Matplotlib savefig image trim

Question

The following sample code will produce a basic line plot with no axes and save it as an SVG file:

import matplotlib.pyplot as plt
plt.axis('off')
plt.plot([1,3,1,2,3])
plt.plot([3,1,1,2,1])
plt.savefig("out.svg", transparent = True)

How do I set the resolution / dimensions of the image? There is padding on all sides of the image beyond the line graph. How do I remove the padding so that the lines appear on the edge of the image?

Sailcloth

Answer 1

I am continually amazed at how many ways there are to do the same thing in matplotlib.
As such, I am sure that someone can make this code much more terse.
At any rate, this should clearly demonstrate how to go about solving your problem.

>>> import pylab
>>> fig = pylab.figure()

>>> pylab.axis('off')
(0.0, 1.0, 0.0, 1.0)
>>> pylab.plot([1,3,1,2,3])
[<matplotlib.lines.Line2D object at 0x37d8cd0>]
>>> pylab.plot([3,1,1,2,1])
[<matplotlib.lines.Line2D object at 0x37d8d10>]

>>> fig.get_size_inches()    # check default size (width, height)
array([ 8.,  6.])
>>> fig.set_size_inches(4,3) 
>>> fig.get_dpi()            # check default dpi (in inches)
80
>>> fig.set_dpi(40)

# using bbox_inches='tight' and pad_inches=0 
# I managed to remove most of the padding; 
# but a small amount still persists
>>> fig.savefig('out.svg', transparent=True, bbox_inches='tight', pad_inches=0)

Documentation for savefig().

Char

is there a way to put these in matplotlibrc? Bad key "savefig.bbox_inches" – Ironstone

You're most welcome. I do not know if it is possible to provide such configuration specifications using a matplotlibrc file. – Char

One other command I like to use (in conjunction with those listed above) from pyplot is plt.tight_layout(), which removes excess white space around the figure. – Henceforward

And if I want to set as resolution in pixel for a png? – Walczak

as for the style, see matplotlib.org/stable/tutorials/introductory/customizing.html: for "tight" use savefig.bbox: tight, for padding use savefig.pad_inches: 0. – Ridgley

Answer 2

The default axis object leaves some room for titles, tick labels and the like. Make your own axis object that fills the whole area:

fig=figure()
ax=fig.add_axes((0,0,1,1))
ax.set_axis_off()
ax.plot([3,1,1,2,1])
ax.plot([1,3,1,2,3])
fig.savefig('out.svg')

In svg format I can't see the line that's right at the bottom, but in png format I can, so it's probably a feature of the svg renderer. You might want to add just a little padding to keep everything visible.

Spindrift

Correct. You can adjust the position of the axes in the figure by making the axes manually. The pyplot (or pylab) command for making an axes includes this in its docstring: axes(rect, axisbg='w') where rect=[left, bottom, width, height] in normalized (0,1) units. axisbg is the background color for the axis, default white – Acroter

Answer 3

A very easy way to trim down most padding is to call tight_layout() before saving the figure.

import numpy as np
import matplotlib.pyplot as plt

x = np.linspace(0, 10, 200)

fig, ax = plt.subplots()
ax.plot(x, np.sin(x))

fig.tight_layout()
fig.savefig('plot.pdf')

Leading

plt.tight_layout() also did the job for me for SVGs. For PNGs, it wasn't necessary but somehow matplotlib seem to treat SVGs differently. – Taipan