When I first became involved in the planning of digital image collections, I maintained that there were three components – creation, storage and management [InformationWeek, November 25, 1995] that “are interlocking pieces that use different equipment and software, but whose configurations affect the technology requirements of the others.”  While I would now substitute the term access for storage, I would still support my main point; each component or process is affected by the others and should be treated as part of the whole. 

This summer I was able to attend a conference that made great strides in the integration of these components and those who work with them.  It was entitled the Summer Summit on Images and was organized by the Visual Resource Managers of the Universities of California system or the UC Sliders. For some time they have been holding summer meetings during which they discussed their own work and ways in which they could collaborate. This summer they chose to expand their annual meeting to share their picture of digital imaging in the state of California and to gather responses from others using images in the visual culture community. Attendees came from the entire western region of the United Stated, including Hawaii. There was other Sliders (VRC) as well as their corresponding Bookers (librarians), educational technologists, vendors and image providers.  The result was an exciting and informative exchange of information that all found to be of great value. We are indebted to the UC Sliders for providing this opportunity for discussion and for openly sharing their own process.

Creators or Providers:

The UC Sliders have been working with the California Digital Library [CDL] in the formation of its Image Service Demonstrator Project. The Universities of California created the CDL in 1997 as a central digital repository to serve researchers in both the public sector and the Universities. Its public component is the Online Archive of California [OAC]. (You can learn more about the CDL and all its projects at cdlib.org [http://www.cdlib.org/inside/projects/].) The UC Sliders contribution has been the Library of UC Images (LUCI) with images from the UCB, UCI, UCR, UCSB, UCSC and UCSD art and architecture visual resource collections. LUCI was the UC Sliders own experiment at developing common standards for file format and metadata.  As LUCI is being integrated into the CDL, the experiment is continued by the CDL.  They are using its data to develop a metadata template that collections from other sources and about other topics can also use.

As Laine Farley, Robin Chandler, and Lena Shelton of the CDL presented the status of the Image Service Demonstrator project, currently a collection of approximately 300,000 images compiled from a combination of institutional (UC), licensed, and free sources, the complexity and the benefits of combining diverse collections was clear.

While CDL has been able to create and exact fairly strict image guidelines on those collections just being created, it can not require that those established collections, which were made before the current technology capabilities existed, meet all these guidelines.  The UC institutional resources include LUCI; Museums and the Online Archive of California (MOAC); SPIRO, the UCB Architecture collection; Tebtunis Papyri from UCB; and the UCSF Medical Collection. The project also provides access to licensed and free collections such as: AMICO; Saskia art history images; Hartill; and the Luna collections including the Rumsey Historical Maps.  This is a diverse collection.  Some of these digital collections are quite old, SPIRO was has been used for more than ten years, while others are just being created. The inconsistency of image resolution is actually less of a concern than the diversity of the metadata and hence the attempt to create a template.  Besides the different quantities of information, there are also the different methods of storing information, the metadata structure. Hopefully, the metadata template will alleviate the difficulty this makes when people are using the metadata to search for items.  The benefits were that those beyond image “owners” or even their “owners” field of studies were using the images.  The History departments now had access to cultural images of their period.  Of course, this expanded use also influences the search patterns and thus the metadata structure.

A major concern of the UC Sliders and the CDL is defining exactly what should a Central repository contain.   The UC Sliders estimated they had only digitized 3-5% of all UC image collections or that is 300,000 of an estimated 11.3 million images.  Of course, these 11.3 million images are not all different images. In the past, each of the nine universities developed their own image collections, each of which not only covered the basics, but also had their own unique characteristics.  Clearly, they must develop a way to avoid duplication of effort while at the same time getting the best image and information for the central repository, but what about the special collections, that contains images that to this date only one school has collected. Should the central repository contain those, which may only be of interest to that one school, or just be the “survey course" of images that everyone needs? Who is to review and select the images?

Then there is that fact that collections are constantly evolving.  Currently almost all the campuses were interested in improving their Japanese art History slides. Where should this growth occur, in the individual collections or in the CDL?  

The even bigger question right now is should it be by acquisition or licensing. Many of the current licenses are for access to the images.  When the license expires, a school or system would no longer have access.  On the plus side, the maintenance of the digital integrity would be done by the licensors. Two groups who are licensing image collections, ARTstor,[ http://www.artstor.org/index.html]  and Research Library Group (RLG) discussed the images that they are providing for educational use.

