Saving a matplotlib animation with imagemagick, and without ffmpeg or mencoder

Question

I want to create an animation of a calculated function. I'm not able to install ffmpeg or mencoder on the cluster I'm using to run the animation, but imagemagick is installed. matplotlib.animation apparently supports imagemagick as a writer for animations (see here, for example). The documentation says that the supported formats are:

['png', 'jpeg', 'ppm', 'tiff', 'sgi', 'bmp', 'pbm', 'raw', 'rgba']

I recognize several of these as non-animated file types, but apparently ffmpeg supports .sgi files in some way. It's fine if the codec that I have to use on the cluseter is obscure if I can convert it on my home computer with ffmpeg or mencoder.

How can I use imagemagick to save an animation with matplotlib?

Answer 1

This old post popped up while I was struggling to get some matplotlib animations exported as an animated gif. It's not rocket science, but still somewhat of a hassle to set up the first time around. I'll leave this post for future reference.

As for the specifics when setting up this environment on a cluster, you'll have to translate each step for the specific needs for your server I guess. Animated gif is at least one of the animation options offered by the ImageMagickWriter, but there may be more.

How-to

• Get the ImageMagick binaries
• Set an environment variable MAGICK_HOME pointing to the '<your-install-dir>\modules\coders' folder (see the docs. When on windows, remember to log off and back on)

• point matplotlib to the install dir of ImageMagick: (mpl docs on rcParams)

  import matplotlib.pyplot as plt
  plt.rcParams['animation.convert_path'] = '<your-install-dir>/magick.exe'
  

• Export your animation! (mpl docs on ImageMagickFileWriter)

  writer = ImageMagickFileWriter()
  your_animation.save('location.gif', writer=writer)
  

Note that by default, errors thrown by ImageMagick will not reach you, apart from a nonzero return code. To get more extensive feedback, you'll need something like

import matplotlib as mpl
mpl.verbose.set_level("helpful")

Bonus

As a bonus, here's a proof-of-life: