A technique of approximating exponential transformations such as gamma correction is disclosed that achieves high resolution without requiring relatively large RAMs for lookup tables or specialized DSP processors. The technique includes the acts of scaling the pixel values according to their values, wherein smaller pixel values are scaled with larger scale factors and larger pixel values are scaled with smaller scale factors; quantizing the scaled pixel values; looking up the quantized pixel values in a lookup table to provide lookup values; and scaling the lookup values responsive to their scale factors to provide the gamma-corrected pixel values.