Which plots backends, currently are the best for article publication?

I’m still a little scarred by my multi-hundred line bash scripts circa GMT 4.5; I’m sure GMT 6 + the Julia interface will restore my love for all things GMT


Yep, GMT evolved enormously since the far GMT4 times (> 8 years ago)

This is very field- and journal-specific. Be thankful for working in a field where people are sensible.

1 Like

I’m thankful. I’m very appreciative of copy editing too. Sometimes copy editors drive me insane, but most of the things they suggest are helpful and correct. Journals that copy edit make your papers better.

When SIAM moved to a LaTeX workflow 20 years ago, we gave authors a break for a bit and converted their word/typewritten(no joke!) … papers into LaTeX. If I recall correctly that lasted for at most a year. After that, we insisted on LaTeX with the SIAM macros. We’ve had very few complaints since.

1 Like

I suggest trying pyplot backend and set familyfont="Serif" with colorpalette=:seaborn_deep. This creates very pleasant figures

l = @layout [
           a{0.3w} [grid(3,3)
                    b{0.2h}  ]

           rand(10, 11),
           layout = l, legend = false, seriestype = [:bar :scatter :path],
           title = ["($i)" for j in 1:1, i in 1:11], titleloc = :right, titlefont = font(8), palette=:seaborn_deep, fontfamily="Serif"


Then you could export this as pdf and fine-tune to your liking. I believe it is not really a competition between backends but the selection of good settings for font, linewidth, and colors.

The benefit of pyplot could be that you can obtain underlying python matlotlib object, and fine-tune with matplotlib’s api. Plots is limited when it comes to minor details, but once you take the matplotlib object, you can do all the tuning provided that you are familiar with matplotlib behemoth.

1 Like

The GMT gallery is quite impressive indeed…

Definitely Gnuplot.jl.
Even if the output may not match the latest trends in plotting appearence, I am confident it can produce even the craziest plot I could think of. And, most importantly, very quickly and without headache!

Examples from a recent paper:

And from an internal report:

They plots exported as PDF have been published with no modification on the Editor side.


Most of the mentioned backends are for 2d plots. Quid of 3d, volume, 3d contour plot, …?

I find that on a non-interactive 2d medium (printed paper, equivalent PDF), 99% of 3d plots would have been better as 2d with a bit of effort invested into highlighting what the author wants to show (eg 2d slices, contours). This is especially true in black & white.

Don’t get me wrong, 3d plots can be amazing, but usually require some interaction (rotation, zoom) and a color screen.


What’s the opinion on VegaLite.jl?

And line-art is even better. That is why most students still prefer blackboard classes than slides.

1 Like

I like it. Quite powerful for ‘tidy’ data but sometimes gets weird for multiple datasets or drawing just a line.

You are lucky! Science specifically ask for Word document submissions, argh!

I still like to create a version of my work that I like, so I can share it on a preprint server at least.