In load_image, assuming the input data domain is [0, 65535] (i.e. 16-bit integer) then the converted output range is [0, 1) because it's being divided by 65536.
Though it would never be visible, it appeared incorrect to me. I created #4 while experimenting with some other changes
In load_image, assuming the input data domain is [0, 65535] (i.e. 16-bit integer) then the converted output range is [0, 1) because it's being divided by 65536.
Though it would never be visible, it appeared incorrect to me. I created #4 while experimenting with some other changes