Redo tutorial images

[bangerth]

A lot of the pictures shown in the tutorials are quite old and show an aesthetic and color scheme that is a bit outdated. In particular, we have many many pictures that were created with Visit's default "rainbow" color scale. But we know that the rainbow scale is a bad choice (why, or why, does Visit not change its default!) because it lumps half of the color range into the greens, and it is not a good choice for people with color blindness either.

We should redo essentially all of the pictures of the tutorial using better color schemes. For example, in Visit the "oranges", "blues", and "magma" scales are pretty good. I think this might make for a nice project at a hackathon, or if someone wanted to spend a few minutes at the end of a long day on being a bit of an artist.

[Rombur]

The people of matplotlib did a lot of work on how to create good colormaps. Here is their page about it and a nice video about it

[drwells]

See also #6948: we should tackle both at once.

[bangerth]

Ah yes, @Rombur, thanks for the link. I had looked for it, but couldn't find it (because I was looking for python, not matplotlib).

[marcfehling]

Here is another reference: "Ten Simple Rules for Better Figures" https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003833

[vweeds]

I've been doing a lot of tutorials as a way to learn the software and am happy to help if there is a still a need. Using the latest version of Visit Parallel as my visualization software.

[bangerth]

@vweeds -- any help is always appreciated! If you see pictures that you think ought to be updated, post here about it and we'll give you feedback!

[drwells]

@vweeds basically anything that uses the rainbow color scheme needs to be updated.

[jpthiele]

Just found this and I wholeheartedly support this. I just went through all current steps and here's the list which still has rainbow/jet plots. Also @bangerth it would be good to know where to "publish" these images so they end up at the right place. Also which colorbar was used for the already changed ones, so it's consistent? (Even though personally I'd prefer viridis or cividis)

• [ ] step-4
• [ ] step-7
• [ ] step-12
• [ ] step-16
• [ ] step-17
• [ ] step-18
• [ ] step-20
• [ ] step-21
• [ ] step-23
• [ ] step-24
• [ ] step-25
• [ ] step-29
• [ ] step-30
• [ ] step-31
• [ ] step-33
• [ ] step-34
• [ ] step-36
• [ ] step-38
• [ ] step-40
• [ ] step-41
• [ ] step-43
• [ ] step-51
• [ ] step-53
• [ ] step-61

[jpthiele]

Cividis is a color vision deficiency optimized version of viridis as described here: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0199239

I'm not sure whether doxygen would support this but if it does maybe there could be a switch between different colorscales, i.e. viridis, cividis and bw maybe.

[bangerth]

Prompted by my graduate student @Sam0h51, here is sort of what I envision:

• For each picture, think through what it is meant to represent, and choose a color scheme that matches that. For example, if it is a temperature, use perhaps a color scale that goes from black-brown-red-orange-white. If it an elevation, use a divergent color scale that is blue for values below zero and then green-brown-yellow-white above zero.
• Make sure that the neutral color of the color scale is mapped to the correct "zero" value. For example, for elevations, even if the range you want to represent is -11km ... 9km, visualize -11...+11 km so that the neutral color in the middle of the scale falls on zero, and not -1km.
• If you show visualizations of different variables in close proximity -- say, temperature, pressure, and density in an Euler code, make sure that they all use different color scales to make it apparent visually that these pictures show different things.
• Think about what pieces of information are useful in a picture. User name and file name are probably not. Axes labels may or may not. Axes ticks may or may not. Min/max of values as part of the legend may be, or may not. Etc.