Nick Charlton

Outputting Matplotlib Plots for the Web

For the past few days, I’ve been working with Matplotlib and collecting together a bunch of notes on animation. It lead to my previous post, Drawing and Animating Shapes with Matplotlib. In doing so, I realised I’d only embedded plots inside PDFs (when they weren’t part of some sort of application) and so, whilst getting bitmap images out of Matplotlib is quite easy, it’s not so optimal for the web. I’d rather use SVGs.

But, Matplotlib was designed to produce plots for publications, and so it’s centred around printing. And so we have to deal with DPIs, inches and boundary boxes and a bit of configuration.

So, given the basic sine wave plot below:

import numpy as np
import matplotlib.pyplot as plt

figure = plt.figure()

X = np.linspace(-np.pi, np.pi, 256, endpoint=True)
C, S = np.cos(X), np.sin(X)

plt.plot(X, C)
plt.plot(X, S)

plt.ylim([-1.0, 1.0])
plt.xlim([-3, 3])

Figure 1: A Simple Sine Wave
Figure 1: A Simple Sine Wave

The last line handles saving in the simplest of forms. This gives us a standard sized SVG file. A nice way to calculate the resulting size is below. This was helped by this Stack Overflow question about page sizes.

dpi = figure.get_dpi()
size = figure.get_size_inches()
print "DPI: %i" % dpi
print "Size in inches: %i x %i" % (size[0], size[1])
print "Pixels: %i x %i" % (dpi * size[0], dpi * size[1])

Typical screen DPI is 72 (with print usually around 300), so that gives us the scaling factor. But, there’s no reason why this cannot be 100 and using this gives more obvious result. We can then use the simple equation of: pixels ÷ DPI to figure out the inches. So, for a 700 x 650 image, you’d want to specify 7 x 6.5 inches:

figure.set_size_inches(7, 6.5)

The resulting inches measurement will be rounded to the nearest integer, but this will still translate to pixels. So, a 7 x 6.5 inch image will report being 7 x 6, even though in the example above it’ll produce a 700 x 650 image.

To keep the aspect ratio correct, 50 pixels are added in the above example to take into account the y axis. Notably, the plot dispayed with show() won’t respect the same aspect ratio as the saved file.

The next thing to look at is the border around the plot. The simplest thing to configure is that of tight_layout. If turned on, this reduces the white border around the outside of the plot. The same will be applied to each subplot. After enabling this, you get Figure 2.

Figure 2: Sine Wave Plot with Tight Layout
Figure 2: Sine Wave Plot with Tight Layout

And so, with a little bit of extra effort, it’s quite possible to get perfectly sized and positioned plots using savefig. Using SVG means that the file size is relatively small and it can be scaled without losing quality — much nicer on Retina displays — and they can be embedded like any other image.