Thread with 3 posts
jump to expanded post
you know that really inelegant solution to the packing squares problem? does it depress you? something that's always bothered me in computer graphics is that converting between 8-bit fixed-point (integer: [0, 255]) and binary floating-point ([0, 1]) is unavoidably approximate 😭
it's a division by a power of two, right? no, it's a division by a power of two minus one… also known as not a power of two. why must the universe curse us with this inelegance
@hikari Non-serious solution: store all your 8-bit numbers as signed negative integers!