Until the 1970's, vision research was quite a unified endeavour. The same researchers often studied several levels from the input to the output, comprising both visual psychophysics and the physiology of the retina ("an accessible part of the brain", John Dowling). The correlations described were often illuminating, even if simplistic at times. As different techniques for recording activity in the ("proper") brain became common from the 1970's, many visual psychophysicists began to regard the retina as an uninteresting device for injecting light into the brain. More recently, the great advances in retinal molecular biology and electrophysiology have given the retina a second coming. On one hand, mechanisms can be addressed at more fundamental levels: sub-cellular, molecular, genetic. On the other hand, correlations with visual psychophysics can be formulated with new quantitative precision, and hypotheses experimentally tested with new tools. Psychophysical modelling may benefit greatly from "factoring out" transformations directly inherited from the retina. Notably, linear modelling is of rather limited value for understanding what the retina does. It is time to revisit the relations between basic aspects of visual performance and retinal signalling. These aspects include absolute sensitivity, discrimination and scaling of achromatic and chromatic contrast, spatio-temporal resolution, summation and antagonism, and the effects of the general illumination level on all of these.