Color Error in Digital Imaging for Fine Art Reproduction

Abstract

Various aspects of digital fine art reproduction imaging were examined to determine possible sources for image color errors. Differences between two color charts used for making profiles were examined and tested with a special test chart. Possible color errors due to the camera’s spectral response were investigated. The use of tone curve adjusted chart images for profiling were compared against using raw images. The type of ICC tables in the profiles were checked as an error source. Profile induced neutral color errors were tested. The results show that some profile creation algorithms and the type of tables available in the profiles may be sources for some of the image color errors. Introduction In the field of digital imaging, fine art reproduction is the most difficult task. The fine art image is compared to the original artwork with the client’s expectation that the image be as accurate a reproduction of the original as possible. Other photographic tasks, such as scenic or portrait photography, have much more relaxed criteria; usually the image only has to be visually pleasing. As the field of fine art reproduction has expanded in size beyond simply recording originals of historic and cultural importance within institutions to the commercial world where artists are now using imaging for their livelihood, color management has found an increasing role in the process. In a previous paper the issue of observer metamerism failure was investigated. While this was found to be helpful for solving one problem with digital imaging of artworks, color reproduction issues still remain. This paper examines possible additional sources for color errors in fine art digital images. Several sources might be the cause of image color errors; choice of the color chart used to make the color management profile, use of images with a tone curve or without one (raw images), the choice of the ICC profile format, the spectral sensitivity of the camera, or the profile itself. Color Chart Differences To check the profile accuracy against an independent reference, a custom test chart was prepared. The chart consists of 60 patches made from artist acrylic paints and consists of 40 saturated hues, 11 miscellaneous colors and a 9 step gray scale. Paints from 4 manufacturers were used, with 34 pigments represented. Profiles were created using a ColorChecker® (24 patch original, matte finish) and a ColorChecker SG® (140 patches, semi-gloss finish) with programs from 3 vendors. The lighting was adjusted to give a variance of 1 RGB unit across the chart image area. Each chart was imaged using a Better Light Super 8k-HS digital scanning camera at 100% resolution (no interpolation). If available, options in the profiling software were selected for “reproduction”, assuming they would produce the most accurate reproductions. The profiles were applied to the test images in Adobe Photoshop CS2, then converted to CIE L*a*b* 1976 using the Adobe Color Engine with the Absolute Colorimetric intent without black point compensation. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 InCamera ProfileMaker Monaco Profiler ∆ E

Topics

    5 Figures and Tables

    Download Full PDF Version (Non-Commercial Use)