🎭 Matplotlib integration

.fig

.rgb, .chans and .plt all have a .fig attribute that returns a matplotlib figure object.

a = numbers.rgb.fig # matplotlib figure
print(type(a))
a
<class 'matplotlib.figure.Figure'>

numbers.chans.fig

numbers.plt.fig

numbers.plt(center="mean").fig

Saving the figure

You can save the figure by calling its savefig method:

numbers.rgb.fig.savefig("tench.jpg")
!file tench.jpg; rm tench.jpg
tench.jpg: JPEG image data, JFIF standard 1.01, resolution (DPI), density 100x100, segment length 16, baseline, precision 8, 196x196, components 3
/home/xl0/mambaforge/envs/lovely/lib/python3.13/pty.py:95: RuntimeWarning: os.fork() was called. os.fork() is incompatible with multithreaded code, and JAX is multithreaded, so this will likely lead to a deadlock.
  pid, fd = os.forkpty()

Using existing Axes

All functions allow an ax= argument that accepts an existing Axes object into which they will plot:

fig = plt.figure(figsize=(8,3))
fig.set_constrained_layout(True)
gs = fig.add_gridspec(2,2)
ax1 = fig.add_subplot(gs[0, :])
ax2 = fig.add_subplot(gs[1, 0])
ax3 = fig.add_subplot(gs[1,1:])

ax2.set_axis_off()
ax3.set_axis_off()

numbers.plt(ax=ax1)
numbers.rgb(ax=ax2)
numbers.chans(ax=ax3);

Without Jupyter

By default, the Lovely functions will call plt.close(fig) on the figures they create.

This prevents displaying the figures twice when running in Jupyter.
If you are not using Jupyter, here are 2 configuration options you might want to set:

fig_close=False

#!/usr/bin/env python
from lovely_jax import config, set_config

...

set_config(fig_close=False)
numbers.chans()

# or, using the context manager:
with config(fig_close=False):
    numbers.chans()

plt.show() # Will show all open figures

fig_show=True

If set, lovely will call plt.show() after each figure creation.
You don’t need to set fig_close=False manually.

set_config(fig_show=True)

numbers.chans() # Figure generated and shown

# Note, you have to use the "call" syntax `( )`, as figure
# generation is not triggerd by just accessing the attribute

numbers.chans  # No figure generated

f = numbers.plt.fig # figure generated, shown, and returned.

Note, plt.show() closes all figures.