Jennifer McGlinchey Sexton, Conservator of Photographs at Paul Messier, LLC, presented on the testing of reference cards and the development of new imaging protocols that are so desperately needed in our field for increased standardization and comparability of photographs taken of UV-induced visible fluorescence phenomena. The project started by private photograph conservator Paul Messier in 2006, under the servicemark name UV Innovations (SM), was taken over by Jiuan-Jiuan Chen, Buffalo State’s Assistant Professor of Conservation Imaging, Technical Examination, and Documentation at Buffalo State College. Sexton has directed development of the Target-UV™ and UV-Grey™ products since 2012.
Many a visual examination is followed by technical imaging, including both Ultraviolet Fluorescence (UV-FL) and Visible-Induced Luminescence (VIL), and Sexton’s talk first reiterated why observing cultural material by using carefully selected wavelengths of light is important: It is non-invasive, relatively inexpensive, accessible, and (largely) commercially available. As a surface technique, UV-induced fluorescence probes outside layers, coatings, optical brighteners, mold, tidelines, and organic-glaze pigments above bulk pictorial films. Although it is a technique we rely on for the large majority of condition assessments and technical studies, our documentation remains unstandardized, and essentially, unscientific. With so much to gain by standardizing our capture and color-balancing process, as well as by taking careful notes on the equipment used, the prospect of the Target-UV™ and UV-Grey™ UV-Vis fluorescence standards is certainly an exciting one.
UV-FL images are unique in that they contain diagnostic color information, hence the need for standardization, which would enable cross-comparison between colleagues and between before- and after-treatment documentation. The beta testing of the UV target which was been carried out for 2 years has attempted to account for the most significant variables in the production of UV-FL images. The talk evidenced the enormous amount of collaboration and communication needed to streamline the significant aspects of equipment choice, the optimization of acquisition, and the documentation of post-processing methods. The goal was to increase reproducibility and comparability. Sexton’s presentation showed that the beta testing of the product achieved demonstrable results in terms of uniformity of output.
Development of the UV target was begun in collaboration with (Golden) to produce stable fluorogenic pigments of known color values and known neutral-gray values (which were evidently produced by mixing the red, green, and blue fluorogenic pigments). Neutral gray was defined as a gray which was interpreted as neutral by many viewers and which performed similarly under many different conditions. Including such color swatches within a photograph–for the purposes of color-balancing and correcting any variation in the Red-Green-Blue channels for each pixel–is a very familiar principle in visible photography.
A second consideration made for the round-robin testing was that of intensity, which is a variable somewhat unique to UV-FL photography. The nature of the emissive source must be noted for purposes of calibration and exposure, especially as all light sources currently used in fluorescent photography lack stability over long periods. The output of a lamp with fluctuate over time, and this makes relative intensities of materials illuminated with some lamp types very difficult to determine. Even when this particular factor is taken into account, other variables, such as the distance of the lamp to the subject and the wattage of the lamp will effect intensity. It is also possible that multiple emitting sources could be present. These factors should be included in the metadata for the exposure.
To control for this intensity factor, beta testers were to divide their sources, distance-to-subject, and wattage parameters into three different intensity levels which were best matched to certain analyses: “Ultra” was beta-tested for analysis of optical brighteners and other products produced specifically to fluoresce. “High” was best for the analysis of natural and thicker fluorescence, perhaps of a paint film such as zinc white, of some feathers (see Ellen Pearlstein’s talk from this year “ Ultraviolet Induced Visible Fluorescence and Chemical Analysis as Tools for Examining Featherwork”), and uranium glass colorants; and “Low” was used to image thin applications of resins, varnish, and sizing films.
A third variable was that of camera sensitivity, which varies with manufacturer (proprietary internal filtration and software), camera type (either DSLR or digital back cameras), as well as with sensor type (CCD or CMOS, modified or unmodified). Different filters were tested (Kodak Wratten 2e pale yellow filter, PECA 918, and an internal blue-green IR (BG-38) filter). These types of internal filtration are typical on digital cameras to block out IR and some red light to bring the camera output closer to the typical photopic curve of the eye and more closely mimic human vision. The 2e filters UV radiation and a small amount of the blue light commonly emitted by UV lamps, while the Peca 918 is used for IR blocking.
The fourth variable tested was the source type. Those tested included low-pressure mercury, high-pressure mercury, arc and metal halide arc lamps. Although LEDs were used at some institutions, many of these have a peak emission at 398 nm, which is barely in the ultraviolet range. Greg Smith at the IMA analyzed Inova X5 UV LED, and found that it does contain UV but is more expensive. Other products show a large difference in emission peaks which often cannot be accommodated by a simple white-balancing operation. Therefore, testing limited the peak emission to the most common types, emitting between 360 and 370 nm.
The last variables that were analyzed were those of post-processing procedures and software and of user perception and needs. An problematic paradigm identified over the testing period was that of the image being readable or resolvable vis-à-vis a particular argument versus the image being strictly accurate and well-calibrated. A photograph may accurately render the intensity of the fluorescence but it may be so completely underexposed so as to be unreadable.
Testing showed that, despite these difficulties of calibration and subjective experience, that the workflow incorporating the UV Innovations standard, showed a marked increase in standardization. Round-robin testing was completed by eight institutions in the US and Europe in May 2013. Fluorescent object sets were shipped along with the UV standard and filters. Each test site collected two image sets, one named “a,” using the lab’s current UV documentation protocol with color balance and exposure set “by eye,” and the other named “b” using the UV innovations protocol. The increased control provided by the use of the standard was evidenced by the average delta E of L*a*b* data points as well as the average standard deviation of RBG data points for both a and b sets as each institution. By way of example, the ‘Low—a” set showed an improvement from a delta E of 18.8 to the ‘Low—b” with a delta E of 4.9. The average standard deviation in-between these two sets showed an improvement from 32.8 to 6.2!
The presentation went into depth about how this data was collected, how variables were controlled for, and how the data was analyzed, and it showed convincingly that despite the high variability of current work flows, the UV Innovations UV-Grey card and Target-UV standards in conjunction with standardization of UV source and filtration can markedly improve the image variability of UV-FL photography.
One variable in “extra-spectral” imaging that was not addressed in this talk were the spatial inhomogeneities of the light source, or the gradient that results from the use of an inconsistent light source. This could be especially problematic if using UV-FL photography for condition imaging, and “flat-fielding” should be considered as a possible augmentation to the ideal image-acquisition protocol.
There is still further research to be done before this product hits the market. A fourth intensity level will be added to increase the flexibility of the product. The current prototype features two intensity levels on the front and two on the back. Notably, artificial aging must be done to determine when the product should be replaced. As this current standard only operates over UV-A and UV-B, UV Innovations looks forward to developing a UV-C standard, as well as a larger format target.
The prototype of the Target-UV and UV-Grey cards were handmade, but the company hopes to overcome the challenges of large-scale production and distribution by Fall 2014.
Skip to content
AIC Blog Archives: Conservators Converse
Former Blog of the American Institute for Conservation