Association of Research Libraries (ARL®)

Michael Ester, President

Luna Imaging, Inc.

Systems which integrate text and images have long captured the imagination of the art world. Initial experimental attempts using electronic images have since given way to a near explosion of image databases and multimedia projects in related fields. Despite the much needed attention paid to technical issues, there has been little inquiry into the uses of visual resources by art professionals or the ways in which computer systems and electronic publications can be made sympathetic to research practices. Drawing primarily from a study conducted by the Getty Art History Information Program, several considerations of image use are developed in the context of the electronic user environment.

Introduction

When I became Director of the Getty Art History Information Program in 1985 it seemed entirely appropriate that an organization whose mission was directed toward the application of automation in the visual arts should be concerned with developments in computer imaging. Although I previously had been involved in computer graphics and geographic information systems, it was obvious that there were no viable systems available to the art community that offered any immediate promise of practical application for large-scale collections. Therefore, about six years ago, we began creating a context where we could explore what it is that people in the arts do with images -- how they use them in their work, and how we should be shaping technology to address the interests of these professionals.

Our basic approach was to set up day-long meetings in which participants could learn about and see key features of image technology, and where they, in turn, could offer their experience in two important areas: their assessment of differences in image quality, and their views and practices of using existing photographic materials. A total of nine such sessions were held at Getty offices in Santa Monica, California, and at the National Gallery of Art in Washington, D.C. Approximately seventy art professionals attended from the United States and Europe. Although the general term "art historian" is used for convenience, the study encompassed a broad range of professionals, including researchers, curators, catalogers, collections managers, photography specialists, visual resources curators, and senior staff of art institutions. Typical of many individuals in the field, participants often divided their time among several of these responsibilities. Although our natural target audience was the art world, much of what we encountered had close parallels in other fields that rely on visual collections.

A more detailed account of the study method and the results on viewer perception can be found in Ester. [1] More recently, I have been speaking about some of the most important outcomes of our work which emerged -- not from our original objectives -- but from encountering unanticipated results in the course of preparing for the sessions. In particular, we found that the image metrics of resolution and dynamic range are insufficient for defining image quality, and that the process of capture is a crucial component in production. This led us to handling some of the most challenging characteristics of art reproductions, and to developing the issues and framework for creating digital archive collections and derivative versions of images for use and distribution.

The context of addressing a mixed audience from the library, scholarly, and publishing communities, gives me an opportunity to discuss more fully the third part of our study. In the final session with the participants, we covered the different photographic and print media they used, the ways they collected and assembled material, and how they employed reproductions in their work. We also explored some of the limitations they encountered with conventional reproduction sources. While an open format was used, the intent was to summarize what was said under recurrent functional themes that could affect the development of electronic publications and their forms of use.

While I cannot cover all of the ground that emerged from these sessions, I can offer a few examples to indicate the range of user issues and implications that seem quite important. Both in the course of covering these points, but more explicitly toward the end of this paper, I want to make a few observations on the suitability of prevailing models for representing and distributing visual collections and associated text information.

Physical Juxtaposition of Images

Consideration of current database systems that incorporate visual materials provides a useful entry into my talk. In these systems images typically exist as a passive field of a text record: finding an object that satisfies certain search criteria and an associated image appearing along with it; alternatively, finding several objects that may produce a screen full of small images. Although the sophistication of interaction, access, and interface vary, this same basic format is nearly a constant. The same static role for images is also true of today's multimedia titles. Regardless of the flexibility or liveliness of the branching interaction, there is a certain inevitability to reaching a destination where an image is locked into a surrounding frame of text.

Yet, one very prominent theme we found is that art historians use physical juxtaposition of images to develop conceptual constructs. Stated simply, professionals use physical arrangements of images as a way to think. Arrangements may reflect stylistic relationships over time, different artistic or historical expressions of a similar theme, separate studies by an artist for an evolving whole, or the visual outline or story board of a scholarly argument. Characteristically, participants recounted diverse methods for ordering material, from tacking images on the wall, to spreading them out on a table, to repeatedly dealing out -- like a deck of cards -- piles of related photographs according to progressively refined criteria. An electronic system which does not incorporate such features is missing an opportunity to transfer a manual activity, both fundamental and natural to research practice, into the automated environment.

The Problem of Scale

Let me choose the simple problem of scale, both as a shortcoming viewers experience with illustrations and photographs, and as an area where the computer context can offer assistance but can also raise unfamiliar problems. A frequent comment from art historians about reproductions is the way that photographic media homogenize the physical size of art objects into standard print and film formats. Sculpture that appears monumental in a transparency may turn out to be, in fact, quite small. In the common situation in which two slides are projected side-by-side for comparison, it may be impossible to tell that two paintings, for instance, are of vastly different scale. Generations of students have been left with false impressions about object size through the lay-out considerations of popular texts. Something of the significance and limitations of illustrations comes across in experiences, which some of you may share, in confronting a well-studied work of art for the first time having only seen it previously in reproduction. These are often strong remembrances, and commonly evoke the mixed experience of finding the great original initially diminished by its actual smaller size, and yet a sense of rediscovery in the real work of art.

