Repository navigation
[Bug]: no longer able to set multiple hatch colors #28990
Description
Bug summary
Possible bug or maybe a usage issue following some recent updates...
In short, it used to be possible to set the hatch color for each filled contour area to a different color by iterating over the collections and then setting the edgecolor for each, but that's no longer the case with Matplotlib 3.10 and the removal of collections.
I've been trying to find a workaround after the recent updates, but am struggling a bit. Using set_edgecolors only sets a single hatch color for all filled contours (which seems unexpected based upon the docs and like a potential bug) and rcParams['hatch.color'] also just sets a single color (less suprising).
It's entirely possible I'm missing something here, but didn't see much in the docs to help either.
Code for reproduction
# The new approach I was hoping would work, but got unexpected results from
import matplotlib.pyplot as plt
import numpy as np
# invent some numbers, turning the x and y arrays into simple
# 2d arrays, which make combining them together easier.
x = np.linspace(-3, 5, 150).reshape(1, -1)
y = np.linspace(-3, 5, 120).reshape(-1, 1)
z = np.cos(x) + np.sin(y)
# we no longer need x and y to be 2 dimensional, so flatten them.
x, y = x.flatten(), y.flatten()
fig1, ax1 = plt.subplots()
cs = ax1.contourf(x, y, z, hatches=['-', '/', '\\', '//'],
cmap='gray', extend='both', alpha=0.5)
cs.set_edgecolors(["blue","grey","yellow","red"])
fig1.colorbar(cs)
# The older approach that now fails with Matplotlib 3.10
import matplotlib.pyplot as plt
import numpy as np
# invent some numbers, turning the x and y arrays into simple
# 2d arrays, which make combining them together easier.
x = np.linspace(-3, 5, 150).reshape(1, -1)
y = np.linspace(-3, 5, 120).reshape(-1, 1)
z = np.cos(x) + np.sin(y)
# we no longer need x and y to be 2 dimensional, so flatten them.
x, y = x.flatten(), y.flatten()
fig1, ax1 = plt.subplots()
cs = ax1.contourf(x, y, z, hatches=['-', '/', '\\', '//'],
cmap='gray', extend='both', alpha=0.5)
colors = ["blue","grey","yellow","red"]
for i, collection in enumerate(cs.collections):
collection.set_edgecolor(colors[i % len(colors)])
fig1.colorbar(cs)Actual outcome
Expected outcome
Additional information
No response
Operating system
OS/X
Matplotlib Version
3.9.2
Matplotlib Backend
'module://matplotlib_inline.backend_inline'
Python version
3.11.7
Jupyter version
4.0.9
Installation
conda
Activity
timhoffm commentedon Oct 18, 2024
This comes in through #25247, which aggregates all contours in one Collection rather than having a separate Collection per contour. While the old approach set one color per collection, the new approach requires setting individual colors per item in the collection. Conceptually, that should be possible, and
cs.set_edgecolors(["blue","grey","yellow","red"])is indeed accepted without complaint. The issue is that the first color is used for all elements instead of cycling edgecolors.I believe, the culprit is the special-casing for hatches in
matplotlib/lib/matplotlib/contour.py
Lines 1255 to 1269 in 9489b93
one gets different edge colors when removing the hatching. The hatch case has to be extended to loop through edgecolors.
rcomer commentedon Oct 19, 2024
It seems that previously if you wanted different hatch colours you also had to accept coloured edges on the contours. Is that the desired behaviour? I’m thinking since
contourfhas a bespoke hatches keyword, it could also have a bespoke hatchcolors keyword. I assume the different colours could be applied at draw usingrc_context, though I have not tried.timhoffm commentedon Oct 19, 2024
Separating hatch color from edge color is a separate topic. See #26074 and the PRs linked therein.