Flint, 'Images in Your OPAC', LITA Newsletter v15n04 URL = ftp://dewey.lib.ncsu.edu/stacks/serials/lita/lita-v15n04-flint-images V15N4.IMAGOPAC LITANEWS ----------------------- Images in Your OPAC Anne Flint This program, sponsored by the PLA Technology in Public Libraries Committee and LITA, included an overview on the subject of imaging and a description of two successful imaging projects that were done in cooperation with local library systems vendors. Imaging Overview Dick Boss (Information Systems Consultants, Inc.) explained that imaging is a group of technologies, including the technologies necessary to capture, store, manipulate, transmit and access images. Electronic imaging can provide access to a wide range of media including prints, photographs and slides. Some advantages of imaging are the reduction of storage requirements, preservation of originals, improved image quality if the original is in poor condition and increased accessibility of material by allowing concurrent users and remote users to view the images. Imaging begins with a scanner to capture the image. There are a wide variety of scanners on the market, although many vendors are only familiar with one kind of scanner, the kind used for document imaging. These scanners are high speed scanners that do not enhance the image and may destroy or damage the original image during the scanning process. At the other end of the scale are the color scanners used by photographers that are used for slower, quality-oriented work. Typically, library companies are not familiar with the full range of scanners available but only know the ones that they have encountered in their work with other clients. Choose the type of scanner that is best for your project. Consider how you must position the original in order to do the scanning and how many dots per inch (dpi) the scanner will scan. Black and white prints are often scanned at 300 dpi, color illustrations at 600 dpi and medical illustrations at 1,000 dpi. It is possible to scan at a high resolution but store and display images at a lower resolution in order to lower costs and reduce transmittal time. Scanning is expensive because of labor costs. It may take a minute or more to scan each image. A high-end PC or Macintosh (or perhaps a Sun workstation) will be needed as a server. It will need to have 16 (or preferably 32) megabytes (MB) of memory for a small project. If you want to manipulate images in order to block them or improve the contrast, you may need up to 1 gigabyte of disk storage. You may also need 2MB of storage for each uncompressed image, or up to 24MB per color image. JPEG is the standard for compressing images. You will need devices for printing and displaying images. You will need a laser printer, and you will need a color printer if you wish to print color images. You may need optical digital disk storage, depending on the project. Estimated equipment costs are $4,000 to $8,000 for a scanner; $10,000 for a server (which usually comes bundled with a display station); $5,000 to $8,000 for a printer; $4,000 for optical digital disk storage. Networking can be a problem for transmitting images, because fairly high bandwidth is often needed in order to transmit images at what most users would consider a tolerable speed. For example it takes 80 seconds per page to transmit a 200 dpi uncompressed image at 9,600 bits per second (bps). One solution is to send images in compressed form, but then it must be decompressed at the end-user's workstation. An IEEE committee is working on a standard for higher transmission rates over twisted pair wiring. Requirements for an end-user system include a 486/33 PC (or its equivalent) with 12 to 16MB memory and a high resolution 17 or 21 inch monitor with at least 1,024 x 768 resolution. Most systems index images in a rudimentary fashion. In some cases, bibliographic records are created for the images; when a user searches, the OPAC alerts him or her that there is an image that can be viewed. There is a link between the server for the images and the automated library system. The OPAC vendor can put a note in the field in the bibliographic record and provide a link to the image server and to an optical digital disk where the images are stored. In most cases library system vendors provide the links to a server only. The other end of an imaging project is the library's problem. There are no widely-available turnkey solutions for imaging projects that interface with the library's OPAC. Most projects to provide images in the OPAC will cost about $50,000. The cheapest ones will be $20,000 to $40,000. One of the largest factors is concurrent users: imaging costs about $20,000 per concurrent user versus about $5,000 per concurrent user for an automated library system. Do an imaging project properly with the right equipment and an adequate budget. Use the appropriate software for the project. Instead of buying one all-purpose scanner, buy several scanners that are each designed for the different media you are scanning. When you buy equipment and software, think ahead to other future applications. Do not rely solely on a library system vendor for expertise, because they do not have much experience working with images and tend to be narrowly focused. Seek out non-library imaging vendors, but look for ones that have expertise in your particular kind of project. Traditional imaging firms are accustomed to working with black and white documents that are delivered over networks. They are