When we view artwork, the experience is dynamic. We look from far away and up close and personal. In addition to the image content, we might notice the artist’s working method: choice of colorants, their optical properties such as opaque or transparent, surface roughness and gloss, and surface topography (impasto). Our research goal is to capture these experiences using imaging technologies, archive the scientific data, and create reproductions using displays and print media.
This website, art-si.org, is shorthand for “Art Spectral Imaging.” Our first research program developed technologies that captured spectral (wavelength) information. The building blocks of color are wavelengths. From the spectral properties of light sources, objects, and observers, color is quantified. The “spectral fingerprint” of a painting can tell us whether the artist used cobalt or Prussian blue. Spectral data are used to improve color accuracy. See the research category, spectral capture.
While developing spectral capture capabilities, a second research program was undertaken that benchmarked color and spatial image quality of typical museum imaging, “Direct Digital Image Capture of Cultural Heritage: Benchmarking American Museum Practices and Defining Future Needs.” We have used these findings to guide our research and to provide a performance baseline.
Our spectral capabilities have been used in a number of correlated research programs. The first is spectral-based color printing. The goal is to produce prints matching artwork where the match is illuminant and observer independent, that is, a print that is minimally metameric. This research has included spectral color management, multi-ink color separation algorithms, and ink design. See the research category, spectral printing. Spectral images can be used for analytical explorations, important for art conservation. We have developed techniques (both imaging and with small-aperture contact spectrophotometry) to aid pigment selection for restorative inpainting (retouching). See the research category, inpainting (retouching). Inpainting research combined with image classification (performed commonly in remote sensing) has been used for pigment mapping. A painting is decomposed into its constituent colorants including concentration. This can be used for color simulation and inpainting in addition to the utility of the pigment maps. See the research category, pigment mapping. Our spectral camera system has been used to develop an abridged two-monochromator imaging system for fluorescence. The system is capable of separating fluorescent materials into their fluorescence and non-fluorescence properties, enabling more accurate image archives and better color reproduction. See the research category, fluorescence imaging.
Related to our printing research is an improved understanding of how the color appearance of artwork changes when viewed as a smaller reproduction, such as catalogs and textbooks. We have completed visual experiments using display technologies to develop a better understanding of the human visual system response to image size. See the research category, size effects.
A natural next step to capturing spectral information is to capture information about gloss and surface properties. “Improving Artwork Reproduction Through 3D-Spectral Capture and Computer Graphics Rendering” is our most current research program. The first phase is to design and build an abridged imaging gonio-spectrophotometer that can be replicated in a museum imaging department. This is used to measure a given spatial location’s bi-directional reflectance distribution function, known as a BRDF. BRDF models are being evaluated for accuracy, both physical and psychophysical, in order to minimize the number of measurements. The second phase is to augment the system with a method to determine topography and use these and BRDF data to spectrally render art in a specific location, such as a gallery, using computer graphics. See the research category, 3-D imaging.