The University of Illinois/Kodak Digital Imaging Project:
A Report on the Use of the Photo Imaging Workstation
and Related Imaging Projects
September 22, 1995-February 14, 1996
Coordinator for Imaging Projects
Imaging Projects Technician*
May 9, 1996
University of Illinois at Urbana-Champaign Library
Urbana, Illinois 61801
*University Archivist, Bloomsburg University, Pennsylvania
|Motley Theater & Costume Slides|
|Various Art Slides (Krannert Art Museum) Asian, European and American, Ancient Mediteranean, Medieval and Near Eastern Art, and Twentieth Century Art|
|Virtual Reality Tour-African Art Gallery Slides|
|Toxic Plants Slides (Vet. Med. Library)|
|Illinois Historical Maps (35mm film)|
TABLE OF CONTENTS
|Purpose of the Project||1|
|Location of Equipment||3|
|Installation and Training||4|
|Description of Scanning Projects||7|
|Figure 1: Images Converted Using the PIW||8|
|Figure 2: Productivity Rates||10|
|Quality of Input Media||12|
|Digital Image Quality||14|
|Evaluation, and Intended Uses of the Photo CD Images||16|
|Related Imaging Activities||17|
|Future Digital Imaging Directions at the University of Illinois||22|
|Appendix A: Final Report of Operating Technician||25|
|Appendix B: Sample Record-Keeping Log||35|
|Appendix C: Feedback from Photo CD Image Users||41|
|Appendix D: Proposed Survey Instrument||46|
The authors would like to thank a number of individuals whose expertise and support have made this pilot project possible. University Librarian Robert Wedgeworth initiated the discussions in 1993 and has provided a strong Library commitment to moving forward with this exploratory project. Brian Mittelstaedt, Research Scientist at Kodak, and Kodak's Technical Manager for the project, provided invaluable technical assistance and collaborated closely with the UIUC participants throughout the project. Kodak Marketing Researcher Deborah Nicklaus collaborated with the authors to perform evaluation interviews and their analysis. Henry Janis, Director, Markets Development, Education Solutions and Services was responsible for bringing together the right mix of people with common interests in exploring these issues. Engineering Librarian William Mischo and his staff provided space and assistance with equipment setup. Each of the users who supplied either 35mm film or slides was most willing to provide feedback and constructive suggestions for improvements. To them we owe a great debt of gratitude for their time and their input: Linda Duke (Krannert Art Museum); Prof. Richard Betts (School of Architecture); Prof. Mitsuko Williams (Veterinary Medicine Library); Prof. Linda Scott (Advertising); Prof. Ellen Sutton (Communications Library); and Prof. John Straw and Prof. William Maher (University Archives).
The Challenges Libraries, archives, and museums have begun to employ digital image capture and conversion methods to explore two critical missions: preservation, and the provision of greater access to scholarly research materials. In the area of graphical works, several challenges face the scholarly community:
The Opportunities In 1994, University Librarian Robert Wedgeworth announced the University of Illinois Library's commitment to begin exploring the challenges associated with the creation of digital images to provide greater access to under-utilized collections of visual materials. In 1995, the University of Illinois Library and the Eastman Kodak Company launched a pilot project to test the feasibility of operating a production imaging workstation in a library setting, and to collaborate with Kodak to determine the best methods for capturing archival, high-quality images from film and through the use of a digital camera. Materials were selected from several areas on campus that have rich graphical image collections to test whether there would be a variation in the amount of resources needed to digitize various types of materials-human, financial, and technical. Approximately 2,000 images were converted from 35mm film negatives or transparencies to Photo CD format during the pilot test.
The results of the pilot project suggest that the Kodak Photo-Imaging Workstation is a highly effective and high quality means for digitizing 35mm format film materials at the University of Illinois. The Photo CD format offered a very convenient and effective means of capturing images once at high resolution, and storing them in multi-resolutions. The pilot test demonstrated that a production imaging operation using Kodak equipment could be feasibly implemented in a Library setting, and that significant levels of production could be reached with a relatively short training and ramp-up phase. Users of the service had very favorable reports about the speed and accessibility of service, the quality of the Photo CD images, and the personalized nature of the service. The expertise in film to digital conversion provided by Kodak's research scientists proved to be extremely useful in developing a basis for imaging projects.
Next Steps: The University of Illinois Library intends to launch a major effort to create a digital image repository of heterogeneous collections that shall serve the research and instructional needs of humanities scholars on this campus. We shall digitize selected materials from special collections, including the D'Arcy Advertising Archive, the Motley Collection of Theatre and Costume Design, and historical maps of Illinois and the Northwest Territory. We look forward to new opportunities to integrate Kodak's imaging technology and expertise into these future plans.
