This talk focused on the combination of two technologies, Reflectance Transformation Imaging (RTI) and Image analysis. Much of the talk dealt with the application of these two technologies to evaluate accelerated aging or Oddy Test coupons in a quantitative manner. As the evaluation of Oddy tests has traditionally been subjective, making reproducibility problematic, I was particularly interested in the potential for quantitative analysis.
Reflectance Transformation Imaging (RTI) is a relatively inexpensive and simple tool that creates a mathematically synthesized image of an object’s surface from a series of image (typically ~36) lit from different angles and directions. The image produced by the RTI software can reveal visual information that is difficult to discern under normal conditions.
Image analysis software utilizes algorithms that enhance the visual separation of features and marks them for analysis, a process known as segmentation, thereby enabling those features to be quantified. The software used by the authors of this presentation was Image Pro Premier by Media Cybernetics, which has previously been used for thin section analysis of ceramics.
The authors used RTI and Image analysis in combination to evaluate Oddy test coupons. The process aided in visual assessment, improved the documentation of the results, and provided quantitative results. Adding RTI and Image analysis to the Oddy test protocol was not a cumbersome addition, requiring only ~ 20 minutes. It was noted that the type of coupon used made a big difference for this technique, as foil and bent coupons were not ideal since the added texture complicated interpretation of the results.
After exposure, the coupons were photographed and processed in batches by metal: silver, copper, and lead. A single image of the coupons was chosen from the RTI viewer and used for image analysis. A different protocol was used for each metal. The image of the lead coupons was converted to grayscale and the colors inverted, background, control, and corrosion areas were defined, and the “Smart Segmentation” tool used to separate and quantify them. The image of the copper coupons was not converted to grayscale and the variety of corrosion types were all treated the same by the segmentation process. The image of the silver coupons was converted to grayscale or pseudo-color to enhance differences before segmentation. The software allows for individual segmentation protocols to be saved and reused. The percentage of tarnished to untarnished surface could be calculated for each metal. Comparison with visual evaluation of test coupons yielded the following results:
Control or clear pass: 1-4% tarnish
Clear Fail: 45-100% tarnish
Pass for temporary use: 7 – 17% tarnish
The “temporary” category is particularly hard to judge when evaluating Oddy tests in the traditional manner, so this method seems to be especially useful in this case.
In addition to Oddy test results, RTI and image analysis were used by the authors to evaluate rapid corrosion tests and coating tests. In each case, like with the Oddy tests, the process provided good documentation as well as the possibility for quantitative results. The combination of these techniques seems to have great potential for a number of applications and their relative simplicity and inexpensiveness make them a great tool for institutions with limited analytical capabilities.
Tag: RTI
43rd Annual Meeting – Textiles Specialty Group, May 14th, “Lights, Camera, Archaeology: Documenting Archaeological Textile Impressions with Reflectance Transformation Imaging (RTI)” by Emily Frank
Documenting textile impressions or pseudomorphs on archaeological objects is very challenging. In my own experience, I’ve found trying to photograph textile pseudomorphs, especially when they are poorly preserved, very difficult and involves taking multiple shots with varying light angles, which still often results in poor quality images. This is why Emily Frank‘s paper was of particular interest to me because it provided an alternative to digital photography that would be feasible and more effective in documenting textile impressions: Reflectance Transformation Imaging (RTI).
RTI is a computational documentation method that allows for multiple images of an object to be merged into one and viewed interactively to allow the direction of light to be changed so that surface features are enhanced. The process involves changing the direction of the light when each photo is taken. Using open source software, a single image is rendered using various algorithms that allows the viewer to move a dial and change the direction/angle of light the image can be viewed at. Additional components in the software allow for the images to be viewed using different filters or light effects that make visualization of surface features easier. RTI is gaining in popularity as a documentation tool in conservation due to its low cost and feasibility and several papers presented at this year’s conference touched on the use of this technique (including this paper I also blogged about).
There are two general light sources used for RTI. One uses a dome outfitted with many LED lights that will turn off and on as photographs are taken. An RTI light dome is pictured on Cultural Heritage Imaging’s website that was used at the Worcester Art Museum (CHI is a non-profit organization that provides training and tools for this technique). However, most conservators use a lower tech method where a light source (a camera flash or lamp for example) is held at a fixed distance from the artifact and manually moved around at different angles when each photo is taken. You can see an example of this method used in the field in this blog post from UCLA/Getty Conservation Program student Heather White.
In her paper, Emily focused on documenting textile or basketry impressions on ceramics and more ephemeral impressions, such as those left in the soil by deteriorated textiles or baskets, using RTI. By using the various tools offered by the RTI software (changing light angle, using diffuse light or changing it so that concave surfaces of impressions look convex), she was able to see fine features not clearly visible with standard digital photography, such as the angle of fibers, striations on the surface of plant material or the weave structure. For impressions of textiles left in soil (these were mock-ups she made in potting soil) she noted that digital photography was not very effective in recording these because there was no contrast and the impressions were so fragile that they could not be lifted or moved for better examination or imaging. However using RTI she was able to clearly see that the textiles were crocheted.
