Printer models and color halftoning

We present new models for color laser printers. They form the basis of `model-based' techniques that exploit printer distortions to increase both the spatial and color-scale resolution of printed images. We consider two model-based techniques, the modified error diffusion algorithm and the least-squares model-based algorithm. The models account for distortions caused by `dot-overlap' and imperfect inks. We show that when the inks are assumed to have perfect light absorption properties, both algorithms are separable. When the inks are not assumed to be perfect, the algorithms are not separable and, moreover, the modified error diffusion becomes unstable. The separable printer models offer computational simplicity and robustness to errors in color registration, while the non-separable models produce images with better colors.