I'm implementing the kmeans clustering algorithm in Python. I would like to plot at each iteration the status (image) of the clusters quality. So, basically I have a cycle which plot at each iteration an image and I want to animate this. I don't know if I made that clear. At the moment I just use the show() command which plot the image but then I have to close it in order to continue the iteration.
So, is there some way to animate the sequence of images computed at each step?
I tried the ion() method and it works fine for small amount of data, but if you have large images or images streaming in relatively quickly, this method is horrendously slow. From what I understand, ion() will redraw everything each time you make a change to your figure, including axes and labels, etc. Which might not be what you want.
This thread shows a much nicer way of doing things
Here's a simple example that I made showing how to do this:
import time
import numpy
import matplotlib.pyplot as plt
fig = plt.figure( 1 )
ax = fig.add_subplot( 111 )
ax.set_title("My Title")
im = ax.imshow( numpy.zeros( ( 256, 256, 3 ) ) ) # Blank starting image
fig.show()
im.axes.figure.canvas.draw()
tstart = time.time()
for a in xrange( 100 ):
data = numpy.random.random( ( 256, 256, 3 ) ) # Random image to display
ax.set_title( str( a ) )
im.set_data( data )
im.axes.figure.canvas.draw()
print ( 'FPS:', 100 / ( time.time() - tstart ) )
I get about 30 FPS on my machine with the above code. When I run the same thing with plt.ion() and ax.imshow( data ) instead of im.axes.figure.canvas.draw() and im.set_data( data ), I get around 1 FPS
Use pause(). For a set of images stored in video[t, x, y], this makes a simple animation:
import matplotlib.pyplot as plt
for i in range(video.shape[0]):
plt.imshow(video[i,:,:])
plt.pause(0.5)
%matplotlib qt to make python plot into an external window. Also this answer is refreshingly compactJust enable the interactive mode:
ion()
show()
And it will work. This is a little bit strange. But remember: At the end of a python script, it will close the windows. You have to call
ion()
show()
at the end of the script, if you don't want the windows to close.
I have implemented an image sequence visualization utility that may be helpful. Try it out here
Below is a example that draws a dynamic sine wave.
import numpy as np
def redraw_fn(f, axes):
amp = float(f) / 3000
f0 = 3
t = np.arange(0.0, 1.0, 0.001)
s = amp * np.sin(2 * np.pi * f0 * t)
if not redraw_fn.initialized:
redraw_fn.l, = axes.plot(t, s, lw=2, color='red')
redraw_fn.initialized = True
else:
redraw_fn.l.set_ydata(s)
redraw_fn.initialized = False
num_time_steps = 100
videofig(num_time_steps, redraw_fn)
