[Discussion] Create a set of rules around cropping and pixel inclusion

When cropping in a cube, how do we want pixel inclusions behavior to be at the bounding box boundary

If you submap(-0.5, 0.5) you get a (1,1) array, ditto for (-0.25, 25)

Currently you can pass pixel coords to submap, but we don't want that to be the case going forward.

Fundamentally, any pixels within your bounding box, should be returned. However there are some very specific edgecases.

Should crop be inclusive of exclusive?
we're defining a pixel as -0.5->+0.499999.

for context to help with our discussions 🫠

Now agreed that submap should be inclusive as is the case currently on NDCube crop. i.e. submap includes top right.

So in the case of submap(bottom_left=(-0.5, -0.5) * u.pix, top_right=(0.5, 0.5) * u.pix) returns a 2 x 2 map.

Because 0.5, 0.5 is in the top right most pixel.

So in the case of submap(bottom_left=(-0.5, -0.5) * u.pix, top_right=(0.5, 0.5) * u.pix) returns a 2 x 2 map.

I strongly disagree with this result

@ayshih

@ayshih can I assume this has been resolved in sunpy/ndcube#874 and this can be closed?

Navigation Menu