Where might the use of computer images figure into a sense of scale? Fortunately, one of the most commonly documented attributes of an object is its absolute size. Although measurements such as dimensions are not without their problems -- for example, was a painting measured with or without the frame -- such information is usually present in the description. Access to the measurement values themselves does provide the essential information, but we can have a strong intuitive sense of scale in the context of the computer display. A screen view of an object outline relative to a fixed scale, or to the size of another object, is a way of giving the viewer this kind of immediate reference point. An approach which the participants voiced themselves was to show the image at the true physical size of the object. Putting aside extreme cases such as standing architecture, participants said this would be extremely useful even if they could only see a portion of the object at actual size which would inevitably occur if the object was bigger than the screen. The potential to visualize the intended scale outweighed the restricted field of vision. I would like to note that the different ways of showing images that evoke strong interest can also be transitory. For example, once having seen an image at object size, there was no particular motivation to continue to work with a full-sized representation for other purposes.

Let me also use the problem of scale to make the point that we still have many lessons to learn about user response to viewing images in a computer environment. Such a situation arose when I was showing an illuminated manuscript page on a computer display to a curator in this particular field. The reaction was a profound double-take: on the one hand, he was captivated by the image quality; on the other hand, there was a strong negative reaction to the "wrong" color which was judged overly saturated and too "hot." These opposing reactions were only heightened as we looked at parts of the page at higher resolutions. However, when we subsequently compared the digital image to the source transparency from which it was derived, we could find no apparent differences. How to account for this contradiction of perceived but not actual difference?

What emerged in the course of further discussion was that the actual page itself was only slightly larger in size than a 4" x 5" transparency, and yet the digital image had been shown unthinkingly at the full dimensions of a 19" monitor. In effect, the manuscript page was seen at an unfamiliar scale that conveyed novel color intensities. Filling the screen with detailed close-ups made the situation worse by presenting even more arbitrary color combinations. When I resized the manuscript image to the actual size of the manuscript page, there was complete satisfaction with the image color. As an afterthought, I asked whether this same problem occurred with projected slides of manuscript pages. The immediate response was that accurate color representation was not expected with slides. The strongly persuasive fidelity in the digital reproduction had made the unanticipated sense of color all the more troubling.

Art in Context

If photographs homogenize their subject, a very similar general theme is the problem of art-in-context. Photographic reproductions are an extremely isolating format. The perspective of a sculpture as it was meant to be seen, the syntax and grammar of murals in a church, and the original articulation of furniture in a room are all relationships that are undermined by standard reproductions of the single object. In the human experience of visiting a church, for instance, we are not likely to be eye-ball to eye-ball with a roof-top gargoyle, or to look down the nave from the church ceiling.

Perspective views, diagrams, physical models, and the inclusion of scales, etc. are among the remedial solutions often used to compensate for these limitations. Yet recognition of the art-in-context theme and the attendant limitations of photographic reproductions raise several new possibilities in the electronic medium where association, resizing, and arrangement of images in space are particular strengths.

One direction to be approached with considerable caution is the use of computers to generate a third dimension of analysis. Steering a course for effective computer use of three-dimensional space will mean focusing on the objectives for showing spatial relationships and keeping applications simple. Providing convincing 3-D models can be an expensive, processing intensive, and highly labor-demanding proposition, on an order of magnitude of more work for each increment of realism. More tractable solutions for individuals and personal computers would emphasize putting two-dimensional images on flat surfaces and articulating these surfaces in a three-dimensional space. While the effect would be something akin to a building facade in a Hollywood set, the articulation of visual content and the opportunity to explore different viewpoints would remain intact.

The Use of Visual Materials as a Process

We learned that what art historians may want to do is often dependent on a phase of their research. It is accepted lore that researchers want access to great quantities of visual material and, moreover, that poor image quality is an acceptable trade-off to large volume. This assumption has been one of the arguments for the reason art historians would be content with videodisk publications which provide only television quality but have high image capacity. Participants in our sessions confirmed this impression up to a point. Certainly one of the distinguishing traits of professionals is their extraordinary visual memory; many placed a high value on familiarity with wide-ranging visual materials. It is also common that as they begin a project they want to immerse themselves in as rich and varied a corpus of material as possible. To the extent it could be articulated, this exercise was characterized as a kind of "refresher" in the subject matter or a way of prompting serendipitous discovery.

Yet, it was even more common for art historians to move past this stage, or bypass it altogether, and reverse priorities. Typical of later stages of research, art historians winnow their material to a relatively small working set of images that remains reasonably stable through to the conclusion of a project. A strong premium is placed on image quality for this working set of images. Bringing this understanding back to user design considerations, a system that functions in concert with art-historical practice would address research and image use as a process with a changing balance of activities and needs. Modes of use and available content should allow the researcher to move effectively through quantities of images and still permit concentrated work on selected groups of images.

