Publication-ready 3D figures from Matplotlib

I’m a big fan of scripting my plot creation. I also use python for all of my data acquisition and analysis. This naturally put me in a spot to dive into matplotlib when it came time to create figures for a paper I’m working on. It took a bit of digging, but I worked through the kinks and put together a 3D surface plot (with contours) that is PDF and publication ready. I addressed the several issues by overriding the defaults. I want to clarify that I like the defaults for on-screen display, presentations, and posters, but they weren’t quite right for a tiny two-pane figure in a two-column manuscript. Things I “fixed” include:

• Too many ticks and labels
• Small fonts (when reduced to publication dimensions)
• Tick labels misaligned relative to ticks (problem seems to come from using larger fonts to fix the previous issue)
• Odd axes label alignment, rotation, and placement
• Busy background (removed the gray panes with gridlines)

Some of the solutions were easy, some are hackish and only one uses the lightly-documented _axinfo dict. I’ll highlight a few snippets that were useful in this process (full code linked below).

Fontsize

To adjust the font sizes, I took the rc approach which modifies the global settings (at least within this script). The following three lines take care of the font style/size issues:

rc('font',size=28)
rc('font',family='serif')
rc('axes',labelsize=32)

Notice that this requires the use of from matplotlib import rc.
I didn’t like the spacing between the tick labels and the axes so I use the following six lines to loop through the tick labels and set their alignment parameters:

[t.set_va('center') for t in ax1.get_yticklabels()]
[t.set_ha('left') for t in ax1.get_yticklabels()]
[t.set_va('center') for t in ax1.get_xticklabels()]
[t.set_ha('right') for t in ax1.get_xticklabels()]
[t.set_va('center') for t in ax1.get_zticklabels()]
[t.set_ha('left') for t in ax1.get_zticklabels()]

Where .set_va is a method to set the vertical alignment, and .set_ha sets the horizontal alignment. Notice that the alignment choice depends on the axis in order to get the label close to the axis.

Background

To clear out the grid, remove the gray fill and outline the panes (i.e. create an empty cubic wireframe around the plot) I use the following:

ax1.grid(False)
ax1.xaxis.pane.set_edgecolor('black')
ax1.yaxis.pane.set_edgecolor('black')
ax1.xaxis.pane.fill = False
ax1.yaxis.pane.fill = False
ax1.zaxis.pane.fill = False

A final note of caution, in my testing, I found that the visibility of parts of the pane edge depends on the viewing angle. I found that setting the view to ax1.view_init(elev=10, azim=135) worked to see all the edge lines. Notice that this does put the x and y axes in a counter-intuitive orientation. For my purposes, those directions are arbitrary, but that may not always be the case.

Tick placement

There are two things I changed about the ticks, first, the way they point and how long they are. I like ticks pointing into the plot that leave a clean edge around the boundary. To achieve this I did have to dive into the realm of _axinfo (here be dragons):

ax1.xaxis._axinfo['tick']['inward_factor'] = 0
ax1.xaxis._axinfo['tick']['outward_factor'] = 0.4
ax1.yaxis._axinfo['tick']['inward_factor'] = 0
ax1.yaxis._axinfo['tick']['outward_factor'] = 0.4
ax1.zaxis._axinfo['tick']['inward_factor'] = 0
ax1.zaxis._axinfo['tick']['outward_factor'] = 0.4
ax1.zaxis._axinfo['tick']['outward_factor'] = 0.4

Notice the order isn’t what I’d expect: I had to set the inward factor to zero and increase the outward factor. This may be a bug so don’t count on it to always work (and don’t count on access to _axinfo at all for that matter).

The final change to the ticks was to use a different tick placement. Matplotlib has several to choose from and I wanted to select an even interval so the MultipleLocator ticker is my tool. Load it with from matplotlib.ticker import MultipleLocator and then implement for all three axes as follows:

ax1.xaxis.set_major_locator(MultipleLocator(5))
ax1.yaxis.set_major_locator(MultipleLocator(5))
ax1.zaxis.set_major_locator(MultipleLocator(0.01))

This is not automagic so you’d have to change the base interval (or make it programmatic) if you want a general-purpose figure script. I personally have some quick and dirty figure-making scripts (using mostly defaults) for analyzing my own data and then I write up a specific script to create a specific plot for the manuscript. Much of those scripts get heavily reused, but this was the first time I had to create a 3D surface plot and also the farthest I’ve had to dig into matplotlib. I figured I’d document it for that reason. Good luck, and I hope you find this helpful.

Thanks to Ben Root for the tip on setting the pane edgecolor… that piece took me the longest to work out.

Original source for the figure.