Closed thawn closed 6 months ago
Cool thanks, Stephane 🎉
just noticed that the bug is fixed for the x coordinate but still persists for the y and z coordinates.
I guess you need to apply similar changes for y and z as you did for x in this commit
@StRigaud would you prefer to reopen this bug, or should I file a new one?
actually, I suggest to fix the bug in line 17 of tier1::multiply_image_and_position_func():
rather than creating an output array of the same type as the input, that function should create an output that can handle positions larger than 255 (e.g. int64 if the input is an integer and float64 if the input is a float).
With its current behavior, tier1::multiply_image_and_position_func() implicitly assumes that the position has the same type as the input, while it is actually of type uint32 (or even uint64?).
uint64
and int64
are not supported in OSx arm64 so I would enforce 32 bits as much as possible for now.
Pre-requist tasks
Describe the bug the results of bounding_box() on a uint8 image are limited to uint8 values. Any value that should be above 255 becomes 255
this is particularly annoying, because threshold_otsu returns uint8 images
To Reproduce