Today I built matplotlib from svn. The answer why I did it is in the post’s title. 3D plots are back to this wonderful tool !
So, if you build matplotlib version 0.98.6svn or later, you can enjoy this. It’s sometimes a little bit cranky. With plots of multiple points I got one big black surface, but for simple stuff it works great.
Here is an example of 3D plot of hydraulic head on a coordinate system:
import pylab as pl
from numpy import *
import mpl_toolkits.mplot3d.axes3d as axes3d
# x y head
head = ((25, 225 , 240.1178), #h1
(75, 225, 242.3238),#h2
(125, 225, 244.8013),#h3
(175, 225, 247.2736),
(225, 225, 248.8057),#h5
(25, 175 , 241.7646), #h6
(75, 175, 242.0468),#h7
(175, 175, 248.2085), #h8
(225, 175, 249.1382),#h9
(25, 125, 243.1239), #h10
(225, 125 , 249.5332), #h11
(25, 75, 244.4780),#h12
(75, 75, 245.1523), #h13
(175, 75, 248.9717),#h14
(225, 75, 249.4562),#h15
(25, 25 , 245.1523), #h16
(75, 25, 245.8214),#h17
(125, 25, 247.1543), #h18
(175, 25, 248.4819),#h19
(225, 25, 249.3144)) #h20
x, y, z = zip(*head)
xi, yi = pl.arange(0, 250, 5), pl.arange(0, 250, 5) #create grid
head = pl.griddata(x, y, z, xi, yi) #interpolate the scattered data !
print shape(head)
print shape(xi)
print shape(yi)
f = pl.figure(1)
pl.scatter(x, y)
pl.contour(xi, yi, head)
pl.colorbar() # draw colorbar
#pl.show()
f = pl.figure(2)
ax = axes3d.Axes3D(f)
#X,Y,Z = axes3d.get_test_data(0.05)
cset = ax.contourf3D(xi,yi,head)
ax.clabel(cset, fontsize=9, inline=1)
ax.set_xlabel('X')
ax.set_ylabel('Y')
ax.set_zlabel('Hydraulic Head')
f = pl.figure(3)
ax = axes3d.Axes3D(f)
#X,Y,Z = axes3d.get_test_data(0.05)
cset = ax.contour3D(xi,yi,head)
ax.clabel(cset, fontsize=9, inline=1)
ax.set_xlabel('X')
ax.set_ylabel('Y')
ax.set_zlabel('Hydraulic Head')
pl.show()
and the output is:
Another 3D plot
old fashioned...
As usual, have fun with python !