When one seeks to characterize the appearance of art paintings, color is the visual attribute that usually focuses most attention: not only does color predominate in the reading of the pictorial work, but it is also the attribute that we best know how to evaluate scientifically, thanks to spectrophotometers or imaging systems that have become portable and affordable, and thanks to the CIE color appearance models that allow us to convert the measured physical data into quantified visual values. However, for some modern paintings, the expression of the painter relies at least as much on gloss as on color; Pierre Soulages (1919-2022) is an exemplary case. This complicates considerably the characterization of the appearance of the paintings because the scientific definition of gloss, its link with measurable light quantities and the measurement of these light quantities over a whole painting are much less established than for color. This paper reports on the knowledge, challenges and difficulties of characterizing the gloss of painted works, by outlining the track of an imaging system to achieve this.