Point operations in float mode

[kamocat]

The documentation for Image.point() says you can pass it functions or lookup tables.
The examples show a simple linear lambda function using multiply and add operators.
However, if I include a divide operator, it all breaks down. For example, im.point(lambda x: 500 / x) results in
TypeError: unsupported operand type(s) for /: 'int' and '_E'

(My actual formula is somewhat more complex, involving several operators and a logarithm to convert radiated IR into measured temperature)

On the other hand, lookup tables are not supported for float type images, as evidenced by the source code, and I do not wish to lose accuracy by casting to integers.

Happily, I found a concise way to do this:
im.putdata(np.vectorize(my_function)(im.getdata()))

I was intending to post this as a question, but I found my answer as I was providing evidence of the problem. Hopefully this will help someone else who is struggling to use Image.point() in the future.
As a side note, I never found a clear example of what format the lookup table should be. Is it just a list of values, addressed by index?

[radarhere]

I never found a clear example of what format the lookup table should be. Is it just a list of values, addressed by index?

It's just a list of values, yes.

Pillow/Tests/test_image_point.py

Line 11 in 12022fe

im.point(list(range(256)) * 3)

Pillow/Tests/test_image_point.py

Lines 44 to 45 in 12022fe

	im = hopper("I")
	im.point(list(range(256)) * 256, "L")