Written by Megan Cook, Research Assistant for ICFA and Library Digitization Projects
Importance of Digital Preservation Standards
It is imperative to digitally preserve historically significant photographic materials, like the collection of original color transparencies from the Kariye Camii, part of the Byzantine Institute and Dumbarton Oaks Fieldwork Records and Papers collection in ICFA. These transparencies document the removed traces of the building’s former life as a mosque or stages of the restoration and cleaning process that are invaluable for understanding the original building and its decoration. Film deteriorates over time, and once this happens, it is impossible to reverse. Therefore, it is also ICFA’s goal to create a digital image collection that adheres to proper digital preservation standards, as this may eventually be the only surviving record of our at-risk materials. ICFA plans to release a reference guide for Digitization Guidelines and Standards, which will detail the specifications for scanning and photographing the items held within the Image Collections and Fieldwork Archives at Dumbarton Oaks. In the meantime, I describe below the challenges I faced when scanning the deteriorated Kariye Camii transparencies and the solution I devised to produce archival scans for long-term preservation.
Complications With Scanning
Scanning film is much more complicated than scanning reflective (or non-transparent) materials. Indeed, only certain scanners are equipped with the necessary transparency unit to scan film.
Ideally, one would use a drum scanner to scan film, which operates differently than the commonly used flatbed scanner, and is able to achieve superior image quality. Drum scanners, however, are exceedingly expensive (in the $20,000 range) and are not widely manufactured.
Alternatively, flatbed scanners are widely produced and much more affordable. For most institutions a flatbed scanner is the only suitable option, and it allows for versatility with regard to the nature of materials that can be scanned. Depending on the model of scanner, one can scan various film formats, in addition to reflective materials such as photographic prints and archival documents. Based on my previous scanning experience at the National Gallery of Art, I recommended that Dumbarton Oaks purchase the Epson Expression 10000XL Large Format Photo scanner, coupled with Silverfast Ai scanning software.
Problem 1: Film Carriers
This Epson 10000XL scanner comes with a set of two film carriers for four standard film formats: 35 mm slides, 35 mm film strips, 2 ¼” (medium format) film strips, and 4″ x 5” (large format) sheet film. For the Byzantine Institute Kariye Camii project, with the exception of a few 2 ¼” transparencies (cut to individual frames), none of the transparencies were among the standard film sizes listed above. In my experience – even when useable – these film carriers are problematic: they are difficult to maneuver film into and also tend to crop out the edges of the film.
The purpose of these film carriers is to hold the film flat and directly above the scanner’s platen (glass bed). Film has a tendency to curl, especially when warmed by the heat of the scanner; this curling can prevent the scanner from properly focusing on the film. Film must also not come in contact with the platen during scanning, since this will result in the appearance of “Newton rings” (or “Newton’s rings”) in the scanned image. Newton rings occur when light hits a convex surface – the film – adjacent to the flat surface of the platen.
Problem 2: Scans not properly focused
In addition to Newton rings, another problem inherent with flatbed scanners is the difficulty in obtaining a sharp image. Although most film scanners come with their own film carriers – which theoretically should position the film at the correct height – our results proved that this is not necessarily the case. The scanner we used to digitize the collection of Kariye Camii transparencies, the Epson Expression 10000XL, is a relatively high-end, professional scanner. The Kariye Camii film transparencies consist of medium and large format film, capable of displaying a high level of detail. Therefore, the scanned images created from this film should also display sharp detail at close to (or at) 1:1/100% zoom.
However, our initial scans were not anywhere near as sharp as they should have been, as verified by comparing the scanned image to the film studied on a light table under a loupe. Another method to determine whether the problem originates in the film or the scanner is to study the scanned image for surface flaws on the film – such as scratches or dust – as these anomalies should remain sharp, even if the detail in the film is not. Below is an example of the lack of detail and focus in the initial scan (left) compared to the much improved scan (right), as observed through a flaw in the film.
Maintaining a high level of detail in these scans was especially important for this project, due to the collection’s subject matter and intended use: to view and study details of the mosaics depicted, and to properly preserve the content for historical and archival purposes.
A frequent practice used by many is to compensate for the lack of detail or focus by using the “unsharp mask” function in the scanning software. There are three reasons why I do not suggest using this feature, or similar functions, when scanning:
- Sharpening functions such as the “unsharp mask” do not actually increase the accuracy of detail in an image. This feature can only sharpen the edges of each pixel, giving the image a grainy appearance.
- Archival scanning should strive to capture the most accurate representation of the analog item, with the goal of obtaining as much information as possible without any additions or alterations, at least in the initial image capture file. Rather, one should use a copy of the original file to make any adjustments to the image.
