As you turn back in time, the historical record is narrower and narrower. We lose artifacts to decay in poor preservation methods. While facsimiles (physical copies or photographs) have always been possible, technology can help us in ways that our unprecedented. Imaging technology has enabled a quality of facsimile that brings out various aspects of the page that would not be visible to the naked eye, and that before were lost to antiquity. However, making a digital copy of an antiquity creates a vast amount of data that needs to be managed. Researchers preserved the Chad Gospels, the Wycliffe Bible, and the Homeric E-3 and E-4 manuscripts to use as a dataset to develop a digital repository.
While the digitization is important, it was also labor-intensive. Researchers had to travel to the manuscripts to image them – Lichfield Cathedral holds the Chad Gospels and the Wycliffe Bible and El Escorial Library houses the E-3 and E-4 manuscripts. The delicate manuscripts cannot be moved, and so researchers worked on site, often in less than ideal conditions. When imaging the Chad Gospels, the team worked in the changing room in the back of the choir stall in the Lichfield Cathedral. The room was only 12 feet by 15 feet, with limited power supply and environmental control capabilities. Additionally, the imaging process was time-consuming. It took two minutes of imaging to capture the date for a single page.
The project’s imaging system started imaging from ultra violet light and worked its way up into infrared light. It exposed the document to a particular wavelength of light, took an image in that wavelength, and then moved through the spectrum. Researchers took 13 images in different wavelengths of light and used these images to reconstruct a digital facsimile of the document. These images were black and white, not full color.
After they collected this raw data, the 13 black and white images needed to be processed by running it through a multi-spectral measurement system, which combined all these images and created different images of the same page. Researchers also gathered 3-D data of the page. To line up these images exactly with each other and 3-D data, the Vis Center team solved the “registration” problem. This work captured the steps necessary to specify registration as a metadata construction that enables a range of granularities in mapping images to each other, and heterogeneous relationship across organizational categories such as time (diachronic), multi-modal, and instances related by a semantic object.
However, the image that is most true to the original page is the RGB images, or images created from exposing the documents to red, green, and blue lights. We can use those three images to recreate the page’s true color. Researchers used the data to digitally restore the images – one can flatten pages that are warped or bend or folded and restore characters that have faded.