph, chart, or figure. Instead if I do use color then I try to also have some other differentiating factor such as a pattern. – Nathan Farrington Jan 24 '12 at 7:19

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.