Only change axes aspect in imshow if image transform is/contains transData #26122
Conversation
lib/matplotlib/axes/_axes.py
Outdated
| @@ -5585,6 +5585,11 @@ def imshow(self, X, cmap=None, norm=None, *, aspect=None, | |||
| that the data fit in the Axes. In general, this will result in | |||
| non-square pixels. | |||
|
|
|||
| This parameter is ignored if the image uses a transform that does | |||
There was a problem hiding this comment.
Is this really what we want? If you do this then aspect=2 no longer will work. This seems like such an edge case, I'd rather just force users to pass aspect='auto', eg keep the current state.
| or im.get_transform().contains_branch(self.transData)): | ||
| if aspect is None: | ||
| aspect = mpl.rcParams['image.aspect'] | ||
| self.set_aspect(aspect) |
There was a problem hiding this comment.
I agree ignoring explicit user input is bad, I think we should only skip doing something with aspect if aspect is None.
There was a problem hiding this comment.
I could also see a case for raising if both transform and aspect are passed...
… transData. If the image transform does not contain transData then setting the axes aspect to 1 will typically not help to make image pixel squares anyways.
|
Fair enough, I only changed the behavior of |
|
Failing tests are an upload to codecov error. I'll merge as I feel @tacaswell and I had the same point and this fixes it. |
If the image transform does not contain transData then setting the axes aspect to 1 will typically not help to make image pixel squares anyways.
Closes #14118.
PR summary
PR checklist