ARTstor, a nonprofit organization established by the Mellon Foundation to create an online archive of cultural images for educational purposes, is currently building their collection by acquiring existing collections to distribute within their system, digitizing projects of existing surrogate collections of specific topics, and direct capture of original material. For example our host, UC San Diego, has been such a contributor to their collection. With a grant from the Mellon Foundation, UCSD developed and consequently submitted their copy stand images and cataloging data for inclusion in ARTstor. While ARTstor’s immediate goal is to deepen their current coverage of art history, they are also strengthening ARTstor’s value to humanists across the board and reinforcing the sense of ARTstor as a “campus-wide” visual  resource.  Their license provides access to their master images; it does not deliver the images to the institutional repository.

To give you an idea of the breadth and diversity of this collection here is a list of contributions in production or about to begin:

• Architecture and allied arts from the Hartill Art Associates. 18,000 images of western architecture from antiquity to the present, made from original color 35mm slides.
• Image of the Black in Western Art. Based on the photo archive of the W.E.B. Du Bois Institute at Harvard, and related to the 4-volume reference work with this title; 30,000 images, reproducing representations of black people in the western artistic tradition from antiquity to the present, made from 4 x 5" and 8 x 10" black and white prints
• Native American Art. Plains Indian ledger drawings and historical photographs from the National Anthropological Archives of the National Museum of Natural History; 12,000 high resolution digital images made from either the original objects
• Visual history of women in America from the Schlesinger Library, Radcliffe Institute, Harvard; 36,000 images now digitized and ready for distribution shortly
• Antiquities, African art, Native American art, and 19th c. sculpture from the Berlin State Museums. Collaboration with the Prussian Cultural Heritage Foundation; 7,500 images from new photography and existing digital images
• French Romanesque and Gothic Architecture, and American Architecture from the Clarence Ward Archive, National Gallery of Art; 7,000 images from large format nitrate negatives
• Islamic art and architecture from the photographic archives of Walter Denny at the University of Massachusetts, Amherst, and Jonathan Bloom and Sheila Blair at Boston College; perhaps 25,000 images
• Paintings from Italian Churches photographed by Sansoni in the 20s and 30s, paintings sold at auction, and historic photographs; 15,000 from large format glass plates

Their collection is approaching 300,000 images and is being promoted for licensing to university library systems. For the UC system, each university is currently establishing its own licensing agreements.

As mentioned before ARTstor’s licensing only provides access. It is actually more restrictive than that.  You can only view their images online or through their offline viewer.  They have developed presentation software into which other images can be incorporated, but not vice a versa. While many find this restriction a negative point, it is giving them a selling point to interest contemporary and international creators to contribute their work as the files are encrypted and only useable by their viewer.  Another perceived negative is the current inconsistent image and data quality.  ARTstor, like CDL, has been forced into the position of using images digitized during the period of digital exploration, in other words, the quality is a little uneven.

In contrast to the CDL and ARTstor, the RLG Cultural Materials is a collection that is developed from primary resources, oral histories and archival materials that are pictorial in nature. RLG is now making it accessible for use in the classroom or lecture hall. Marilee Proffitt, from RLG discussed how they too are dealing with the problem of metadata merger further complicated by diverse descriptive practices (Library - MARC, Archive -  EAD, VRC – Dublin Core), and trying to integrate terminology and meaning. A very simple search strategy is used and the Rights Framework for Education is consistently applied.

Other image sources that were discussed were those that were available without cost, such as the Library of Congress (LOC) which is now over a million images and also includes oral history and music and favorite web sites, some institutional and some personal. There was also some discussion of the difficulty of merging individual faculty’s private collections with them with the other collections. Overall the groups present represented over one million images that could be available to the UC system.

The consensus was that, while they need to reach a scalable, sustainable, critical mass of images, the UC system also needs to create a curatorial board with representation from visual resource managers, special collection librarians, and art and natural history museum curators.  This board would work with the Joint Steering Committee to prioritize which content to license and which to build. It would also evaluate the unique content available from the individual campuses.

