The project process begins with the selection and organization of materials for digitization by the Collections Team. Materials are selected based on a variety of criteria, including (but not limited to) historical interest and uniqueness, institutional relevance, completeness of organization and description, and access or use restrictions. Each selected item is entered in a database prepared by the Infrastructure Team with corresponding institution, collection, descriptive, and technical information, while filenames and directory locations are automatically assigned.
Scanning follows the selection and organization of materials and consists of several standard techniques and procedures. All items, including black and white materials, are scanned in RGB, 24-bit color. By scanning in RGB, we leverage several advantages over grayscale.
First, RGB is a truer representation of the image. Because most historic photographs show some signs of age, digitizing in color allows us to capture this aspect of our images, the colorations, subtleties, and artifacts of aging. The goal is to capture the image as it is, without manipulation. Second, RGB images have 3 8-bit channels (24-bit image) while grayscale images have a single 8-bit channel (8-bit image). If adjustments need to be made to the image, RGB provides more data with which to work before the image begins to degrade. Third, if needed, an RGB image can be converted to grayscale; the reverse is not true.
Both the Project Management and Infrastructure Teams feel confident that by capturing our collections in RGB today, we provide the project, our institutions, and our users with greater flexibility and potentially broader uses in the future.
Using the Collections database as a guide to which items have been prepared and are available and ready to scan, digitization begins. Digitization occurs locally at each partner institution. Using SilverFast scanning software, images are scanned at 600 dpi; for slides and other smaller mediums, 2400 dpi is used. Scans are imported into Photoshop where alignment, rotation, and cropping occur, as needed. No other manipulation occurs. For archival purposes, margins are included in the scan to indicate that the entire original object, without doubt, has been scanned. Images are saved as uncompressed TIFFs, using the filename indicated in the database. Utilizing Photoshop’s scripting environment, a custom script verifies dpi and logs activity, providing a check mechanism.
Members of the Infrastructure Team conduct regular site visits to collect the digital images, the database, and the log files. The database records are used to create basic Qualified Dublin Core records which are loaded into the project’s digital object repository. During this record creation and migration process, image conversion occurs: TIFFs are converted to access JPEGs and thumbnail images. These images are uploaded to a web server for use in the web-based cataloging module.
Each digitized item and its associated repository record are made available online through a web-based cataloging interface where a metadata technician gathers descriptive data from existing documents; verifies and inputs descriptive, administrative, and technical elements; and enhances certain metadata elements, using descriptive facets provided by a taxonomy created by the Collections Team.
Various sequences of quality control also occur throughout this process. Training and training documentation are also provided for the digitization and cataloging processes.