While access to reproductions is essential to the training and professional life of art historians, I would mention for balance that we had among our participants art historians who never used reproductions in their research or who needed only a handful of photographs or publications which were readily available. Moreover, participants were very clear that no matter how good a reproduction was it could not substitute for the experience of directly confronting a work of art.[2]

Relationship to Current Directions in Electronic

Publications for Higher Education

Against this backdrop of the ways people use visual materials in their work, what are the current directions in electronic media for new publications for higher education? I would maintain that it is by no means inevitable that the arts and humanities will witness image resources comparable to what they currently find today on the shelves of universities and study centers.

To look at one branch of development, I spend a good amount of time at conferences, special roundtables, and demonstrations on multimedia. Several of the media, communications, and computer giants are plowing the rocky furrows of mass interest in new technology with decidedly mixed results thus far. A couple of general observations can be made, however:

• First, these companies are neither directed technically nor are they noticeably interested in higher education or the research community. They and producers of multimedia titles are looking at consumer audiences and would much prefer being offered at Blockbuster Video than in a major university;

• Second, the vast majority of what is called multimedia is really uni-media, in that it consists of video-based stills, motion, and audio. Compare this to current print equivalents -- the collections catalogue, the catalogue raisonée, the compendium by artist or theme. The multimedia industry is ill- positioned to address what might be taken as the print standard: the single-page plate in a high quality art book. In a very meaningful way, in terms we know art professionals can appreciate, significant penetration of multimedia in higher education would dilute or impoverish what already exists.

The technology of multimedia is also innocently misleading as a body of information. If one pulls an encyclopedia volume from the shelf, there is an intuitive idea of the amount information in it and of the larger whole of which it is a part. However, sitting in front of a computer, it is very difficult to know how much information is in a multimedia work. Implicitly, it always represents itself as a major collection with endless branches of access. Yet, all of these works have edges, and are in every way an authored title and not a resource. The very idea of a coherent collection is inappropriate--take away the interactive script, and the images, clips, and sound have no conceptual integrity. Neither as they are captured nor as they are integrated do they represent an accruing body of material. Despite rich navigational aids a user is not searching through a resource collection in the way I have described scholarly activities earlier.

Another important development are initiatives to build national electronic libraries in the humanities. The Getty has been a strong participant in these efforts and there are powerful incentives for the many constituencies involved to move forward. But the idea is almost as basic as wheeling up our libraries to the electronic loading dock and shoveling in all of the books. It has been a considerable effort for us to get people to include primary research materials and visual collections as these are not as tidy or as common as books. More pointedly, there is virtually no thought given to the forms and functional capabilities that should exist for new productions. The ironic implication of this omission is that we may have to create a print publication to know what form the electronic version should take.

Conclusion

Closely related to the issues I have discussed in this paper, the Getty Art History Information Program is initiating an information and standards organization on imaging that will draw on an array of partners from academics, museums, archives, and industry to study the application of this technology in the humanities. There are various issues to consider, including the creation of image archives, clarification of rights and reproductions, documentation of images, technical prospects for imaging, and questions of publication and image use.

At a very concrete level, I have just left my position as one of the directors of the J. Paul Getty Trust to head a new company, called Luna Imaging, which has sponsorship from both the Getty and Eastman Kodak. The company's objectives are to put in place the services, publication resources, and relationships that will build a long-term future for image collections in the visual arts, history, and cultural heritage communities. From the perspective of practical projects, I will mention two of our objectives:

Production. Quite apart from attaining the image quality levels that will continue to serve institutional and scholarly needs, current production practices for capturing images require a radical change in approach if they are to address the very different demands of image collections. In the pre-press world, highly trained technicians typically hand-adjust each scan to anticipate a publishing objective. For building digital archives quality must be accompanied by consistency and throughput: consistency to maintain the same image as it passes through a wide potential of systems and devices and throughput to accommodate volumes of images at tolerable cost.

Publication and use. A first project in collaboration with the Frank Lloyd Wright Foundation will produce and publish 5,000 of Wright's presentation drawings and associated documentation; only several hundred of these drawings have ever before been seen. In conjunction with this initial release we are building a visual environment which inverts current emphasis on text, and creates a working context that is at once more natural to the intrinsic character of visual information and builds upon the way professionals use images in their work.

In closing, I would like to leave you with the same key questions that motivated the initial study conducted by the Getty Art History Information Program, that shaped the focus of subsequent work and presentations such as this paper, and that guided priorities at Luna:

Will conversion of large photographic collections into electronic form satisfy museum and archive needs, and provide visual resources of lasting value to the field?

Will images be distributed into the academic context, and be of a quality that will serve a broad range of research interests?

Will computer environments complement and support the ways art professionals use visual materials in their work?

Will the path of electronic images lead audiences to interact with the original objects in the institutions that house them?

References

[1]Ester, Michael. "Image Quality and Viewer Perception,." Leonardo 23, no. 1 (1990): 51-63, and Visual Resources 8 (1991): 327-352.

[2]Ibid.

Bakewell, Elizabeth, Beeman, William O., Reese, Carol McMichael and Schmitt, Marilyn. Object, Image, Inquiry: The Art Historian at Work. (Santa Monica, 1988): 7-22.

McClung, Patricia A. "Costs Associated with Preservation Microfilming: Results of the Research Libraries Group Study." Library Resources and Technical Services, (October/December 1986): 363-374.