Methods, devices and systems for overflow error diffusion in image processing are provided. A method includes calculating error values, e_m, corresponding to differences between modified intensity values for a number of pixel locations and one or more threshold values, where e_m represents an error value associated with an m^th pixel. The method further includes diffusing a calculated overflow error value associated with each calculated error value, e_m, to at least one pixel neighbor of the m^th pixel location.