Digital imaging provides a practical means of both revealing and recording obscured letters in the manuscripts. The digital camera easily captures many other features, which otherwise require special equipment to see and are difficult or impossible to record in conventional facsimiles. For example, readers will be able to examine in great detail the colour and texture of the vellum leaves and to investigate modifications and erasures in the manuscript by magnifying the images, or to view once adjacent leaves side by side. Images are scanned with bright visible light, with fibre optic back-lighting and with the sometimes more revealing aid ultraviolet light. With the help of image processing software it is even possible to restore or at least improve the legibility of faded passages. Readers of the electronic facsimile will thus acquire a reproduction of the manuscript that reveals more than the manuscript itself does under ordinary circumstances.
The equipment used to capture images is a Roche/Kontron ProgRes 3012 digital camera using WinCam and Roche Image Manager software. The maximum image size is 2320 by 3072 pixels in 24-bit colour. The resulting TIFF format image files each take up about 20-25 MB and are enough to tax the capabilities of most PCs. We have found that no single image of this size can be processed in real time without at least 64 MB of memory. Currently the digital camera is attached to a Pentium PC with 96 MB of memory.
The images captured can be of any text, from a single letter or word to an entire page, however shots of whole folios at the size mentioned above represent a printed resolution of only 200 dpi. To achieve a resolution of 600 dpi requires an array of 9 shots to be taken - resulting in almost 200 MB of image data.
Apart from the problems of capturing images, the question arises of how best to provide would-be readers access to the images. As an initial measure, a PC has been installed in the Reading Room running a Powerpoint presentation that cycles through images taken of the manuscripts, however this does not allow any user interaction.
Many of the readers are not computer-literate and so we are looking for solutions that are as straightforward to use as possible. We looked at using Mosaic and the World Wide Web to allow readers to browse through the images, as it provides a simple point and click interface. It should be noted that the images are the essential element of this application and so access with line mode browsers are totally is totally inappropriate. We haven't yet made a final choice of client. For the Reading Rooms we require a client that can be configured to disallow termination of the program and opening arbitrary URLs.
Obviously the size of the archive image files precludes transmitting them to remote clients (we even have to choose our time carefully to transfer sets of archive images over the internal Ethernet), however analysing the problem more closely, we realised that the client programs did not need to receive full image data, but only what was needed to put an image on the screen, i.e. a relatively small GIF file. The problem then became how to deliver that information quickly while maintaining image quality. The initial approach taken was to prepare a set of image files off-line, that could be used to quickly produce image views to be sent to the Web clients.
Users of the Electronic Beowulf image delivery system are initially presented with a ``Welcome Page'' offers a choice of folio to be viewed. Once a complete image has been displayed, the reader can click on the image and a magnified partial image centred around the selected coordinates will be generated and returned. This process can be repeated until the maximum permitted zoom factor is reached. We have experimented setting this parameter such that each pixel is magnified by a factor of four.