Purpose of the Project
This project is the result of a series of discussions, refinement in thinking about digital imaging, and mutual interest in studying the process of digitizing materials in the academic library setting. In 1994, the University of Illinois Library and the Eastman Kodak Company began discussing how we could use Kodak digital technology to help us capture and preserve surrogate images of rare and fragile collections, and to provide broader access to under-utilized graphic materials both for research and instructional purposes. The initiator from Eastman Kodak was Henry Janis, Director for Midwest Educational Marketing. In September 1994, University Librarian Robert Wedgeworth appointed an advisory team of teaching faculty and librarians to explore these issues with Kodak imaging experts in a series of meetings. Various aspects of digital imaging were explored, including a potential three-way collaboration involving another corporation.
As the conversations evolved two points emerged that served to solidify the relationship between Eastman Kodak and the University of Illinois Library. First, the University Library was interested in exploring the ways in which Kodak's Photo CD digital imaging process could assist it in producing high-fidelity digital images in an efficient and cost-effective manner. The University Library's second goal was to become familiar with the process of producing digitized images, as well as the variety of challenges inherent in attempting to digitize different sizes and formats of graphic materials. It was important for us to be able to experiment within the Library setting to determine whether and to what degree digital imaging might be integrated into the Library's existing operations. We sought to understand what equipment and technical expertise, as well as human resources, were necessary to begin selecting, building, and organizing digital image collections.
In order to accomplish this work we first needed to begin building a local image collection. In the summer of 1995, Henry Janis and Brian Mittelstaedt of Eastman Kodak proposed that Kodak install a PIW in the University Library so that we could carry out a series of pilot digitizing projects. In doing so we would be able to explore the related issues of staffing, media, image quality, and intended uses of the images once they were digitized. The University of Illinois agreed to digitize approximately 1,000 images over a period of several months utilizing the PIW. A project team was formed, including personnel from Kodak--Henry Janis, Project Director, and Brian Mittelstaedt, Project Technical Director, and personnel from the University of Illinois Library--Beth Sandore, Project Director, and Robert Wedgeworth, University Librarian. An agreement was drawn up between Kodak and the University which outlined the common goals as well as the individual interests of each party. The agreement specified that Kodak would loan the University of Illinois a PIW and supplies over the period from September 25, 1995, to January 15, 1996. The University of Illinois Library agreed to provide a suitable location for the PIW, to staff and operate the equipment, and to provide selections of materials from both Library and other campus collections to be digitized using the Photo CD format.
One collection--the D'Arcy Advertising Archives--was of particular interest to both Kodak and the University of Illinois. Kodak had expressed an interest in digitizing a portion of the approximately 1,800 magazine advertisements in the D'Arcy collection for Kodak cameras and equipment. The D'Arcy Advertising Archives is one of the largest collections of American advertisements, with over 750,000 items dating from 1880 to 1970. The University Library also had an interest in digitizing the ads to make them more accessible and to capture as much of their color and brilliance as possible before additional paper deterioration occurred. Furthermore, the manner in which we handled questions of copyright permissions, where they arose, were of interest to both parties due to the newness of legal work in this area.
This White Paper provides an account of what we learned during that period about the quality, technical aspects, staffing, training, record-keeping, organization, and intended use of the Photo CD images produced by the PIW.
Location of Equipment
The PIW was shipped to the University of Illinois in mid-September, 1995. After preliminary consultation with Kodak and Library personnel, we decided to place the PIW in the Digital Imaging Laboratory of the newly constructed (1994) Grainger Engineering Library Information Center. One of the objectives of the Library's digital imaging initiative was to extend the Grainger Library's facilities to support exploratory technology initiatives for the Library system as a whole. This project provided an excellent opportunity to use our Digital Imaging Laboratory as a testbed for Kodak's technically advanced digital imaging equipment. The Lab's primary use has been for multimedia development of locally produced information retrieval software for the Library. However, there was a fortuitous overlap in terms of the equipment and software needed for in-depth image conversion, manipulation, and editing. Other equipment in the Lab included a Hewlett Packard (HP) ScanJet IIc flatbed scanner, a Nikon slide scanner, an HP color Laserjet printer, approximately four high-end Pentium PC workstations with high-resolution oversize monitors, and three Pentium PCs with graphics accelerator cards and 15"-17" monitors. Image processing software already in place in the Lab included Adobe Photoshop, Adobe Premiere, Lview Pro 1.0, ImageKnife, and other assorted utilities. All workstations except the PIW were connected to the campus' TCP/IP network (100 MB/second between buildings and 10 MB/second within the building), as well as a Novell network. Over the course of the fall semester access to a server was arranged with at least two Gigabytes of storage allocated for digital images. This arrangement was made to support the future placement of Photo CD images on the Library's Web site.
The PIW was placed in a corner of the Digital Imaging Lab that had very little traffic. The scanner, thermal printer, SunSparc workstation, and CD writers occupied approximately 150 square feet of floor space. Three modular tables, approximately four feet in length, thirty inches in depth, and twenty-nine inches in height, were used to house the equipment. The arrangement was comfortable and could accommodate two people seated working simultaneously, with standing room for one additional person.
