These posters were identified in 1995 as ideal material for a digitization project because of their pictorial interest and the enriched access potential which an online database could provide. Their large, full-color format led to the decision to photograph them first on professional color-negative film and then scan from the film. Before photographing, the posters were removed from mylar encapsulation used for conservation, in order to avoid glare and image distortion. However, subsequent experience has shown that the use of polarizing filters on fresh, uncreased mylar can yield acceptable results and avoid the extra task of re-encapsulating.
The archival digital images were saved in photo CD format which offers cross-platform compatibility and multiple levels of resolution. The original scans were cropped by library staff to remove reference targets (including grayscale and color cards as well as inventory numbers). They were re-saved as low-compression, high quality JPEG images using batch processing, and through further processing they were also saved in GIF format to create smaller-scale "thumbnail" images. Original proportions were retained rather than forcing uniform dimensions or restricting the JPEG images to screen size.
An important aspect of the workflow was to develop a numbering scheme which tracked the migration of images through the several processes. Object numbers were assigned to the posters based on their filing order within a broad organizational scheme, and these were correlated with film and photo CD disc frame numbers; ultimately the digital file numbering created for the web server was based on the photo CD's, not the original object numbers. Batch processing was used for file re-naming.
It was decided to create individual records for each poster in both the library's online catalog (NUcat) and an online web database to allow complete and flexible access. Full-level MARC records were created and then converted to SGML for use in the web database where they are searched by OpenText's Livelink software, chosen for its customizable results page, flexible SGML indexing, and robust search features. The data structure for the web database incorporates selected fields from the MARC record, and the conversion was conducted through a customized batch process. In the 856 field of each MARC record, two links appear to the user: one directly to the JPEG image and the second leading to the URL for the web image database if the user prefers incorporating visual browsing with search results. Hidden links allow for the incorporation of the thumbnail images in the various summary displays.
For a list of the various staff members and departments involved in this project, see the credits page.