Digitizing the Intimate:

The multiple techniques for attributing works of art, whether done with the naked eye or using X-rays or infrared, rely on the same premise: an artist, in a given period (tied to a coherent selection of works), does not depart from a given number of habits such as his artistic gesture, composition, or in the choice of materials used. Multispectral analysis is based on this same principle and pushes it even further. Using the entire colorimetric spectrum it measures a determined number of data for a corpus of works clustered in time and attributed without any possible doubt. This data includes light and color contrasts, pictorial layer thickness, stroke and underlying drawing characteristics, and finally, the colorimetry specific to the painter as well as other details,.

Starting from the quantification of each of these characteristics, it establishes the statistics that represent all the selected works. These statistics then map what we can call the territory of the artist. A painting placed outside this territory – even if it had all the appearance of coming from the concerned artist’s hand – would nevertheless be very unlikely to be authentic. We know that forgers are capable of seeking period materials such as pigments and supports, of reproducing the looks of a stroke, of using color tints resembling those observed in the artist’s work, of painting in thickness like him or not, of relying on an underlying drawing or not in the manner of the master they aim to copy. By relying on objective elements – pigments or tools for example – and others which are linked to appearance like type of stroke, range of colors, general balance of the composition, the forgers can still, if they are particularly skilful and knowledgeable about the artist, deceive a specialist.

Multispectral analysis precisely quantifies data to which forgers do not have access such as thickness – to the nearest micron – of the pictorial layer or that of the pencil stroke. Moreover, we know that color contrasts perception is eminently subjective, a forger will therefore reproduce his own perception of the contrasts made by the painter. This perception will not correspond to the contrasts actually measured in the original work. Multispectral analysis is therefore not only based on the basic premise that an artist does not deviate or ever so slightly from a certain number of habits, but also on the possibility of modeling the very intimacy of his practices. Length of stroke, frequency of colors, shadow and light distribution, all of which the artist himself is probably only partially aware because it comes down to practice, instinct, the moment, brilliance, but also, to perception, the idea, the hand.

At bottom, multispectral analysis transcribes into objective data the way in which sensitivity, in its uniqueness, is embodied in a work of art.

Anne Malherbe

A.M