AIC's 44th Annual Meeting

Aged and discolored varnishes on paintings are known to decrease the brightness of the white, brighten the darks and give an overall color shift towards yellow. These alterations often cause a loss of three-dimensional appearance. Digital ‘virtual cleaning’ utilizes color images and ad hoc information about the optical properties of the varnish and the painting surface to provide conservators and curators an intuitive feel of how the appearance of the painting will change if the varnish is removed/replaced. Virtual cleaning with more spectral information (e.g. multispectral data) on mock-ups paints glazed with an artificially aged varnish or comparing the reflectances of degraded parts of the painting and parts that did not degrade since they were protected (e.g. by the frame) have been attempted to generate more realistic appearance of cleaned paintings. In the present paper we will report on further studies to develop a more complete and accurate model that describes how the aged varnish alters the color appearance of a painting. This is done through hyperspectral imaging of paintings undergoing conservation treatment, following the change in reflectance with the varnish removal process and developing a mathematical model that describes the change. We will report on two panel paintings that were followed during their conservation treatment: the first painting is a small impressionist panel by Georges Seurat entitled “Haymakers at Montfermeil” c. 1882, and the second is a Dutch still life by Jan van Huysum entitled “Flowers in an Urn” c. 1721, both paintings with a large range of colors, exhibiting vibrant pure colors as well as dark passages. The experimental method adopted is to measure the diffuse reflectance spectra before and after the removal of the aged varnishes on the paintings, as well as after application of a fresh varnish. The removed varnish that has been absorbed in cotton swabs is solubilized and the transmission is also measured. This allowed us to define a mathematical model for the varnish/painting system that describes the optical phenomena associated with aged varnishes, adapting a formula derived from the Kubelka-Munk theory to include the effects of the interface modification with the exposure of the rough air/paint interface, and including the contribution of grime. The comparison between the colorimetric values of the painting after real cleaning and after the virtual cleaning with our varnish/painting model shows that satisfactory results can be achieved with few point-based diffuse reflectance measurements before and after actual varnish removal. However, residual discrepancies are ascribable to spatial variations in the characteristics of the varnish (optical density, thickness) and of the exposed painting (roughness).