In describing her set up and work flow, Emily took photos of the impressions indoors, as well as outdoors (for the soil impressions). She was able to take good images outdoors, but it was better to do RTI at dusk with lower light. She took a minimum of 12 shots per impression at 3 different angles. For her light source she used a flash. In all, she said it took her about 10 minutes to shoot each impression.
When compared to digital photography, RTI is a useful and feasible technique for the documentation of impressions, and worked well for most of the impressions Emily tried to record. It seems that RTI worked well as the stand alone documentation method for impressions in about 40% of the images she took, but is more effective as an examination and documentation tool in combination with standard digital photography. RTI is on its way to becoming a more standardized documentation method in conservation. It appears to be effective for recording low contrast, low relief surfaces, such as textile impressions, and may be the best method to record ephemeral or extremely fragile surfaces that are not possible to preserve. I’m excited about the potential of RTI for impressions and look forward to trying it out the next time I have to record textile impressions or organic pseudomorphs on an archaeological object.
43rd Annual Meeting – OSG Tips Session, May 16, "Plaster Cleaning Tests" by Kathryn Brugioni
In this tip presented during the OSG Tips session luncheon, Kathryn Brugioni discussed the use of Reflectance Transformation Imaging (RTI) for evaluating whether certain dry cleaning methods for plaster abraded or damaged the surface of the object.
Dry methods are preferred over wet methods for cleaning plaster because of the risk of solubilizing the substrate during treatment. When Kathryn was presented with a heavily soiled plaster bust that required cleaning, she turned to the use of vinyl erasers as a cleaning method. Using previously published information that evaluated various types of erasers (Williams and Lauffenburger 1995; Pearlstein, et al. 1982) she decided to test two different PVC-based erasers made by Staedtler: the 526 50 Mars plastic eraser and the 527 05 Mars eraser strip refills, to evaluate not only how well they cleaned soiled plaster, but whether they abraded the surface.
Once she chose what dry cleaning method/materials to test, Kathryn was left with the question of how to evaluate surfaces after cleaning to determine the level of abrasion or scratches resulting from the treatment. Examination using SEM imaging has been used (Wharton, et al. 1990) , but it is a technique that may not be available to all conservators. So she looked to a method that could be more accessible: RTI.
RTI, or polynomial texture mapping, is an imaging technique that allows for an interactive display of an image under different lighting conditions. Multiple images are taken of an object where the object is kept in a fixed position, but the light source moves. The images are processed using using freely available software which combines all the images taken into a single image presented in an interface that allows for the direction of light to be moved across the image at different angles highlighting surface features. (The non-profit organization Cultural Heritage Imaging (CHI) is one of the leaders in this type of imaging for cultural heritage and has lots of information on its website about this technique, steps on how to do it and the software needed to process the images). The software also allows for different types of light or shadow effects to be rendered which may improve or further highlight surface examination. All you need for RTI is a camera, moveable light source and some metal spheres (ball bearings) as markers that help the software determine the direction/angle of the light. These are all things that conservators have on hand or can readily purchase (like the ball bearings) making this type of surface examination/imaging more accessible and much cheaper than an SEM.
Kathryn cleaned the surfaces of plaster test coupons using the erasers and imaged them with RTI before and after cleaning. She soon saw that it was possible to see scratches on the surface using this technique. However, she wanted a way to quantify the scratches and determine what the limit was in terms of scratch size observable using RTI. She abraded plaster coupons with a range of grades of micromesh, from 400-1800, and then examined the surfaces using RTI. She noted that you could detect scratches made with up to 800 grit micromesh, but higher grits, like 1800, created more subtle scratches that were not as easily discernable.
Comparing the scratches made by the two erasers on the plaster coupons to those of different micromesh grades, the scratches made by the Mars plastic eraser were similar to those made by 1200 grit micromesh (measured to be about 34μm size scratches) and the eraser strips made scratches similar to 466 grit micromesh (measured to be about 60μm sized scratches). So the eraser strips are much more abrasive to plaster surfaces than the plastic eraser.
Based on Kathryn’s findings, it looks like RTI can be used to evaluate any surface scratches or changes caused through the abrasive action of erasers used for dry cleaning plaster. Though there are limitations to the use of this technique, and fine scratches may not be readily visible, RTI is a useful, and accessible, examination tool and can provide important information on surface changes caused by certain cleaning methods.
References
Pearlstein, E., D. Cabelli, A. King, and N. Indictor. 1982.The Effect of Eraser Treatment on Paper. JAIC 22(1): 1–12.
Wharton, G., S. Lansing Maish, W.S. Ginell. 1990. A Comparative Study of Silver Cleaning Abrasives. JAIC. 29(1): 13-31.
Williams, J. and J. Lauffenburger. 1995. Testing Erasers used to Clean Marble Surfaces. Objects Specialty Group Postprints, Vol. 3: 118-124.