not focused on retaining the original or on handling originals that are a variety of different sizes. Photograph supply imaging firms have expertise in single-user color applications. The library system vendor will know how to deliver the images over a network and how to index the images. Response time and system performance will not be the same for images and bibliographic records. It is not unusual for it to take 15 to 25 seconds to deliver an image to the end user. Both staff and patrons must be prepared for this. Comments and Questions There was some discussion of contracting out scanning jobs. You may need to hire a firm with graphics experience for certain projects. They should take a random sample of the work to be done so that they can give you a fixed price for the job. You and the contractor should set a maximum time per image and the contractor should flag items that take the maximum amount of time. Enhancing of images can be deferred if the project is taking a long time. When images are compressed you can choose to lose some image resolution in order to reduce the requirements for storage and display. What is the likelihood that there will be a "bibliographic utility" for images some day? The Elsevier TULIP Project is FTPing images to universities. UMI's microform republishing is moving toward images, although there are still some technical problems. Once a link between images and the OPAC is in place, what do people see if they do not have a high-end workstation? They do not have access to the images if they do not have a high-end workstation. Imaging at Boulder Public Library Ron Abbott (Boulder Public Library) described an imaging project that used CARL's Image Access System, installed in the spring of 1991. So far they have digitized about 5,200 images from a 200,000 piece collection of maps, photographs and other images. The images are available at four branches of the library. Design goals were that the project had to fit into the existing budget, it had to help preserve the collection by reducing the handling of materials, the system had to be easy to use and display the images quickly, images had to be stored on the library's LAN and the project had to reduce staff time. Other requirements were that the project had to use an existing system for storage, it would be limited to use on the library's LAN (not available to dial-up users) and that it had to use IBM-compatible PCs. The Carnegie branch, where the images are stored and the originals are available, gets displays of images very quickly, but the other branches do not. Initially, the library replicated the image database at other branches to speed up the display, but had to stop because of the expense. The library uses library volunteers to do scanning. They use a flatbed scanner and Picture Publisher software, and store images on a local drive. Images are not enhanced at all. The images are uploaded to a server. Bibliographic records are created if they do not already exist, and the image's file name is inserted in this record. The average time for this process is 5 to 8 minutes, not including time for cataloging. Each item takes about 163,000 bytes of storage. Images are stored in compressed form, but the library is considering compressing them more in the future. Black and white images are stored at 32 levels of gray. The cost per image for storing is $0.24 per record. For display, Microsoft Windows is used with a VGA display at 640-by-480 pixels. Staff feel that they spend less time retrieving and filing originals. Since scanning started with the most popular items in the collection, this has greatly reduced the amount of time spent handling originals. Instead, they spend more time showing people how to use the system. Users without computer experience need a great deal of help at first, but they like the system. Print quality is poor and people want it to be better. Staff would like to have a better user interface with fewer keystrokes needed to display images, especially for multiple images attached to one bibliographic record. They would like to be able to scroll through the images in those cases. Limitations of the system are that dial-up users cannot see images, images are using a great deal of storage space, print quality is bad and some sort of graphical user interface (GUI) is needed. Display time is also a limitation. It takes three to five seconds to display an image at the Carnegie branch, but it takes 60 to 90 seconds at the other branches, even with a T1 line. The are problems with the Novell local area network; the T1 line is not the only problem. InfoStation at the National Agricultural Library Vinod Chachra (VTLS) spoke as a last-minute substitute. He demonstrated the VTLS InfoStation at the National Agriculture Library. Images are a component of the generalized multimedia system. The InfoStation includes an authoring system that allows the user to enhance and rotate images and create a personal study desk of images. There are hypermedia links to sounds and images from text. The InfoStation is designed to operate over the Internet and uses a subfield of the USMARC 856 field for the Internet address where an image is stored. The InfoStation uses the NeXTStep operating system. VTLS will supply workstations with 17-inch monitors for about $5,000. Chachra also touched on a project for Princeton University Libraries, where VTLS digitized images of 6.5 million catalog cards. Anne Flint is Assistant Director of Library Systems for OhioLINK.