The Digital Imaging Lab is located on the lower level of the Grainger Library. The Lab contains approximately 1200 square feet, and has one bank of glass windows that runs approximately 12 feet on one side facing a hallway. It does not have windows that provide direct sunlight to the room. The Lab's lighting is adjustable, although not by rheostat dimmer. Two types of light switches can be operated either separately or together--1) fluorescent lighting, positioned so that it illuminates upward and is mounted on bars hung from the ceiling, illuminates the room completely; and 2) halogen recessed lights in the ceiling which provide a low level of indirect illumination to the work area. Electrical connections (110v) were available. However, we needed to work around the PIW's requirement of a 220v connection with an adapter plug. In planning meetings with Kodak personnel we determined that this site would be the most favorable of all possibilities because of the flexible arrangement of furniture and lighting, and the availability of imaging-related software, hardware, networking, and technical expertise.
Installation and Training
The installation of the PIW and the training of Library staff occurred over a two-day period at the end of September. The PIW equipment was set up by Kodak technician Jack Kester and Research Scientist Brian Mittelstaedt. The Kodak PIW configuration included the following equipment and software:
The hardware itself is approximately three years old. Kodak's estimate of the current value of the components is approximately $65,000.
A five-hour training session was conducted by Jack Kester and Brian Mittelstaedt. Two members of the Library staff, Robert Dunkelberger and Beth Sandore, were trained initially in the operation of the PIW. The training session was used to familiarize us with the PIW software, hardware, and the process of scanning, editing, and writing images to Photo CD. Potential problems related to hardware and software, and their probable solutions, were also reviewed during this session. We were trained to operate three attachments to the 35mm scanner, including the slide gate, the film roll automatic gate, and the film roll manual gate. Kodak provided us with a technical support hotline number. We also had the option of contacting Brian Mittelstaedt, Kodak's Technical Liaison, with questions about the PIW and image or film quality. This arrangement was mainly satisfactory, although there were several instances, as noted in Appendix A, the Report of the Operating Technician, pp. 28-30, where the help line was unable to provide all the necessary assistance. One service call was made by Jack Kester to service the printer. In the second instance, a complete shut-down and re-start of the unit served to resolve a software problem.
Kodak supplied printed training and reference guides, including a quick-reference flip chart, a reference manual, and copies of Kodak PIW-related newsletters. Of all the materials the flip chart proved to be of the most value. The manual provided was actually produced for a newer version of the PIW software (PCD 4200), which was not installed in the PIW at the University of Illinois. The newer version offered more features, such as additional image editing and cropping capabilities. In several cases referring to this manual with in-depth questions about the capabilities of the PIW-resident software proved to be confusing, since the software did not possess the same capabilities that were referenced in the manual. This was not, however, a significant problem. Brian Mittelstaedt was very accessible by telephone and e-mail, extremely knowledgeable about the capabilities of the PIW software and hardware, and always willing to discuss questions that would arise during the course of our work.
The training in the operation and troubleshooting for the PIW itself was straightforward. Interesting complexities were posed by the work of getting the film terms and image editing settings correct in the cases where editing was required before image conversion. In those cases where we needed to correct for brightness/contrast or color density it was important to have at least a rudimentary understanding of the principles of imaging, and which types of changes would interact positively with the type of film being scanned. In these cases, Brian Mittelstaedt worked with us to impart a solid understanding of the operations we could perform to correct for film or photographic inadequacies before scanning and writing to disk, whenever possible.
Many start-up supplies were provided by Kodak as part of the agreement. Some of these included Kodak products, such as Photo CDs, thermal ribbons, and thermal paper. Kodak furnished an initial supply of 25 Photo CDs, 100 sheets of thermal paper, and one thermal ribbon for the printer. Other useful office supplies included a paper cutter for cutting index prints to fit the CD jewel cases, a small camel hair brush and cans of compressed air for cleaning film, and cotton gloves for handling the film. A portable Photo CD player and a video monitor were kept close at hand for viewing the finished Photo CDs. We did not often use the Photo CD player, mainly because PCs with image display and editing software were readily accessible, and users expressed a preference for viewing their images on the computer screen. Although our lack of direct access to a light table for viewing slides and film was somewhat inconvenient we were able to resolve most film quality questions with very few problems. Nevertheless, access to a light table is recommended.
A limited amount of funding was available to staff and operate the PIW, because of the fact that the project plan was presented to the University after the fiscal year 1995-96 budget had been allocated. The time for planning and selection of projects, as well as staff, was quite limited. Due to these limitations we determined that we could commit approximately 100 hours over a period of several months to the operation of the PIW. Additional time above the 100 hour limit would have to be contributed by staff supplied from other units. While we welcomed the opportunity to train staff from other units in the operation of the PIW, we also felt that this scenario presented supervision and quality control challenges that could not be adequately addressed during the course of a short pilot project. Bearing these limitations in mind, we adopted the approach of staffing the PIW with an individual who could work in tandem with the UIUC Project coordinator to manage the PIW work schedule and space, establish production priorities, review film quality and consult on image editing decisions, and participate in preparing the White Paper.
