Matplotlib hide bar in bar3d if height is zero

Question

I have seen matplotlib invisible bar if height is zero - however, that is for the 2D case of bar plot, so I couldn't quite apply it to my problem.

I have also seen Make transparent color bar with height 0 in matplotlib which should be usable for the 3D case - however, I do not want to make the zero-height bars transparent, I want to hide them, to speed up rendering when interacting (rotating) the 3D plot.

Here is an example, based on Create 3D histogram of 2D data — Matplotlib 3.5.0 documentation:

import matplotlib.pyplot as plt
import numpy as np

fig = plt.figure()
ax = fig.add_subplot(projection='3d')
x = y = np.concatenate([np.array([5,4,3,2,1]), np.zeros(35)])
hist, xedges, yedges = np.histogram2d(x, y, bins=40, range=[[0, 40], [0, 40]])

# Construct arrays for the anchor positions of the 16 bars.
xpos, ypos = np.meshgrid(xedges[:-1] + 0.25, yedges[:-1] + 0.25, indexing="ij")
xpos = xpos.ravel()
ypos = ypos.ravel()
zpos = 0

# Construct arrays with the dimensions for the 16 bars.
dx = dy = 0.5 * np.ones_like(zpos)
dz = hist.ravel()

ax.bar3d(xpos, ypos, zpos, dx, dy, dz, zsort='average')

plt.show()

Running this creates this plot:

Most of the bars are height 0, but there are many of them, so rotating the plot is a bit sluggish.

I have attempted to "mask" like this, as in Make transparent color bar with height 0 in matplotlib:

import matplotlib.pyplot as plt
import numpy as np

fig = plt.figure()
ax = fig.add_subplot(projection='3d')
x = y = np.concatenate([np.array([5,4,3,2,1]), np.zeros(35)])
hist, xedges, yedges = np.histogram2d(x, y, bins=40, range=[[0, 40], [0, 40]])

# Construct arrays for the anchor positions of the 16 bars.
xpos, ypos = np.meshgrid(xedges[:-1] + 0.25, yedges[:-1] + 0.25, indexing="ij")
xpos = xpos.ravel()
ypos = ypos.ravel()
zpos = 0

# Construct arrays with the dimensions for the 16 bars.
dx = dy = 0.5 * np.ones_like(zpos)
dz = hist.ravel()

mask_dz = dz == 0 # SO:60111736, 3d case

ax.bar3d(xpos[~mask_dz], ypos[~mask_dz], zpos[~mask_dz], dx[~mask_dz], dy[~mask_dz], dz[~mask_dz], zsort='average')

plt.show()

... but this fails with:

Traceback (most recent call last):
  File "D:\msys64\tmp\test.py", line 22, in <module>
    ax.bar3d(xpos[~mask_dz], ypos[~mask_dz], zpos[~mask_dz], dx[~mask_dz], dy[~mask_dz], dz[~mask_dz], zsort='average')
TypeError: 'int' object is not subscriptable

How can I get the same plot as shown earlier - except with all zero-height bars hidden?

Answer 1

While writing this question, I stumbled upon the right solution - basically, not all of xpos,... dx,... are arrays, so we should not apply the mask where they are not:

import matplotlib.pyplot as plt
import numpy as np

fig = plt.figure()
ax = fig.add_subplot(projection='3d')
x = y = np.concatenate([np.array([5,4,3,2,1]), np.zeros(35)])
hist, xedges, yedges = np.histogram2d(x, y, bins=40, range=[[0, 40], [0, 40]])

# Construct arrays for the anchor positions of the 16 bars.
xpos, ypos = np.meshgrid(xedges[:-1] + 0.25, yedges[:-1] + 0.25, indexing="ij")
xpos = xpos.ravel()
ypos = ypos.ravel()
zpos = 0

# Construct arrays with the dimensions for the 16 bars.
dx = dy = 0.5 * np.ones_like(zpos)
dz = hist.ravel()

mask_dz = dz == 0 # SO:60111736, 3d case

ax.bar3d(xpos[~mask_dz], ypos[~mask_dz], zpos, dx, dy, dz[~mask_dz], zsort='average')
# after the above command, the plot range xlim and ylim gets set to [0,6] - so re-set it back to same as on original image
ax.set_xlim([0, 40])
ax.set_ylim([0, 40])

plt.show()

... which results with this plot:

... which indeed is a lot easier to interact with.

Note that after applying the mask, the xlim and ylim of the plot gets reset to a different range than the one shown in the OP, so the range needs to be re-applied, to get the same plot range visualization as in the OP.