General comments on digital reproductions of pictorial items for American Memory


Applicants for DLI-Phase II who would like to make use of still visual materials converted by the National Digital Library Program (NDLP) for American Memory should be aware of the heterogeneity in digital format.

American Memory includes different genres of original pictorial material, primarily photographs, but also including prints, architectural drawings, and pictorial maps. They have been converted at various times over the last 6 years, during a period when scanning and delivery technology advanced rapidly. Most conversion for the materials listed has involved photographic intermediates, but direct digital capture is being used for some materials now being scanned. The approach used for capture has sometimes been controlled by the size or condition of the original materials or the prior existence of surrogates such as copy negatives or digital intermediates used for preparation of a videodisc. The Prints and Photographs Division of the Library of Congress first used digitization to enhance general reference access to the materials in its unequalled collections, which include over 15 million items. For this purpose, images of moderate resolution can act as surrogates; users can order photographic reproductions of materials identified online. As appropriate technology has developed and costs for digital storage have fallen, the Library has started making higher-resolution (museum-quality) digital reproductions of pictorial materials. The NDLP believes that digital libraries must support effective access for a variety of purposes and tasks users will have for pictorial materials, allowing for the variety in digital format and resolution that will inevitably be present in digital collections built cumulatively over the years.

Image formats

In most cases, a photograph has been captured and stored in an uncompressed TIFF format (using 8 bits for grayscale and 24 bits for color). This image is considered the archival version of the digital reproduction (not to be confused with a preservation-quality reproduction of the original). From this image, an access-quality service image (in JPEG format) and thumbnails have typically been derived. Thumbnails for use as inline images in web-presentations are in the GIF format. Many thumbnails are also available as uncompressed TIFFs. However, the Library is aware that some graphics software can not open the TIFF thumbnails. In some instances, the best image is not an uncompressed TIFF, but a compressed JPEG. As the Library increasingly captures images at higher resolution, intermediate serviceimages of differing spatial and tonal resolutions may be created for future collections.

Descriptive records

For most of the pictorial collections listed, item-level MARC records are available. In an 856 field, the combination of subfields $d and $f provides a unique identifier for the associated digital reproduction. Files representing the images are available in a directory structure (known at the Library of Congress as an "aggregate") for which $d identifies the root. Depending on the size of the collection, the directory structure may have 1, 2, or 3 intermediate levels of directory, using a pattern that can be derived automatically from the $f value. Filenames for the different digital versions of an image are created by combining the $f value (which identifies the item) with distinguishing final letters and file extensions.

In some cases, the bibliographic record has links (in additional 856 fields) to alternative digital reproductions that are not currently made available through American Memory. The most important instance is in America's First Look into the Camera, a collection of daguerreotypes from the mid- 19th century. For several items in this collection, two or more digital reproductions have been made, using different photographic intermediates, often one color and one black-and-white.

Collections with pictorial items available for use in DLI - Phase II

The American Memory collections listed below, currently released or in an advanced state of production, include photographs and other pictorial images. Technical summaries with considerable detail and links to samples are available for each collection by clicking on the title.

Please notice that a separate technical note describes the map collections, which include large high-resolution images appropriate for testing compression schemes and approaches for presentation to users with display and navigation options that are effective over networks.

Collection title:     Click for technical summary


America's First Look into the Camera: Daguerreotype Portraits and Views, 1839-1862

599 items, image quality good

Architecture and Interior Design for 20th Century America: Photographs by Samuel Gottscho and William Schleisner, 1935-1955

ca. 29,000 items, image quality moderate

Around the World in the 1890s: Photographs from the World's Transportation Commission, 1894-1896

880 items, image quality moderate

Black & White Photographs from the Farm Security Administration and the Office of War Information, ca. 1938-1944

ca. 170,000 items when complete, image quality moderate, not all individually described

Built in America: Historical American Buildings Survey & Historical American Engineering record, 1933-Present

architectural drawings

By Popular Demand: Portraits of the Presidents and First Ladies, 1789-Present

156 items, image quality good

By Popular Demand: Votes for Women Suffrage Pictures, 1850-1920

38 items, image quality good

Color Photographs from the Farm Security Administration and the Office of War Information, ca. 1938-1944

ca. 1600 items, image quality moderate

Creative Americans: Portraits by Carl Van Vechten, 1932-1964

ca. 1400 items, image quality moderate

The Evolution of the Conservation Movement, 1850-1920

ca. 170 items, image quality moderate

Selected Civil War Photographs, 1861-1865

1,118 items, image quality good

Taking the Long View: Panoramic Photographs, ca. 1851-1991

4,200 items, high-resolution images

Touring Turn-of-the-Century America: Photographs from the Detroit Publishing Company, 1880-1920

ca. 25,000 items, image quality moderate

Washington as It Was: Photographs by Theodor Horydczak, 1923-1959

14,350 items, image quality moderate

Challenges faced by NDLP

Providing access to large collections of photographs, such as the working archive of a photographer, has always posed challenges. The information that helps users identify the pictures they want (what, where, when, who) must be represented in an associated textual description. However, providing detailed descriptions for each item is inordinately expensive. In many cases, it can be argued that individual descriptive records are of questionable value (for example, when no information is available to identify the location or the event or people pictured or when it is hard to distinguish among a group of very similar pictures).

Although a cataloger in the Prints and Photographs division of the Library of Congress may not be able to identify individuals or precise locations in photographs in its collection, there are often remote users who can. Such users may range from scholars familiar with an individual, event, or period to family members of an individual in a portrait or schoolchildren living in the neighborhood of an architectural structure or landscape. What workable systems can be developed to allow users to suggest descriptive metadata that might enhance catalog records. Any such system must incorporate appropriate safeguards against malicious suggestions and ensure that controlled vocabulary and name authorities are used when appropriate, without placing undue burden on cataloging staff.

Physical archives of pictorial materials often make use of grouping containers, storing many related items in a single folder or box. Items not meriting individual description can be found through a hierarchical categorization or group-level descriptions. Users can browse through the containers or through a document listing the containers with brief descriptions of each container's contents. Such a document is often called a finding aid or an archival register. The physical organization or a corresponding structured (but linear) listing, provides effective access if the physical organization matches the users' conceptual view of the collection or particular information need. To provide additional access points, supplementary indexes to large physical collections may be prepared, perhaps by name, place, or subject, as appropriate for the particular collection. In a digital library, there is the potential for much more flexible access, but the advantages of traditional access aids should not be lost. How can we build systems and interfaces take full advantage of whatever information is available -- combining visual browsing through structured organizations with searching descriptions that may apply to an individual item or at different granularities of grouping? How do we balance the power and flexibility for retrieval with performance and avoid confusion for users?

Similarly, how do we best provide access to alternate formats and resolutions for expert or specialist users, without making the interface awkward or confusing for novice, occasional, or casual users?

[Digital Library Initiatives] [LC NDLP]
Challenges] [Collections] [Site Map]

[List] [Rights] [Notes] [Samples] [Handover]

[Documents] [Images] [Maps] [Sound Recordings] [Motion Pictures]