Save plot to image file instead of displaying it

python

matplotlib

image-processing

plotting

byNikita Barsukov·Sep 18, 2024

We directly save a plot to an image file using matplotlib’s savefig() function. Just remember, if you plan to also display the plot, make sure savefig() is called before plt.show() since show() can clear the whole canvas. Here's a quick example:

import matplotlib.pyplot as plt

plt.plot([1, 2, 3], [4, 5, 1])  # Creating a simple plot
plt.savefig('myplot.png')  # Saving it to a PNG

In this, myplot.png is your masterpiece. Modify the file extension (like .jpg, .pdf, or .svg) to adjust the image format.

Choosing the right format

The choice of image file format really depends on the application. For instance, a .png file delivers a decent quality-to-size ratio for online content. However, for professional documents or presentations, .pdf would be a better choice.

plt.savefig('plot.png')  # Web-friendly
plt.savefig('plot.pdf')  # Print-friendly

Fancy rasterized images? Use .png. Need vectorized graphics? Go with .pdf. The only plot twist here is that the format hinges on the file extension in savefig.

Sick of the whitespace in your plots? Use bbox_inches='tight' within savefig to clip it. Your masterpiece deserves a closer crop!

Dealing with backends

Matplotlib can sometimes be too eager to show your plot in a GUI window, especially in development environments like Jupyter or Spyder. If you want to stop this, use a non-interactive backend like 'Agg' before importing pyplot.

import matplotlib
matplotlib.use('Agg')  # Backend set to 'Agg' for no GUI interaction

import matplotlib.pyplot as plt
plt.plot([1, 2, 3], [4, 2, 6])
plt.savefig('myplot.png')

Uncommenting plt.show() is better when you don't want to see your plot right away.

For iterative tasks or multiple figures, include plt.close() to prevent memory overloading. It's like closing the fridge door - let's save some energy, shall we?

Automated scripts and backend juggling

If your script is designed to run automatically, without any available display (like a web server or on a background job), you'll need to silence matplotlib's default GUI behavior:

import matplotlib.pyplot as plt
# The next line might throw an error in a display-less environment
# plt.show()

To avoid this, switch to the 'Agg' backend:

import matplotlib
matplotlib.use('Agg')  # Now you're in charge!

For interactive plotting (plt.ion()), always close the figure with plt.close(). It's like turning off the TV when you finish watching - keeps you out of trouble.

Catching errors and switching backends

Like life, your code may not always go as planned. If your plots return errors or blank image files, you may need to switch backends dynamically. Here's how:

try:
    plt.plot([1, 2, 3], [2, 3, 1])
    plt.savefig('plot.png')
except SomeError:  # Replace with your specific error
    plt.switch_backend('Agg')  # Time for plan B!
    plt.plot([1, 2, 3], [2, 3, 1])
    plt.savefig('plot.png')

Taking it up a notch with matplotlib.image

For broader image processing tasks, using matplotlib.image brings extra flexibility. You can load and manipulate images directly, making matplotlib an ideal choice for such operations:

import matplotlib.image as mpimg

# Load an image
img = mpimg.imread('input.png')

# (Insert cool image transformations here!🪄)

# Save the result
mpimg.imsave("output.png", img)  # Voila!