[Bug]: inset_locator is broken when used with subfigures

[amcpherson]

Bug summary

When using inset_locator functions in subfigures, the plots are not placed correctly. Adapting this demo
https://matplotlib.org/stable/gallery/axes_grid1/demo_colorbar_of_inset_axes.html
to use subfigures shows incorrect plots.

Code for reproduction

from matplotlib import cbook
import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1.inset_locator import inset_axes, zoomed_inset_axes


fig = plt.figure()
fig1, fig2 = fig.subfigures(nrows=2, ncols=1, height_ratios=[1, 1], squeeze=True)

ax = fig1.subplots()
ax.set(aspect=1, xlim=(-15, 15), ylim=(-20, 5))

Z = cbook.get_sample_data("axes_grid/bivariate_normal.npy", np_load=True)
extent = (-3, 4, -4, 3)

axins = zoomed_inset_axes(ax, zoom=2, loc='upper left')
axins.set(xticks=[], yticks=[])
im = axins.imshow(Z, extent=extent, origin="lower")

# colorbar
cax = inset_axes(axins,
                 width="5%",  # width = 10% of parent_bbox width
                 height="100%",  # height : 50%
                 loc='lower left',
                 bbox_to_anchor=(1.05, 0., 1, 1),
                 bbox_transform=axins.transAxes,
                 borderpad=0,
                 )
fig.colorbar(im, cax=cax)

plt.show()

Actual outcome

Expected outcome

Something more similar to https://matplotlib.org/stable/gallery/axes_grid1/demo_colorbar_of_inset_axes.html

Additional information

No response

Operating system

OS/X

Matplotlib Version

3.6.1

Matplotlib Backend

module://matplotlib_inline.backend_inline

Python version

No response

Jupyter version

No response

Installation

pip

[jklymak]

That seems likely as axes_grid1 is not very subfigure-aware. Did you try with ax.inset_axes instead?

[anntzer]

From a quick try, the fix seems to be simply

diff --git i/lib/mpl_toolkits/axes_grid1/inset_locator.py w/lib/mpl_toolkits/axes_grid1/inset_locator.py
index bb90f8273c..0771efd71f 100644
--- i/lib/mpl_toolkits/axes_grid1/inset_locator.py
+++ w/lib/mpl_toolkits/axes_grid1/inset_locator.py
@@ -73,7 +73,7 @@ class AnchoredLocatorBase(AnchoredOffsetbox):
         bbox = self.get_window_extent(renderer)
         px, py = self.get_offset(bbox.width, bbox.height, 0, 0, renderer)
         bbox_canvas = Bbox.from_bounds(px, py, bbox.width, bbox.height)
-        tr = ax.figure.transFigure.inverted()
+        tr = ax.figure.transSubfigure.inverted()
         return TransformedBbox(bbox_canvas, tr)

[tacaswell]

Labeling this as easy and good first issue because there is a proposed patch and a clear reproduction case.

The tasks here:

• verify the proposed patch fixes the issue and commit it
• write a test (based on the bug report) This may be a bit tricky to do without adding an image test and I suspect be a vast majority of the work on this issue

[matt256]

mind if I grab this one? Long time user. Would love to be a first time contributor

[amcpherson]

ax.inset_axes works great btw

…

On Fri, Dec 2, 2022 at 8:09 PM Jody Klymak ***@***.***> wrote: That seems likely as axes_grid1 is not very subfigure-aware. Did you try with ax.inset_axes instead? — Reply to this email directly, view it on GitHub <#24589 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAI5WOUEPP2N4LBAAHLEOGTWLKMVHANCNFSM6AAAAAASSMYI5E> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[oscargus]

Please go ahead @matt256 ! We do not assign issues though.

[matt256]

Understood. Thank you and will do. Fix confirmed. Working on the test

[matt256]

PR submitted, but i had some issues with the styling of the test image through the decorator not being classic so i just made the comparison image in the classic style. Naturally i figured out i had to change it to mpl20 as soon as I submitted the PR. I assume i can push another commit to a PR?

[melissawm]

Sure @matt256, you can update your PR as needed. If you need help feel free to ping 😄

[Gairick52]

@amcpherson Hello sir,i want to work on this issue,please assign this to me