An algorithm which exploits binary data representation for efficiently interpolating lookup tables (LUTs) is developed. Some implementation cost and accuracy trade-offs among trilinear, tetrahedral and binary proportional interpolation (BPI) when constrained to no more than one LUT access per pixel are compared for color space conversion of digital images. A simpler sample dither implementation of BPI, called Neighborhood Mask Dither Interpolation (NMDI), is also described.