Robert Dunkelberger was appointed as the project Technician. Mr. Dunkelberger possesses a Master's degree in Library and Information Science, and has several years' experience in a professional appointment in the University of Illinois Archives. Mr. Dunkelberger was the ideal candidate for this position because he has a keen sense of bibliographic organization and archival materials handling. Mr. Dunkelberger quickly understood the scope of both the technical operations, as well as the need to provide unique identification for materials, and the care with which original and surrogate images needed to be handled and converted.
We had the opportunity to train Mary Shultz, a Graduate Assistant in the Veterinary Medicine Library, to digitize several hundred of the slides in conjunction with the Horse Diseases project of 962 images. Mr. Dunkelberger trained Ms. Shultz in approximately two hours, and she was able to attain a production rate of approximately 50 slides per hour. Ms. Shultz worked ten of the twenty-six hours it took to complete this particular project. She is a graduate student completing a professional degree in the UIUC Graduate School of Library and Information Science.
Description of Scanning Projects
Seven projects were completed during the course of the pilot project. The initial objective was to choose a group of projects that would represent a wide range of imaging needs, including archival and special collections, instructional, and multimedia uses. This need for a number of diverse projects was supported by the fact that slides (35mm) were the primary medium for many of the existing non-digital image collections on the UIUC campus. Included among these is the School of Art and Design's Slide Library, which possesses over 500,000 slides used for teaching and research. Individual faculty in other teaching departments have also amassed extensive personal slide collections, which are maintained in a decentralized fashion. Other units on campus have slide collections, such as the Krannert Art Museum, which filmed its holdings, and the Rare Book and Special Collections Library's set of slides of theatre design. Libraries and Museums commonly employ 35mm slides and 4"x 5" transparencies to provide access to surrogates of fragile and valuable materials for students and other user groups. The PIW was necessary for completing these projects, because even though a number of units on the UIUC campus were known to have slide scanners, most of this equipment does not support high volume scanning activity. We were interested in testing how the PIW could accommodate projects that had different objectives and different quality and resolution expectations. The seven projects that were carried out during the course of the study included:
FIGURE 1: IMAGES CONVERTED USING THE PIW
|Project Title||Number of Images Converted:|
|1. Architecture Slides||100|
|2. Motley Theater & Costume Slides||114|
|3. Various Art Slides (Krannert Art Museum) Asian, European and American, Ancient Mediteranean, Medieval and Near Eastern Art, and Twentieth Century Art||186|
|4. Virtual Reality Tour-African Art Gallery Slides||132|
|5. Toxic Plants Slides (Vet. Med. Library)||174|
|6. Horse Disease Slides (Vet. Med. Library)||962|
|7. Illinois Historical Maps (35mm film)||37|
|Total images converted:||1705|
The slides that were scanned for the Architecture Department and the two Veterinary Medicine projects will be used for instruction and research purposes in the classroom, the office or home, and the library or clinic. The images scanned from the Motley collection and the Illinois Historical Maps were selected from the Library's special collections. The slides scanned for the Krannert Art Museum will be included in the Museum's Web site. The set of images from Krannert's African art gallery are to be incorporated into a Quicktime VR virtual reality tour with the collaboration of Apple computer.
A total of 1705 images were selected and scanned for conversion to Photo CD format throughout the course of the project. The bulk of the digitizing was completed by January 15, 1996, the initially planned date for the project to conclude. Although the project was technically complete, the PIW remained operational until February 15, when it was disassembled and packed for shipping by Jack Kester of Kodak and Robert Dunkelberger. During the week of February 12 the seventh project was completed, which involved the conversion of 37 images of 35mm film of historical maps of Illinois. This final project gave us the opportunity to work with roll film and to test the feasibility of the manual film gate.
During the course of the project a total of 31 Photo CDs were used. Six Photo CDs in addition to the initial set of twenty-five were supplied by Brian Mittelstaedt in December when the order for the Horse Disease slides was begun. Approximately 120 images can be stored on a Photo CD, depending on the number of sessions in which the images are digitized and the number of images that are to be digitized for each job. Although the total number of images digitized amounted to approximately 1700, the discrete nature of the jobs did not allow us to optimize the number of images placed on each Photo CD. Furthermore, we agreed to make copies of a number of the CDs and to furnish these copies to Kodak for technical review. Copies were not made for all the projects completed; instead a representative sample was selected. This included copies of the Photo CDs of Architectural photographs and drawings, the Toxic Plants, the Krannert Art Museum Asian Art photographs, and the special collection of Motley Theater and Costume Designs. We felt these Photo CDs would provide Kodak with a good representation of the types and condition of the film encountered during the course of the project.
As mentioned in the section on Staffing we were limited in the number of hours we could operate the PIW, based on internal budget constraints. The total amount of time in which the PIW was in operation was approximately 80 hours, from September 25, 1995-January 15, 1996. During that time, 1705 slides were scanned, selected, edited if necessary, and written to disk, index prints produced, record-keeping lists typed, and film quality decisions executed. Since the total number of hours represents time spent on all aspects of the project, it is important to separate this information from the data about the rate of scanning. The information in Appendix A, the Final Report of the Operating Technician, suggests the following scanning rates per hour for each project:
FIGURE 2: PRODUCTIVITY RATES
|Project Title:||No. of Images:||Images/hr.:|
|Motley Theater & Costume Slides||114||16|
|Various Art Slides (Krannert Art Museum) Asian, European and American, Ancient Mediteranean, Medieval and Near Eastern Art, and Twentieth Century Art||186||23|
|Virtual Reality Tour-African Art Gallery Slides||132||30|
|Toxic Plants Slides (Vet. Med. Library)||174||30|
|Horse Disease Slides (Vet. Med. Library)||962||42 (Ave.)*|
|Illinois Historical Maps (35mm film)||37||N/A|
*Average of all rates provided
In viewing the rate of scanning over time, it could be reasonably stated that the productivity level did rise over the course of the project. Only projects one through six are considered here, because the seventh project used the manual film loading gate, and required that considerable time be spent in choosing the best of two frames for each image that was digitized. It is difficult to determine whether or not the rate of production increased over time simply due to familiarity with the scanning process. We believe it is possible, as Mr. Dunkelberger suggests in Appendix A, that the nature and condition of the material being scanned also had an effect on the rate of scanning.
For example, the Horse Disease slides, the final slide order completed, required very few adjustments. The film quality was excellent, but more important was the fact that the operator did not need to scrutinize each frame to ensure that no fine tuning was required for detail and film quality. The total number of hours devoted to each project also included the time spent compiling the image list. If the operator was not compiling an image list either before, during, or after the job, the rate of scanning could possibly be higher. However, in the case of slide scanning, there is a wait time while an image is being scanned and brought up on the monitor. This time period usually allowed ample opportunity to note the information about the image that was included on our record-keeping instruments, which were Excel spreadsheets.
Records of the imaging work were kept using a Microsoft Excel spreadsheet program. The Excel spreadsheet program was adapted from a spreadsheet used by the Kalmara group at the Canadian Museum of Civilization (CMC) in Ottawa, Ontario. On a visit in late August to the CMC, Beth Sandore observed the photography and PIW imaging operations of this group, which proved useful in planning for the setup, staffing and record-keeping for the UIUC project. An example of the spreadsheet used for the projects completed at the University of Illinois is included in Appendix B, Sample Record-Keeping Log. The following elements were incorporated into the log:
Often the user requesting the images did not supply a list of the items to be scanned. In the one case in which a list was supplied, most of the detail provided was not useful for our purposes. In the cases where there was some type of unique numbering system provided with the film or slides, that system was used in the Item # field. The Motley and the Asian Art slides all had unique accession numbers which made them easily identifiable for us and the user. The two instances in which image lists were not compiled included the Horse Disease slides and the Illinois Maps film. In the case of the Horse Disease slides, because of the sheer size of the order (962), a summary of the work performed was provided. Only the problems were listed on the summary sheet, utilizing the same conventions used on the spreadsheet. In virtually all cases any film or slide image that appeared at less than "good" quality had an accompanying "problem" description on the spreadsheet. This point will be discussed more fully in the next section, "Input Media." It bears emphasizing that although tedious, record-keeping was useful because it enabled us to pinpoint both specific and recurring problems with the input media. It also assisted us in reviewing the finished product with the user.
Quality of Input Media
The quality of the film or slides directly affected the quality of the scanned image. The first encountered problem with film or slides occurred when the type of film used for the photography could not be identified. In the case of slides, the default "E6" film term was used when the type of film could not be specified or identified on the slide. This guesswork appears to have resulted in satisfactory output, based on the feedback from the majority of our users. In the single case where roll film was scanned, the correct film term was supplied (Kodak Royal Gold 25). Specific problems with input media could be classified in three categories--exposure errors; problems with film/lighting/color; and other physical problems that were not the result of the photographic process. Some examples of problems encountered with the film or slides included:
Over- or under-exposed:
"night photo, slightly dark"
"faintly yellowish image"
"slightly reddish image"
"slide mounted in reverse"
"header obscured by frame"
"hand at top of frame"
For each project we requested that users clean their slides thoroughly before presenting them for PIW work. Only in one case did we encounter excessive dirt buildup on the slides, which were approximately fifteen to twenty years old and had been mounted in Gepé glass mounts. Most of the dirt was trapped between the film and the mounting, which could not be removed for cleaning. The resulting digital images contained dirt which was embedded in the emulsion. This produced scanned images with a significant number of dirt specks which would pose even a post-scanning editing challenge.
The initial discussion with the users about the PIW work always took place in the Digital Imaging Lab. In all cases, we invited the user to preview one or two slides in the scan preview mode to determine whether the image would be acceptable. We also offered to let users see examples of other Photo CD images that had been scanned using the PIW. During that initial conversation we asked the user to identify any known problems with the quality of the film. Each person gave candid, and usually accurate assessments about the quality of the film or slides.
