Bug summary

plt.plot() with None / np.nan row spacers doesn't allow for alpha-d lines to weigh on top of each other; instead, the collective lines all display with the same alpha value everywhere, regardless of overlap.

(I can of course fall back to the LineCollection to regain single line + alpha capabilities, but it would be nice if that weren't necessary.)

Code for reproduction

import matplotlib.pyplot as plt
import numpy as np

dat = np.arange(10).reshape(-1, 2)

fig, axes = plt.subplots(2, 2, figsize=(8, 8))


axes.flatten()[0].plot(dat[:, 0], dat[:, 1], alpha=1.0, c="blue")
axes.flatten()[0].set_title("Full Alpha")

axes.flatten()[1].plot(dat[:, 0], dat[:, 1], alpha=1/10, c="blue")
axes.flatten()[1].set_title("1/10 Alpha")

for i in range(10):
    axes.flatten()[2].plot(dat[:, 0], dat[:, 1], alpha=1/10, c="blue")
axes.flatten()[2].set_title("10 Lines of 1/10 Alpha")

temp = np.r_[dat, [[None, None]]]
temp_to_plot = np.row_stack([temp for _ in range(10)])
axes.flatten()[3].plot(temp_to_plot[:, 0], temp_to_plot[:, 1], alpha=1/10, c="blue")
axes.flatten()[3].set_title("Supposed to be\n"
                            "10 Lines of 1/10 Alpha = Full Alpha")
fig.subplots_adjust(hspace=0.5, wspace=0.5)
fig.set_facecolor("white")
plt.show()

Actual outcome

Expected outcome

The hope was the bottom right line would be full alpha, due to it being 10 identical lines at 1/10 alpha on top of each other, but instead it comes out as 1/10 alpha.

Additional information

The key conditions for replicating here are having multiple lines separated in a single (n, 2) array by a [None, None] / [np.nan, np.nan] row, and choosing to plot with alpha < 1.0.

Operating system

Ubuntu

Matplotlib Version

3.5.3

Matplotlib Backend

module://matplotlib_inline.backend_inline

Python version

Python 3.8.13

Jupyter version

6.4.12

Installation

conda