( Programming ranting ahead, with gorey technical details about CPU architectures. )
But gah... so far I'm managing to reduce the memory load from 513k to 16k, AND reducing the code-size and sub-function calls by over half, not counting the colour-blurring functions which I'm rebuilding to handle full 32bpp input using bit-shifting instead of bit-masked two-stage multiplication combinations that the compiler can't optimize. And that will use the same 'block' of code with a single switch between 3 4k-entry function-pointer tables to switch between 2x, 3x, and 4x scaling, instead of being entirely seperate functions with fully duplicated internal logic.
Would this be considered re-writing procerdural code into an object-oriented format, or am I mis-understanding what object-oriented means? =^.^=