Brian Mittelstaedt's expertise proved invaluable in determining the appropriate level of brightness or contrast adjustment, and color density adjustments that ought to be made using the PIW. If we discovered any problems during the scanning process that were not brought to light during the initial discussion, we would experiment with alternative settings and offer a potential solution if the PIW capabilities offered them, such as adjusting brightness/contrast, or color density. If the user chose to work with the proposed solution, we would implement it. If not, we simply scanned the images as they appeared. In the case of the Toxic Plants slides, Prof. Mitsuko Williams requested that we not make color density adjustments. She indicated that she had compared and found the color of the plant leaves that appeared in the slides to be very close to that of the actual toxic plant leaves. She explained that the color of the leaves is critical to the correct identification of the plants.
In the case of the African Art Gallery slides Mr. Mittelstaedt's expertise was crucial, because the lighting used in the gallery for the shoot was insufficient and produced a yellowish tint in the slides. An attempt was made to correct for this by adjusting the controls for brightness and color density (See Appendix A).
Digital Image Quality
Kodak and the UIUC agreed at the outset that acceptable standards for the quality of the images produced using the PIW would vary by case. We agreed to establish and modify standards in an ongoing manner. We also agreed that we would not attempt to convert images that required a level of image quality that exceeded the capability of the PIW. We welcomed the opportunity to consult with Brian Mittelstaedt for advice in several cases. Throughout the course of the project we developed two common sense guidelines to determine whether 35mm film or slides would produce acceptable quality Photo CD images:
These guidelines have been simplified so that they can be applied to a variety of graphic materials. They are the result of several months of discussion and work with photographs of three-dimensional objects, line drawings, water color paintings, maps, and magazine advertisements. In the case of complex images we experimented in the early fall with a sample of images from the D'Arcy collection. We found that for large and complex ads, shot with 35mm Kodachrome film, the resulting Photo CD image appeared grainy and did not produce clear, sharp details in Base * 4 beyond with a 50% zoom factor. Mr. Mittelstaedt worked with us in three experiments to select both film and lighting that would render clear, high quality images of these same subjects. For the Illinois historical maps project, Mr. Mittelstaedt assisted us in determining the best combination of film and lighting to produce high quality Photo CD images. We used Kodak Royal Gold 25 film with six 500watt Blueflood lights. That combination produced high quality map images with a low level of grain in the film image. The two other experiments included more images from the D'Arcy Advertising Archives and photographs of sculptures from the Lorado Taft Papers managed by the University Archives. These are described briefly in the section "Related Imaging Activities."
In several cases, users anticipated that they would need to edit the images from the Photo CD to achieve any specific effects that the PIW's image editing software could not provide. Users often indicated that they intended to use Adobe Photoshop or the older Aldus Photostyler.
Evaluation, and Intended Uses of the Photo CD Images
During the course of the pilot project images were digitized for a total of seven projects: two art museum instructional outreach projects, one member of the teaching faculty, two projects in the Veterinary Medicine Library, and two from special collections. Both Kodak and the University of Illinois felt we could obtain valuable information about the usefulness and the quality of the output from the users of the Photo CD images. The initial plan for obtaining this information involved administering a survey form and gathering informal comments and suggestions from the faculty and staff who had benefited from the digitizing of their materials. A survey form was constructed at UIUC and was reviewed by Deborah Nicklaus, a Marketing Researcher at Kodak (Appendix C, Proposed Survey Instrument). However, in late November we learned that both Ms. Nicklaus and Brian Mittelstaedt planned to visit the UIUC campus to carry out another filming session. At that point, we decided that the ideal way to obtain information about the usefulness of the Photo CD images was through individual interviews. Since we were working with a small population it would be possible to conduct interviews over a period of two days. We withheld the survey in favor of conducting these individual interviews.
The interviews were carried out from January 16-19, 1996, on the UIUC campus with faculty who had requested the images. The purpose of these interviews was to determine the intended use of the Photo CD images, the satisfaction of the users with the quality of the Photo CD images, and to explore further academic needs related to digital imaging. The interviews were conducted by Deborah Nicklaus and Brian Mittelstaedt of Kodak, and Beth Sandore of the UIUC Library. Each interview was informally structured, with several base questions posed during the course of the interview, including the following:
1. For what purpose(s) do you intend to use the Photo CD images?
2. What is your level of satisfaction with the quality of the images?
3. What additional digital imaging needs do you have, or do you anticipate having in the near future (software, hardware, services)?
Each interview lasted approximately forty-five minutes to one hour. A summary of each interview, as well as informal comments from users over the course of the project are presented in Appendix B, Feedback from Photo CD Image Users. One further interview was conducted with Robert Dunkelberger and Beth Sandore to determine their general impressions of the accomplishments of the project, and the suitability of the hardware, the software, and the process of image digitizing using the PIW.
Related Imaging Activities
In early January, 1996, Mr. Mittelstaedt shot approximately 450 frames of materials from the D'Arcy and the Lorado Taft Sculpture photograph collections. The D'Arcy images were shot using a 35mm camera equipped with a macro lens and Kodak Royal Gold 25 film, with Speedotron Brown 800 WS electronic flashes fitted with Photoflex Halfdome (small) diffusers. The Lorado Taft photos were shot using black and white film, with a few shot in color for comparison purposes. An addendum to this report shall be released after the results of that experiment have been reviewed and analyzed.
