Bug report

Bug summary

Regardless of an AxesImage clip_box property value, they are always additionally clipped to their parent Axes' bbox, unlike other artists (e.g. Line2D) who can have a clip_box completely independent of their parent axes.

Code for reproduction

from matplotlib import pyplot as plt

fig, axs = plt.subplots(2, 2)

l, = axs[0, 1].plot([0, 1])
l.set_clip_box(axs[0, 0].bbox)

im = axs[1, 1].imshow([[0, 1], [2, 3]], aspect="auto", extent=(0, 1, 0, 1))
im.set_clip_path(None)
im.set_clip_box(axs[1, 0].bbox)

plt.show()

Actual outcome

The line in axs[0, 1] can be seen in axs[0, 0] by panning axs[0, 1] to the side.
The image in axs[1, 1] cannot be seen in axs[1, 0] (or anywhere else, for that matter) by panning.

Expected outcome

im is correctly clipped to axs[1, 0].

This is likely due to the implementation of AxesImage.make_image

    def make_image(self, renderer, magnification=1.0, unsampled=False):
        # docstring inherited
        trans = self.get_transform()
        # image is created in the canvas coordinate.
        x1, x2, y1, y2 = self.get_extent()
        bbox = Bbox(np.array([[x1, y1], [x2, y2]]))
        transformed_bbox = TransformedBbox(bbox, trans)
        return self._make_image(
            self._A, bbox, transformed_bbox, self.axes.bbox, magnification,
            unsampled=unsampled)

(passing self.axes.bbox as clipping bbox to self._make_image is the problem).

Matplotlib version

• Operating system: linux
• Matplotlib version: master
• Matplotlib backend (print(matplotlib.get_backend())): qt5agg
• Python version: 37
• Jupyter version (if applicable): none
• Other libraries: