It is hardly surprising, then, that those who are entrusted with the management of image collections, electronic and non-electronic, greet these presently ungovernable developments with a mixture of eager anticipation and uneasy apprehension. While the Web cannot yet reliably provide a font of usable images to support the institutional missions of these collections, the rich variety of its online holdings, along with the tremendous versatility of its navigational tools, afford a laboratory of sorts for the interfacing of images and text.
Visual resources professionals are currently faced with the challenge of incorporating digitized images into day-to-day operations that, for a long time, have functioned around the manual storage, retrieval, and projection of slides. Logistical realities dictate that this transition be incremental; critical choices will have to be made in order to determine functional objectives for successive phases of deployment. This is especially true for the many visual resources collections that have already invested a tremendous amount of time and money in the development of textual databases for automated collections management systems.(1) As collections management begins to coincide with electronic image management, pressure will come to bear upon visual resources curators to integrate the two, or face the discouraging proposition of starting over with a new application. Off-the-shelf image management software products, with increasingly sophisticated database capabilities, promise to effect such an integration; but the commitment of resources to relatively untested software in a fast-changing technological environment remains a risky business.
Art history imaging projects initiated by educational institutions have typically provided visual material supplementary to course lectures. This can be accomplished most simply by putting images onto CDs that are made available for student perusal at library workstations. A more ambitious, but ultimately more powerful, extension of this concept is to place such images on a network. However, network implementation presupposes a means of access more refined than following a simple sequence of frame numbers; some kind of data-oriented retrieval is required. One might attempt to address this problem by writing an image retrieval application in- house, but the products of such efforts are costly and may quickly become outdated. Commercial image management software may also provide a solution, but such packages may not necessarily co-exist easily with an institutional networking environment. On the other hand, the advent of web-browsing tools such as Netscape and Mosaic have helped favor a more incremental approach to image system development by offering a flexible front end to any number of image/text configurations.
At Princeton University, the Department of Art and Archaeology, in conjunction with the Office of Computing and Information Technology, has made images relevant to selected art history courses viewable over the campus computer network since the beginning of 1994. This ongoing project was designed in a minimum amount of development time and is intended to take maximum advantage of pre-existing infrastructure, a set of criteria for which web-browsing software is particularly well-suited. Image provision has been accomplished via the following steps:
From my standpoint as an image collection manager, the most salient feature of this system is its open-endedness. The digitized imagebase, the textual database, and the hypertext interface are, in effect, separate modules; the structure of one does not necessarily inflect the structure of the other. This may not be the most efficient means of text and image transfer, but it is the most flexible. The set of methods described above should remain relatively adaptable as technological conditions change. This is not to say that the development of HTML scripts comes without overhead, but this is certainly less expensive than redesigning a database system. Of course, this kind of configuration presumes the existence of a well-established textual database; a hypertext front end without a database back end is only of superficial value.
The methods used to support art history courses at Princeton have helped form the basis for even more ambitious efforts of an interdisciplinary scope. The Program in Medieval Studies has developed a hypertext application that links not only images of art and architecture, but also maps, dynastic genealogies, biblical passages, and a glossary of terms.(4) Another project, developed in conjunction with the planned celebration of the University's 250th anniversary in 1996, will document the architectural history of the campus in a multimedia presentation.
Because of concerns over issues of fair use, the art history course images are not available on the World Wide Web. However, images of Princeton University architecture, presented with a somewhat similar interface, may be viewed (caveat surfer: this site is still under construction).
NOTES
(1) CHArt, v.4, no.2, presents a sampling of the wide variety of database applications currently being used in visual resource collections in North America.
(2) The cost effectiveness of Photo-CD vs. in-house scanning has been a matter of debate. Photo-CD processing costs range between $1.00 and $2.00 per image, depending on volume. The purchase of a scanner represents a fixed cost that can be amortized over time. However, labor costs for scanning (usually student help) also have to be factored into one's calculations. I have not actually sat down and costed out these methods; my general sensibility is that the two are a "wash". If one can find a trustworthy processor, the use of Photo-CD offers digitized images in a relatively quick turn- around time without the tedium of in-house scanning. On the other hand, in-house scanning allows for closer quality control. It is also convenient to have a scanner available in order to accommodate last-minute changes and additions to batch-processed images on Photo-CD.
(3) Of the five resolutions stored in a Photo-CD "image pac", the middle, or "Base", resolution (512 x 768 pixels) was used for conversion into JPEG format. Uncompressed "lossless" JPEG image files could be rather large (approximately 500 KB or higher). It was found that converting at a rate of 75% of full resolution (in Adobe Photoshop parlance, "Good" on a scale of "Fair", "Good", and "Excellent") yielded image files that were significantly smaller.
(4) This project has been funded by a grant from the Culpeper Foundation.