The availability of high quality, production-level imaging equipment such as Kodak's PIW provides the most effective solution to production-oriented imaging needs. This study demonstrated that a high level of production and quality control could be achieved in a library setting using the PIW, for archival as well as for instructional level imaging needs. Experimental approaches to workflow and staffing suggest that startup and training to begin operating the PIW takes approximately three days. This represents considerably less time than configuring, training, and operating separately purchased and installed peripherals to produce the same fidelity and quality product.
Exceptionally high image quality was obtained using the PIW with 35mm film and slides. All users for whom we digitized film or slides of reasonable condition were highly satisfied. The user whose slides were in fair condition recognized that the condition of the transparencies significantly affected the quality of the digitized image. All users felt that the images they received could be incorporated into multimedia web applications with minor editing.
The ability to scan film once and capture an image at high resolution, then produce five output files of varying resolutions, is both time and labor saving. This is a significant benefit of the Photo CD format. For large-scale imaging operations it represents an attractive solution.
Kodak's imaging production system holds great promise for fulfilling academic imaging needs in the library, archival, museum, and instructional communities. Furthermore, the more recent configurations that Kodak has debuted offer even greater flexibility to address varied imaging needs. We found the PIW 2200 worked well for digitizing most materials in 35mm format. The quality of photography proved to have the most significant impact on the quality of the resulting Photo CD image. We also found that it was possible to achieve the highest production rates scanning images for which little or no image editing was needed. The projects that required little or no individual image editing were for instructional use. For example, we found that only four percent of the slides digitized for the Horse Diseases project required individual editing. In comparison, we observed that archival preservation projects required considerable attention to the detail of specific images. The slides digitized for the Motley Theater and Costume Design Collection required close scrutiny. Approximately a dozen of the slides were re-shot, due to filming quality problems with the first set of slides.
There are essentially two perspectives on the use of digital imaging in the areas of libraries, archives, and museums. The first viewpoint suggests that digital images are intermediaries between the user and the original image, recognizing that the best attempt is made to accurately reproduce the fidelity of the original. This perspective incorporates most accepted standards for high image resolution. The second perspective places digital imaging as a new step in the preservation process. That step involves the capture of a high-fidelity image of the original item, to assist in restoration or in the recollection of the state of a deteriorating artifact. The second viewpoint involves stringent quality and fidelity standards. While both viewpoints recognize that it is impossible to produce an exact replica of an original, those who hold the second perspective are more willing to invest the additional time and resources needed to produce the highest quality images that the technology will currently support. During the course of the pilot project we found that the University of Illinois' imaging needs fall into both of the above categories. The ability to work with the PIW on site, combined with the research expertise provided by Kodak's Brian Mittelstaedt, enabled us to examine the complex and critical relationship between film capture and conversion to digital format. This experience has played a crucial role in our preparation and planning to digitize substantial selections of graphical materials from the Library's collections.
In Appendix A, Robert Dunkelberger suggests that some changes ought to be considered if the PIW is to be used for widescale production of digital images by libraries and archives:
...(M)ore is needed if this system is to be adapted for use by libraries and archives for the digital preservation of images. The digital image to be written on a Photo CD should be of extremely high quality and as accurate a reproduction of the original item as possible, so that it can serve as the archival copy, and remain in that state long after the original image is gone. The excellent, detailed images of the historical maps do show the promise of the PIW system for archival preservation...
The Kodak PIW or a similar imaging workstation produced by Kodak has considerable promise for use in the library and archival community. To this end, we make several suggestions for enhancements to the PIW that would enable it to more fully address the needs of archives and libraries:
These recommendations reflect the variety of questions that we encountered during the course of our own work, and in our discussions with the users for whom we produced Photo CDs.
Expanding the internal record-keeping component would enable the terminal operator to add image information to the records while waiting for images to write to disk. We could see many benefits to changing the PIW software so that it could take advantage of the multi-tasking operations that can be executed on a Unix platform like the SunSparc and its successors. Students and faculty alike are familiar with both these workstations and with Unix applications in a networked environment. It would be interesting to consider whether the Kodak scanner, writers and software could be bundled separately so that it can be installed as one of several packages that an institution could purchase and install on an existing workstation.
Adding an attachment for batch processing of slides would make the PIW a more attractive option for academic institutions. Many faculty, libraries, and archives perform their digitizing work from existing slide collections. This will be the norm for at least the next several years, as institutions work to digitize their current collections.
Expanded image editing tools would benefit those who want to use the Photo CD images as authoritative archival images, from which other high-fidelity, identical image copies could be made. That way, the most critical editing is performed once, and captured on Photo CD. This ensures an element of consistency in the type and the amount of editing that is performed on the image the institution saves for archival purposes.
A larger, higher-resolution video display monitor can aid the PIW operator in making image quality judgments. At times, it was nearly impossible to make definitive image quality judgments between two apparently identical images because of monitor limitations. The only choice was to digitize both images and to review the images later using image editing software.
