In this paper we present an algorithmic framework for spectral separation of multispectral images. The proposed iterative method inverts the widely used cellular Yule-Nielsen spectral Neugebauer printer model so quickly that model inversion can be considered for pixel-by-pixel application. Simplification and precalculation make improvements and small modifications of the algorithm significantly reduce the computational effort. Using today's available standard hardware allows the separation of high resolution multispectral images in reasonable time.