The lib gives different hash values on different platforms

[TrungKhoaLe]

Thank you for developing such a wonderful library. I wonder why the lib gives me different values on different machines. In my Macbook M1, it gives me one value. Then on the container that I build on that Macbook machine, it gives me another value. I am using all hash types which are p, a, w and d and 3 hash sizes which are 8, 16, and 32.

[JohannesBuchner]

That's surprising. Most hashing functions start with numpy.array(image.convert('L').resize((img_size, img_size), ANTIALIAS)), see for example the code at https://github.com/JohannesBuchner/imagehash/blob/master/imagehash/__init__.py#L261 . Can you check whether that returned array is machine-dependent?

If yes, this is a pillow issue, probably related to anti-aliasing.