+// NOTE: x_ratio, y_ratio, x_diff, and y_diff should really be float values;
+// however, I've found that my 32-bit Mac Mini has a buggy EFI (or buggy CPU?), which
+// causes this function to hang on float-to-UINT8 conversions on some (but not all!)
+// float values. Therefore, this function uses integer arithmetic but multiplies
+// all values by FP_MULTIPLIER to achieve something resembling the sort of precision
+// needed for good results.