Matplotlib video zooming and quality

Question

I already figured out how to save a video animation and zooming a picture with matplotlib. I want now to merge the two things and understand how to introduce zoom in an animation: reading some documentation I noticed it's not straightforward for just a picture, I expect to be the same or worse for a video.

In the following I write a simple working code, relating to that you can find on the first link

import numpy as np
import matplotlib.pyplot as plt    
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.animation as animation
from matplotlib import cm

#Define x,y vectors and meshgrid with function u on it

x = np.arange(0,10,.1)
y = np.arange(0,10,.1)
X,Y = np.meshgrid(x,y)

#Create a figure and an axis object for the surface

fig = plt.figure()
ax = fig.add_subplot(111,projection='3d')

#Animation without axes and with colormap

def animate(n):
    ax.cla()

    u = np.sin(X+Y+(n/10))
    plt.axis('off')
    plt.grid('off')
    ax.plot_surface(X,Y,u,cmap=cm.inferno)

    return fig,     

anim = animation.FuncAnimation(fig,animate,frames=63)
anim.save('A.mp4',fps=20)

Here the output

As you can see is not bad zoomed, but it's not enough, I want it more!

In the actual code I'm using, video animations are very very small, and I don't know why because it's very similar to this. I hope also that this way I can increase video quality, that is quite poor. Thanks for any help.

Image size is figure size in inches times dpi (dots per inch). If you want more resolution increase either one or both. — Jody Klymak, Jan 24 '22 at 19:47
But my problem is, fundamentally, all the unuseful white space around the image. I'm not sure that matplotlib consider it something different from the image itself — Rob Tan, Jan 24 '22 at 20:14
You could probably edit the original to clarify. Did you try passing bbox_inches='tight' to `anim.save`? — Jody Klymak, Jan 24 '22 at 20:37
True, that `bbox_inches` is ignored (and even produces a message telling you this) in anim.save although the docs say it can be passed as a dictionary. Unless you find a better way, I would generate as a workaround individually cropped images and animate them with [ArtistAnimation](https://matplotlib.org/stable/api/_as_gen/matplotlib.animation.ArtistAnimation.html), matplotlib's flip book equivalent. — Mr. T, Jan 25 '22 at 11:08
@Mr.T How would you do this? You mean with the same method you explained for the picture case (second link)? — Rob Tan, Jan 25 '22 at 13:20
Yes, crop each image, then animate the image collection via [ArtistAnimation](https://stackoverflow.com/a/65622369/8881141). Still, I cannot believe there is no better method. — Mr. T, Jan 25 '22 at 13:26

score 0 · Accepted Answer · answered Feb 02 '22 at 10:46

I finally got to a raw, but effective solution. The code almost doesn't change

import numpy as np
import matplotlib.pyplot as plt    
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.animation as animation
from matplotlib import cm

#Define x,y vectors and meshgrid with function u on it

x = np.arange(0,10,.1)
y = np.arange(0,10,.1)
X,Y = np.meshgrid(x,y)

#Create a figure and an axis object for the surface

fig = plt.figure(frameon=False)
ax = fig.add_subplot(111,projection='3d')

#Definition? and zooming

fig.set_size_inches(10,10)                
fig.subplots_adjust(left=0,right=1,bottom=0,top=1,wspace=None,hspace=None)

#Animation without axes and with colormap

def animate(n):
    ax.cla()

    u = np.sin(X+Y+(n/10))
    plt.axis('off')
    plt.grid('off')
    ax.plot_surface(X,Y,u,cmap=cm.inferno)
    print(n)

    return fig,     

anim = animation.FuncAnimation(fig,animate,frames=63)
anim.save('A.gif',fps=20)

As you can see the zooming is good. The bad quality is due to compression I did in a second moment, because the actual output gif is quite heavy with those parameters, in fact the quality is very good.

Matplotlib video zooming and quality

1 Answers1