@orena1 Thank you for the report :).

I can confirm that the dots were already blue since at least Matplotlib 1.5.3. (they probably never turned red...)

Here is a smaller MWE, with a possible workaround for the moment (though I am not 100% sure why it works):

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

plt.figure()
ax = plt.subplot(1, 1, 1, projection='3d')

pos = np.random.uniform(0, 1, (10, 3))

scat = ax.scatter(pos[:, 0], pos[:, 1], pos[:, 2],
                  edgecolors="none",
                  facecolors="blue")

scat.set_facecolors("red")

# Possible workaround that seems to trigger
# the update of the facecolor values
scat.set_3d_properties(pos[:, 2], "y")

plt.show()

Attn @WeatherGod

Moved to 2.2 as this is not a regression.

This is still an issue. Prior to finding this bug report I posted my own examples as a new bug report because the examples I found were all related to animation just like the context given here. Animation is not the cause.

But, as pointed out in those examples, I believe the issue should be handled together with the issue of coexisting collection-level alpha value and alpha values coming from RGBA colors.

Ran into this issue when updating color limits with scat.clim(), the displayed image from plt.show() is correct, but saved image with plt.savefig() shows wrong (different) facecolor and edgecolor. Using the suggested workaround set_3d_properties() works for now.

Matplotlib version: Ubuntu 16.04, 3.3.0rc1+149.g8c2fbda, Qt5Agg backend, Python 3.7.6

@liuzhenqi77 Do you have a mininal example where it looks right on the screen but saving it comes out wrong?

@tacaswell I tried more cases, and find that my case could be related to plt.colorbar(), but I'm not 100% sure. This MWE is adapted from above. They look the same on screen but different as files.

import numpy as np
import matplotlib.pyplot as plt

pos = np.random.uniform(0, 1, (10, 3))

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
scat = ax.scatter(pos[:, 0], pos[:, 1], pos[:, 2], c=range(10), cmap="Reds")
plt.colorbar(scat, ax=ax)
fig.savefig("test1.png")

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
scat = ax.scatter(pos[:, 0], pos[:, 1], pos[:, 2], c=range(10), cmap="Reds")
# plt.colorbar(scat, ax=ax)
fig.savefig("test2.png")

My original user case was,
When setting color with c=, there could be two ways to adjust the color limit:

• setting vmax and vmin, this works for show but not for savefig
• setting scat.clim(), this is not working for both cases, although the colorbar changes.