The smoothness of a print is one of its main image quality attributes. Here smoothness can refer to the level of unexpected changes or discontinuities in color transitions (at a macro scale) or the level of local variation (at a micro scale), sometimes also described as grain. This paper starts with presenting an approach to building a first-ever set of metameric printed samples that match in color but vary in grain, followed by a psychovisual study of smoothness perception based on a large number of evaluations by both experts and non-experts. This data shows high levels of intra- and inter-observer correlation and can therefore serve as a robust ground truth for understanding and modelling the print smoothness phenomenon. Then, a previously published predictive smoothness model is revisited, that estimates smoothness from a digital halftone before it is printed, and it is shown to result in high degrees of correlation between observer assigned smoothness judgments and computationally predicted scores. The paper also reports the results of tuning the smoothness metrics parameters to further enhance is alignment with the psychovisual ground truth.