- I do not recommend using any of the features such as filters or exposure adjustments through the software used for scanning. In addition to the second point above, it is also my opinion that digital image alterations and adjustments should be performed using software created specifically for this purpose, such as Adobe Photoshop.
Research and Testing
To find a different solution, I researched the problem and conducted tests to try and isolate variables. I first determined that the problem did not originate in the transparencies themselves, by using the previously described methods of studying the film under a loupe and locating surface flaws on the scans.
As an additional control method, I performed test scans using the color target that came with the Silverfast Ai software, instead of the transparency film. This allowed for an additional measure of consistency throughout the testing process, and eliminated any question about whether the original transparency was in fact inherently sharp.
ICFA also has a second flatbed scanner, an older and smaller Epson V750 Pro model, which I included in the scan tests. Using the two models of scanners, I performed numerous rounds of “Focus Tests.” With the large format 10000XL scanner, I was able to manually make adjustments to the scanner’s focusing function through the Silverfast scanning software (a feature not available on the smaller V750 model). Adjusting the manual focus settings for the large format scanner proved to be a complicated operation, and was ultimately ineffective in my test results.
During my research it was recommended that I look into an online company, Betterscanning.com, creators and distributors of accessories for flatbed scanners. Unfortunately, this company does not manufacture products for the large format scanner we have, but does offer products for the smaller V750 Pro model. On my recommendation, ICFA purchased a “Variable Height Mounting Station” from BetterScanning, which can be used to adjust the height at which film is held above the scanner.
By adjusting the height of the film above the scanner bed, it is possible to improve the focus and clarity of the scans. Each flatbed scanner has a different point of optimal focus, and the challenge is determining what that exact point is, and how to suspend the film at this height for scanning. Test results using the Variable Height Mounting Station showed an improvement in focus from scans made on the V750 without any height adjustment, compared to scans made using the V750 with a height increase of .8mm, as seen below.
Surprisingly, even these improved scans were not as sharp as the initial scans made using the large format scanner. The goal was now to find a way to replicate the adjustable height test with the large format scanner. Since we did not have a Variable Height Mounting Station for the Epson 10000XL, I had to find my own method of mounting the film so that the height could be adjusted.
After much trial and error, I was able to formulate a solution for achieving sharper detail in the scanned Kariye Camii transparencies, while avoiding the formation of Newton rings. Using this system, I was also able to work with the large format scanner (Epson Expressions 10000XL), which I believe to be superior to the alternative V750 model. I borrowed concepts from the Variable Height Mounting Station, and from an existing scan method called “wet mounting.” (For various reasons, wet-mounting itself could not be used for our purposes.)
- Film splice tape (we used 35mm, but other sizes would likely work)
- Anti-Newton ring glass (or ANR glass, which has been treated on one side to prevent the formation of Newton rings when touching film surfaces)
- Magnetic vent covers (cut lengthwise into 4 narrow strips)
Using film splice tape, adhere the transparency film – emulsion side down – on top of the etched (treated) side of the ANR glass. In our case, the sheet of Anti-Newton ring glass included in the Variable Height Mounting Station served as the surface for the film to be mounted on. The film emulsion is flat against the treated side of the ANR glass, and the splice tape is stuck onto the outer edges of the non-emulsion side of the film, creating enough tension so that the film remains as flat as possible.
Cut four strips from magnetic vent covers. Place two on the left side of the scanner bed, one on top of the other. Repeat on the right side of the scanner bed.
Place the ANR glass with the film taped to its surface down on the scanner bed – film side down – so that the vent strips support the ANR glass on either side, effectively holding the glass and film just above the scanner bed.
This method places the film high enough off of the glass bed so that the film does not touch the scanner, and holds the film at a height that is also optimal for the scanner to focus. Please note that is not a universal solution for all flatbed scanners; this is the optimal height for our scanner, but not necessarily any another scanner – even if it’s the same model.
Lastly, it is imperative to perform regular quality control checks at frequent intervals, in order to catch flaws and problems as early as possible.
After much research and numerous tests, I found that our situation was not isolated. The problems we encountered, such as lack of sharp detail in scanned film, are complications inherent in flatbed scanning. I was surprised at the disparity of information related to inadequate scan quality, and the lack of resources available to surmount these issues. With the evolution of digital image technology, there has been an enormous push for institutions to digitize materials for preservation and access. Since flatbed scanners are the most commonly used tool for digitizing archival and special collections such as the Byzantine Institute’s Kariye Camii fieldwork photography, that would suggest the potential loss of valuable image information on a large scale.