In amplitude modulated halftoning, Moiré refers to the low frequency textures created by superimposing the monochrome halftones of cyan, magenta, yellow, and black inks. Recently, a stochastic Moiré phenomenon has been discovered that accounts for the low frequency graininess created by superimposing frequency modulated (FM) halftones. A particularly interesting property of stochastic Moiré is that it is most visible when overlapping FM patterns are uncorrelated and are of equal intensity. So in a printer that cannot guarantee perfect alignment in the component screens, the only way to minimize the visibility of Moiré is to introduce clustering such that two overlapping dither patterns of equal intensity have different principal frequencies. In this article, we introduce an adaptive error diffusion halftoning algorithm that produces green-noise dither patterns with coarseness varying according to color content.