matplotlib - duplicate plot from one figure to another?

Question

I'm somewhat new to matplotlib. What I'm trying to do is write code that saves several figures to eps files, and then generates a composite figure. Basically what I'd like to do is have something like

import matplotlib.pyplot as plt
from matplotlib import gridspec

def my_plot_1():
    fig = plt.figure()
    ...
    return fig

def my_plot_2():
    fig = plt.figure()
    ...
    return fig

def my_combo_plot(fig1,fig2):
    fig = plt.figure()
    gs = gridspec.GridSpec(2,2)
    ax1 = plt.subplot(gs[0,0])
    ax2 = plt.subplot(gs[0,1])
    ax1 COPY fig1
    ax2 COPY fig2
    ...

where then later I could do something like

my_combo_plot( my_plot_1() , my_plot_2() )

and have all the data and settings get copied from the plots returned by the first two functions, but I can't figure out how this would be done with matplotlib.

It's possible to move them around (to duplicate them you can pickle then unpickle). https://stackoverflow.com/a/46906599/5267751 shows how to move it from one figure to another, but I've yet to figure out how to put it in a particular subplot — user202729, Jun 26 '22 at 02:21

wim · Accepted Answer · 2019-04-03T23:43:14.423

7

Since pyplot kind of works like a state machine, I'm not sure if what you are asking for is possible. I would instead factor out the drawing code, something like this:

import matplotlib.pyplot as plt

def my_plot_1(ax=None):
    if ax is None:
        ax = plt.gca()
    ax.plot([1, 2, 3], 'b-')

def my_plot_2(ax=None):
    if ax is None:
        ax = plt.gca()
    ax.plot([3, 2, 1], 'ro')

def my_combo_plot():
    ax1 = plt.subplot(1,2,1)
    ax2 = plt.subplot(1,2,2)
    my_plot_1(ax1)
    my_plot_2(ax2)

edited Apr 03 '19 at 23:43

answered Apr 12 '13 at 04:21

wim

338,267
99
616
750

1

This is the best solution IMHO. I just wanted to add that, now that figures are picklable, technically it is also possible to copy a figure instance - you can even do this from one backend to another. – pelson Apr 12 '13 at 16:27
4

This might be a quick work around but is not an answer. If you have "plotting instructions" that take a lot of time, this solution is not feasible. Say that I have a 3D plot and want to show, in 3 subplots, XZ,YZ and XY views. In principle you could use the same plot and change the view. With this solution, you simply plot 3 times instead of once and change the view. (Unfortunately, I also don't know how to achieve this). Someone? – user989762 Mar 26 '14 at 00:23

score 0 · Answer 2 · answered Jun 26 '22 at 02:32

Using the answer https://stackoverflow.com/a/46906599/5267751 it's possible to move the axes from one figure to other (using pickle it's also possible to keep the old figure).

Add set_subplotspec to position the resulting axes:

import matplotlib.pyplot as plt
from matplotlib import gridspec

def my_plot_1():
    fig = plt.figure()
    plt.plot([1, 2, 3], 'b-')
    return fig

def my_plot_2():
    fig = plt.figure()
    plt.plot([3, 2, 1], 'ro')
    return fig

fig1 = my_plot_1()
fig2 = my_plot_2()

def my_combo_plot(fig1,fig2):
    fig = plt.figure()
    gs = gridspec.GridSpec(2,2)

    ax1 = fig1.axes[0]
    ax1.remove()
    ax1.figure = fig
    fig.add_axes(ax1)
    ax1.set_subplotspec(gs[0, 0])

    ax2 = fig2.axes[0]
    ax2.remove()
    ax2.figure = fig
    fig.add_axes(ax2)
    ax2.set_subplotspec(gs[0, 1])

    plt.close(fig1)
    plt.close(fig2)

my_combo_plot( my_plot_1() , my_plot_2() )

plt.show()

The code assumes that each figure contains exactly one axes, however.

matplotlib - duplicate plot from one figure to another?

2 Answers2

Linked