The two final suggestions relate to networking and enabling the PIW software to incorporate a variety of image file formats for writing to disk. We understand that the current PIW's are produced with TCP/IP and other networking capabilities. Networking the PIW would allow for the use of the workstation for several purposes, and for the import of image files from other sources. Importing and writing these image files would greatly facilitate practical needs of teaching faculty who want to combine images from many sources, which have been digitized in various formats, on one disk for greater portability. Image arranging and organizing software does not yet provide the same degree of physical convenience and portability as does a slide tray.
Clearly there are other approaches to digital images for libraries and archives. Another logical approach is to out-source the capture and conversion of images to a commercial firm, with the capture being done on-site and conversion done off-site. This model would likely work best for institutions who cannot justify the cost of the initial equipment and software investment. However, a number of large institutions are currently seeking digital imaging solutions, and could possibly cost-share a purchase such as this across campus units (consider the potential with approximately 120 Association of Research Libraries member libraries in the United States and Canada).
Future Digital Imaging Directions at the University of Illinois
In the short term, we intend to integrate the digitized images that were produced using the PIW into the instruction and research programs at the University of Illinois. The Library collections will be made accessible through the Web. At this time, most of our development projects view the Web as the ubiquitous vehicle for making multimedia materials accessible across the campus network and the Internet. The Library's long-term goal is to build an image repository, consisting of images and descriptive metadata, which serves the UIUC user community across the Web. We are excited about the recent breakthrough by Kodak and NCSA (National Center for Supercomputing Applications, UIUC) making Photo CD format images accessible through a special version of the Mosaic software. Similarly, we have high hopes that Netscape, being the Web application of popular choice, will soon incorporate a Photo CD viewer.
There are several challenges to making images accessible through the Web, including search software improvements and file transmission delays. With the current focus on Web application development and image compression and delivery innovations, we believe these problems shall soon be resolved.
The Library, in tandem with faculty in the Beckman Institute has begun developing prototypes for image search and retrieval systems. One prototype has been developed as an extension of the software prototype for Illinois' Digital Libraries Initiative (DLI) project. The DLI is a multi-year $4 million grant project funded by the National Science Foundation (NSF) to provide access to the full-text and drawings of over twenty Engineering and Computer Science journals. The prototype that is being developed for the NSF project has been adapted to work for image and text retrieval. The search queries in this prototype are text-driven. A second prototype is being developed in conjunction with the Beckman Institute for Cognitive Science and Artificial Intelligence. This prototype focuses on content-based image retrieval, and the use of novel methods for image compression and decompression. This project is supported as a supplement to the NSF Digital Libraries Initiative project at Illinois. Over 4000 images and accompanying descriptive texts licensed from seven U. S. museums have been used to populate these prototype databases. The art and cultural images have been made available through the University's participation in the Museum Educational Site Licensing Project, sponsored in part by the Getty Art History Information Program (AHIP). Illinois' work with prototype system development shall be in progress for the next eighteen months.
In addition to the above research, many teaching faculty are building their own personal collections of digital images to use in classroom teaching and for individual research. Photo CD and other CD-ROM image products are a preferred medium, because they are portable, can be used with stand-alone image organizing software if desired, and are not dependent on network performance or availability. Some faculty are using stand-alone image management systems to organize their own image collections. Kodak Shoe Box and the Canto Cumulus software are perhaps the two most commonly used. Many faculty on this campus who work extensively with images depend on the Web for mounting their own image exhibitions.
Several faculty members have experimented with the Kodak Portable Photo CD player, being enamored of its portability, with the added potential for viewing a CD of images at home on the TV screen. This presents a technology even more flexible than the slide carousel, with virtually every household having a TV, and a growing number of campus classrooms being equipped with video monitors. Ideally, a faculty member could take a CD of images home the night before a lecture, select, mark, and save the images in a sequence to complement the next day's lecture. The portable Photo CD player and the CD could be easily carried to class and hooked up in a matter of minutes.
Work with digital imaging and image collections is under way at the University of Illinois. With the generous assistance of the Eastman Kodak Company, we have had the opportunity to test some important technical and production aspects of digital imaging. We feel that we now have a better understanding of our needs. Through this project we have demonstrated that it is possible to integrate a production-level imaging installation into a library's operational structure. It is clear to us that the Photo CD format has great potential as the high fidelity format of choice for a significant percentage of materials in the library and archival realm. We have also found from this exploratory project that Photo CD works extremely well as a format to support the development of multimedia instructional projects. The unique combination of Kodak expert technical advice with access to the PIW will enable us to expand our planning to anticipate using digital imaging to capture and convert larger portions of our rare and special collections. We look forward to the opportunity to test further new technological developments by Kodak.
Appendix A: Final Report of Operating Technician
Appendix B: Sample Record-Keeping Log
Appendix C: Feedback from Photo CD Image Users
Appendix D: Proposed Survey Instrument
Return to Home Page of the Digital Services and Development Unit