mrbaggins wrote:MD5 isn't what you want at all... one bit of difference will completely change the entire result.
I'm really not sure if what you want exists. It's tricky to do reliably, and you have to go on confidence levels because two things might be similar but not the same, or the same picture but different enough that the program gets confused...
I started with the general assumption that two copies of the EXACT SAME FILE should generate the same md5 hash, but that seems to have failed.
I wonder if I can somehow resize the image to a smaller size, reduce color quality, and then generate a hash