Access or Educational Technology

Luna Imaging currently hosts the CDL, providing its presentation solution, Insight, which is very good for educational purposes. There are several other similar presentation or display solutions, such as MDID ( James Madison University’s free program), ARTStor’s OffLine viewer and Almagest. Almagest was developed by Kirk Alexander while at Princeton and Princeton is planning to release it.  Fortunately, Kirk is now at UC Davis and an attendee, so he was able to demonstrate it. Will Groppe, Technical Developer at ARTstor, also demonstrated the Offline Image Viewer with which you must present images in ARTstor.  

With the programs you can create slide shows, zoom, split the screen, show data or not, and create personal (individual) collections. If you have never seen one of these programs run through their paces, you are missing something.  All four programs manage images that are contained within a repository and search across collections. They also allow for data input, but that is not their strength.

Vickie O’Riordan, of  UCSD demonstrated how she put together her lectures, using images from ARTstor, her hard drive and the campus imaging service.  The flow and use of a variety of image types was impressive. Also impressive are some of the other uses that UCSD is developing with imaging technologies. One exciting use was a Physics professor’ who makes videos of his class concentrating on the formulas he is writing on the board.  He then posts these to his web site for course review.

Management

As mentioned earlier the CDL is planning for the merger of diverse collections of metadata.  Lena Shelton of the CDL discussed her attempt to create a metadata template that is  based on the (Visual Resource Association) VRA Core and Cataloging Cultural Objects guidelines. Using these standards she has developed five search fields that should suit the entire image search basic needs. In addition, because there is an inconsistency in completed data, the CDL will also use keyword search.

Emerson Morgan, Metadata Analyst at ARTstor, spoke on behalf of the VRA Standards Committee about the VRA Core and XML. He reminded us that a core element set is not a database structure in and of itself. The Core implies not an exhaustive set of elements, but only the essential ones, the common denominator.  About ten years ago, the VRA began to study what was necessary for its members to share records. This exploration led to the development of the VRA Core, which is based on the Dublin Core, but more closely responds to real life practices in the visual arts. The VRA Core is now in version 3.0, which is an XML schema and has maintained its relationship to the evolving Dublin Core.  Union Catalog of Art Images (UCAI) project is using VRA Core to create a process for developing a union catalog.

Trish Rose, Project Librarian at UCSD, presented an update on the UCAI project which has been supported by a Mellon Foundation grant. The UCAI team created a technical infrastructure for about 750,000 records that maps to the VRA Core and uses identification clustering techniques. The Value Added Services of UCAI include mapping for data conversions, clustering records and merging records to create composite display records, and enhancing searching through controlled vocabulary. The benefits of a union catalog are sharing the workload, reducing redundancy, encouraging good cataloging practice, leveraging complementary streams of action, facilitating contributions to an authoritative collection, and improving access.

This project is a “proof of concept” proto-type. The partnering institutions are: Cleveland Museum, the Harvard museums, UCSD, and more recently Princeton University, University of Minnesota, and the University of Pennsylvania. The goals for UCAI, at this point, are to conduct a needs assessment for the production environment, greater database augmentation, refine and develop new database processing tools and develop a clearer concept of the work unit and composite records.

Jan Eklund, UC Berkeley, presented a brief overview of the VRA Cataloging Cultural Objects project. The guidelines are for the administrative environments of libraries, visual resources, archives, and museums. The CCO creates standards for data structure, values, and content. The document is in three main sections, general guidelines, the elements, and the authorities; plus the appendices, bibliography, and glossary. The first section provides an excellent discussion about database structure in general terms, how to use the manual, relationships of entities in a record, and includes examples. In section two, each element is discussed; giving a description, rules, display and indexing, and examples to show relationships between the record and authorities. Section 3 includes name, geographical, concept, and subject authorities.

The CCO will play an important role in providing a common language between all communities who are cataloging images. It has the flexibility to be used with different data structures, such as Dublin and VRA Core, but will impart a consistency that will allow for more efficient discovery and retrieval.

Did I mention that this was a three day conference. It was amazing what the UC Sliders were able to pack into that time. Between the speakers who themselves were a diverse group of professors, librarians, curators and Information technologist and the audience who represented the same mix, we were able to really dig in to the whole enchilada - creation, access and management - of a digital repository.

I want to acknowledge the UC SLiders, in particular, Maureen Burns, Jackie Spafford and Vickie O’Riordan, including all the people at UCSD who helped her, for organizing this exciting and informative summit.