The idea of using pseudo-random spatial structures in graphic arts (color and black & white rendering algorithms) was suggested by psychologists and biologists about ten years ago. They observed that some natural pseudo-random structures such as the spatial distribution of receptors in the human retina play an important part in the perceptual process. Our modelization of pseudo-random spatial structures resembles the retinal mosaic. It starts by obtaining the quasi-random distribution of tile centers according to some well-defined spectral characteristics. We then obtain the desired tesselation of the output device space by applying the Voronoi polygonization process. Two slightly different approaches to the output image computation are being explored. In the first approach, an analytic black-dot curve is calculated according to the resampled input signal level and the area of each given tile. This analytic curve is scan-converted to obtain the blackened pixels. In the second approach, we associate threshold values to all pixels inside every tile according to some specially tailored analytic spot function. Then, the standard threshold comparison process is applied.