FeatureRequest: colour-blind compatible way to visually differentiate between implicants

[Dettorer]

This package produces beautiful visuals for karnaugh maps and the automatic colors for implicants greatly help learners with identification, especially for maps with more than 4 inputs, I use it to make corrections for my lessons.

I just realized though this approach won't work for colour-blind people, maybe we can make the outline of the implicants use different line styles (dotted, dashed, waved, …)?

[oscargus]

It should be possible to more or less mimic the way colors are produced here:

karnaugh-map/karnaugh-map.dtx

Lines 109 to 117 in a9d84f3

	\newcommand{\@karnaughmap@func@decimaltocolor@}[1]{%
	\ifnum#1=0 red\fi
	\ifnum#1=1 green\fi
	\ifnum#1=2 yellow\fi
	\ifnum#1=3 cyan\fi
	\ifnum#1=4 blue\fi
	\ifnum#1=5 magenta\fi
	\ifnum#1>5 cyan\fi
	}

karnaugh-map/karnaugh-map.dtx

Lines 830 to 848 in a9d84f3

	\ifnum0=\@karnaughmap@var@bw@
	\fill[
	rounded corners=3pt,
	fill=\@karnaughmap@func@decimaltocolor@{\@karnaughmap@var@colorindex@},
	fill opacity=0.25,
	] {
	($(\@karnaughmap@func@decimaltobin@{\@karnaughmap@local@northwest@}.center)+(-0.3,0.3)$)
	rectangle
	($(\@karnaughmap@func@decimaltobin@{\@karnaughmap@local@southeast@}.center)+(0.3,-0.3)$)
	};
	\fi
	\draw[
	rounded corners=3pt,
	draw opacity=1.0,
	] {
	($(\@karnaughmap@func@decimaltobin@{\@karnaughmap@local@northwest@}.center)+(-0.3,0.3)$)
	rectangle
	($(\@karnaughmap@func@decimaltobin@{\@karnaughmap@local@southeast@}.center)+(0.3,-0.3)$)
	};

However, it would be nice to get some sort of configuration possibility to be able to specify if plain or "special" line-types are to be used. (And to be able to specify colors.)

[Dettorer]

Yes I had a quick glance at the code and that's where I thought I would have to add something. I plan on trying to see if I can make a PR but no promise.

For the configuration part, would it be a good idea to have maybe a package configuration option that says if non-star versions of karnaug maps should use color, line style, or both, to distinguish between implicants? With a default on either color or both?

[oscargus]

I'd imagine a package configuration option like fancyborders (or a better name) that leads to that the non-solid-line-borders are used. But then I have no idea about the use case you consider and which combinations are suitable (nor do I really have anything to say about it, maybe @2pi has an actually valid opinion).

While looking at the code, I got the impression that one way is to define
\newcommand{\@karnaughmap@var@linetype0@}{solid} and \newcommand{\@karnaughmap@var@color0@}{red} etc and use them in the (new) \@karnaughmap@func@decimaltolinetype@ and \@karnaughmap@func@decimaltocolor@ functions. In that way it is possible, although a bit hacky, to redefine them if/when required.

[Dettorer]

That makes sense, I'll try a POC if I find time for it, it should be quick enough.

[2pi]

Thank you for the suggestion! At first glance I'm a bit worried that the output image might get a bit too messy. Having said that, I welcome the initiative and suggest that the first step should be to create examples of designs using real world maps. Are there better ways than changing the border? Ideally we should try to get input from someone that actually is color-blind.

[Dettorer]

I'm a bit worried too, but I think that's the kind of thing we need to see to evaluate.
The idea of having different line styles was from a friend of mine who is color-blind indeed, but that's just "a friend of mine who is", more input would be very nice. Ideas from other people included embedding a small number or letter inside each bubble (could be worth a try) or just drawing arrows when the bubbles span over multiple submaps (really not a fan of that one).

[Dettorer]

Using basic tikz line styles, it's better than I expected! Here is the render on the "karnaugh-implicant" test case (color vs bw with linestyles):


links to the PDFs: https://dettorer.net/karnaugh-implicant.pdf and https://dettorer.net/karnaugh-implicant-linestyles.pdf.

I also tried with a lesson document of mine that has a bit of every kind of implicants (sorry it's in french):

• original (with color): https://dettorer.net/td6_color.pdf
• with bw linestyles: https://dettorer.net/td6_bw.pdf

One thing I particularly like with this approach is that I can reference the implicants in text by underlining parts of the equation with the same style, in the same way I colored them with the same color in the original version.

What do you think?

I should be able to submit a PR soon, I'm just struggling to find a clean way to add a package and environment option.

[Dettorer]

I just got a small input from someone with dichromacy (friend of a friend, sorry for the two levels on indirection) and watched the above documents, they confirm the black and white version with different line styles is easier to read than the colored version.

[2pi]

Thanks for working on this and talking to people with colorblindness! Much appreciated! We should absolutely try to add support for this. I'll get back to you with a more detailed response later. In the meantime a question. How would you use this? Are you going to use these linestyles together with colors? Or supply different versions to your students?

[Dettorer]

That's a good question, the PR I made allows both (it's just a package and environment option) but I think I'll make two versions of each document (one in color, one in bw with linestyles).
If colour-blindnes was simply seeing things in gray levels I think the combination of color + linestyles would be better than just linestyles, but since it's more like getting two or more colors mixed up I think having any color at all would just add confusion for that population.
I also think that adding linestyles to a colored map don't add anything for people who aren't colour-blind, so there's really no interest to having it both ways. But who knows, maybe if I encounter enough people with different kinds of colour-blindness I'll realize having both it actually better than just bw + linestyles. For now I think having the choice to combine color and linestyles or not is the best way to handle this feature.

[2pi]

Here are my thoughts. Please let me know what you think!

Regarding the design: I think we should go with line styles for now, maybe change it if we get other relevant feedback later. However I'm not happy with all of the line styles used in the example above and in the PR. The default line styles need to look good, work well together, and be easy to distinguish from each other. Sorry for being so picky.

Regarding API: I think we should emphasize that this feature is intended for colorblindness as it might not be obvious and gives us a possibility to adjust the design later should we learn more.

Since working with this new options feature I've wanted to add the black and white option there too and deprecate the star. So my suggestion is a package and environment option with a name like implicantstyle with the available options color, bw, and colorblind(colorblindness?).

Extra: Correct me if I'm wrong, but adding support for changing colors of implicants should also improve this package with regard to colorblindness. So if any of you have any feedback in this regard I would appreciate some API feedback at #13 (comment). Thanks!