The easiest way to make a live animation in Matplotlib is to use one of the
A base class for Animations.
In both cases it is critical to keep a reference to the instance
object. The animation is advanced by a timer (typically from the host
GUI framework) which the
Animation object holds the only reference
to. If you do not hold a reference to the
Animation object, it (and
hence the timers) will be garbage collected which will stop the
See Helper Classes below for details about what movie formats are supported.
The inner workings of
FuncAnimation is more-or-less:
for d in frames: artists = func(d, *fargs) fig.canvas.draw_idle() fig.canvas.start_event_loop(interval)
with details to handle 'blitting' (to dramatically improve the live performance), to be non-blocking, not repeatedly start/stop the GUI event loop, handle repeats, multiple animated axes, and easily save the animation to a movie file.
'Blitting' is a standard technique in computer graphics. The
general gist is to take an existing bit map (in our case a mostly
rasterized figure) and then 'blit' one more artist on top. Thus, by
managing a saved 'clean' bitmap, we can only re-draw the few artists
that are changing at each frame and possibly save significant amounts of
time. When we use blitting (by passing
blit=True), the core loop of
FuncAnimation gets a bit more complicated:
ax = fig.gca() def update_blit(artists): fig.canvas.restore_region(bg_cache) for a in artists: a.axes.draw_artist(a) ax.figure.canvas.blit(ax.bbox) artists = init_func() for a in artists: a.set_animated(True) fig.canvas.draw() bg_cache = fig.canvas.copy_from_bbox(ax.bbox) for f in frames: artists = func(f, *fargs) update_blit(artists) fig.canvas.start_event_loop(interval)
This is of course leaving out many details (such as updating the
background when the figure is resized or fully re-drawn). However,
this hopefully minimalist example gives a sense of how
func are used inside of
FuncAnimation and the theory of how
The zorder of artists is not taken into account when 'blitting' because the 'blitted' artists are always drawn on top.
The expected signature on
init_func is very simple to
FuncAnimation out of your book keeping and plotting logic, but
this means that the callable objects you pass in must know what
artists they should be working on. There are several approaches to
handling this, of varying complexity and encapsulation. The simplest
approach, which works quite well in the case of a script, is to define the
artist at a global scope and let Python sort things out. For example:
import numpy as np import matplotlib.pyplot as plt from matplotlib.animation import FuncAnimation fig, ax = plt.subplots() xdata, ydata = ,  ln, = ax.plot(, , 'ro') def init(): ax.set_xlim(0, 2*np.pi) ax.set_ylim(-1, 1) return ln, def update(frame): xdata.append(frame) ydata.append(np.sin(frame)) ln.set_data(xdata, ydata) return ln, ani = FuncAnimation(fig, update, frames=np.linspace(0, 2*np.pi, 128), init_func=init, blit=True) plt.show()
The second method is to use
functools.partial to pass arguments to the
import numpy as np import matplotlib.pyplot as plt from matplotlib.animation import FuncAnimation from functools import partial fig, ax = plt.subplots() line1, = ax.plot(, , 'ro') def init(): ax.set_xlim(0, 2*np.pi) ax.set_ylim(-1, 1) return line1, def update(frame, ln, x, y): x.append(frame) y.append(np.sin(frame)) ln.set_data(x, y) return ln, ani = FuncAnimation( fig, partial(update, ln=line1, x=, y=), frames=np.linspace(0, 2*np.pi, 128), init_func=init, blit=True) plt.show()
A third method is to use closures to build up the required artists and functions. A fourth method is to create a class.
The provided writers fall into a few broad categories.
The Pillow writer relies on the Pillow library to write the animation, keeping all data in memory.
The pipe-based writers stream the captured frames over a pipe to an external process. The pipe-based variants tend to be more performant, but may not work on all systems.
Pipe-based ffmpeg writer.
Pipe-based animated gif writer.
The file-based writers save temporary files for each frame which are stitched into a single file at the end. Although slower, these writers can be easier to debug.
File-based ffmpeg writer.
File-based animated gif writer.
The writer classes provide a way to grab sequential frames from the same
Figure. They all provide three methods that
must be called in sequence:
setupprepares the writer (e.g. opening a pipe). Pipe-based and file-based writers take different arguments to
grab_framecan then be called as often as needed to capture a single frame at a time
finishfinalizes the movie and writes the output file to disk.
moviewriter = MovieWriter(...) moviewriter.setup(fig, 'my_movie.ext', dpi=100) for j in range(n): update_figure(j) moviewriter.grab_frame() moviewriter.finish()
with moviewriter.saving(fig, 'myfile.mp4', dpi=100): for j in range(n): update_figure(j) moviewriter.grab_frame()
to ensure that setup and cleanup are performed as necessary.
Animation Base Classes#
A base class for Animations.
A module-level registry is provided to map between the name of the
writer and the class to allow a string to be passed to
Animation.save instead of a writer instance.
Registry of available writer classes by human readable name.
Writer Base Classes#
To reduce code duplication base classes
Abstract base class for writing movies, providing a way to grab frames by calling
Base class for writing movies.
Mixin class for FFMpeg output.
Mixin class for ImageMagick output.
See the source code for how to easily implement new