Jump to:
Digitizing the data page images
The data pages in the Historic American Building Survey/Historic American Engineering Record/Historic American Landscapes Survey (HABS/HAER/HALS) collections were captured at the Library as digital facsimile images by Systems Integration Group of Lanham, Maryland, using a Minolta PS3000 scanner, a device with an overhead camera design. A series of tests established that the typescripts (from the early period) and computer printouts (from later years) would be fully legible if reproduced as bitonal images with a spatial resolution of 200 dpi and this specification has been applied to the scanning. At 200 dpi, a typical 8.5x11-inch page yields an image of 1700x2200 pixels. The resulting images are saved in the TIFF (Tagged Image File Format, version 6.0) format and compressed with the lossless ITU Group IV algorithm and a typical file is about 35 kilobytes. These are referred to as the archival or master images.
The Library's online presentation of the data pages employs a "page turner" or "electronic binder" that permits the user to navigate the data pages, turning to the next or a previous image, or selecting an image by number. The inline service images displayed in the browser are GIF images with a spatial resolution of about 495x640 pixels; a hyperlink on the image provides access to the master image. Using the master images as a source, the Library produces the inline service images in a batch mode that includes these steps:
- adding tonality (transform the bitonal image into a 16-color grayscale image),
- blurring the image to soften the edges of the characters;
- reducing the image from about 1700x2200 pixels to 495x640 pixels;
- applying an image enhancement algorithm to suppress the speckling that sometimes results from the addition of tones and reduction in size, and;
- sharpening the image to restore some clarity to the characters.
Some of the data pages consist of photo captions associated with particular HABS/HAER/HALS surveys. After capturing images of the caption pages, Systems Integration Group sent them to a subcontractor for rekeying. The rekeyed texts did not receive any coding or markup. After delivery to the Library, the rekeyed lists were formatted to create a small dataset for the survey to which they pertain. These datasets are the source for the captions displayed with the images for some surveys. (For other surveys, the HABS/HAER project recorded the captions on photo mount cards but did not include them with the data pages. The photo mount cards are not being scanned in the current round of work but the Library plans to rekey and present these captions in the future.)
Specifications for the data pages
- Archival or master images
- Spatial resolution: 200 dpi
- Tonal (pixel-depth) resolution: 1 bit-per-pixel
- File format: Tagged Image File Format (TIFF) ver. 6.0
- Compression: ITU Group IV
- Inline service images
- Spatial resolution: about 495x640 pixels
- Tonal (pixel-depth) resolution: 4 bits-per-pixel
- File format: GIF (Graphics Interchange Format)
Digitizing the photographs
The original negatives and color transparencies in the Historic American Building Survey/Historic American Engineering Record/Historic American Landscapes Survey (HABS/HAER/HALS) collections consist primarily of 4x5-, 5x7-, and 8x10-inch cut film exposed in view cameras. By scanning the original negatives and transparencies, higher quality digital images can be produced than would be obtained by scanning the collection's prints, which are one generation removed from the original exposure and, especially for the early years, have faded or deteriorated.
The HABS/HAER/HALS photographs are one of the first American Memory pictorial collections whose images have been produced by direct capture with a digital camera at high levels of spatial resolution. Most of the online collections of photographs and prints for which production began prior to 1998 were produced via photographic intermediates and, with the exception of the Panoramic Photographs, at lower levels of resolution.
The Library does not call the archival or master digital images of the HABS/HAER/HALS photographs "preservation copies," but these digital images will play the same role with respect to the originals as copies made on photographic film have in the past.
The digital images are being produced by JJT Incorporated of Austin, Texas, using their new overhead-capture MARC II digital camera. For the HABS/HAER/HALS photographs, the initial capture uses a 7000x7000 pixel matrix at 12 bits-per-pixel for grayscale or 36 bits for color. Image processing is then executed at 16-bits-per-pixel for grayscale or 48 bits for color. The bit depth is increased because having two even bytes per channel facilitates post-processing. For the archival or master image as saved, the spatial resolution is being reduced to 5000 pixels on the long side (with the short side scaled in proportion) and the tonal resolution reduced to 8 bits per pixel for black and white items and 24 bits for color. JJT's post- processing step also creates the three service images that complete the set of four delivered to the Library.
The digital images are intended to serve research, particularly research pertaining to historic buildings and engineering structures. Although aesthetic considerations played a role in determining the look of the digital images, it was also important to retain highlight and shadow detail so the maximum amount of information can be derived from the image. This means that the images are made at lower contrast than might otherwise be the case. Through the years, the contact prints served to researchers in the Prints and Photographs Reading Room were made in a similar manner, i.e., at a contrast level that reveals as much information as possible, and with little or no burning and dodging. Users who require digital images that are cropped, have higher contrast, or are improved by local adjustments, should download the archival versions and manipulate them in graphic arts software.
Specifications for the photographs
- Uncompressed Archival Images
- Spatial resolution: About 5,000 pixels on the long side with the short side falling where it may. This spatial resolution was employed regardless of the size of the original negative or transparency.
- Tonal (pixel-depth) resolution: Grayscale — 8 bits-per-pixel, color — 24 bits-per-pixel
- File format: Tagged Image File Format (TIFF) ver. 6.0
- Compression: none
- Compressed service images
- Spatial resolution: Medium resolution — 640 pixels on the long side with the short side falling where it may; higher resolution — 1024 pixels on the long side with the short side falling where it may.
- Tonal (pixel-depth) resolution: Grayscale — 8 bits-per-pixel, color — 24 bits-per-pixel
- File format: JFIF (JPEG File Interchange Format)
- Compression: Medium resolution — JPEG at a quality setting that yields an average compression of 15:1; higher resolution — JPEG at a quality setting that yields an average compression of 8:1
- Thumbnail images
- Spatial resolution: About 150 pixels on the long side with the short side falling where it may.
- Tonal (pixel-depth) resolution: 8 bits-per-pixel; palettes optimized (adaptive palettes) for each image.
- File format: Archived copy — Tagged Image File Format (TIFF); online copy —GIF (Graphics Interchange Format)
- Compression: Archived copy — uncompressed ; online copy — compression native to the GIF format
Digitizing the measured drawings
In 2000-2001, a contractor, JJT, Inc., of Austin, Texas, used an Océ 9800 scanner to scan the 55,000 original measured drawings that the HABS/HAER/HALS programs of the National Park Service had sent to the Library of Congress since 1933. In 2004-2005 the Library of Congress Digital Scan Center used an IDEAL/Contex Premier TX36 scanner to scan approximately 4,800 additional drawings that had been transmitted to the Library since 2000. Since 2004, the National Park Service has included digital images of the architectural drawings as part of their transmittals to the Library.
HABS/HAER drawings are either B (18x24-inch), D (24x36-inch) or E (34x44-inch) size.
The drawings were scanned at 400 dpi. Since the originals are line drawings, the images are bitonal and compressed with the lossless ITU Group IV algorithm. Typical compressed file sizes are about 700-800 kilobytes. Uncompressed file sizes range from 9 MB (B size) to 30 MB (E size).
After the master images were delivered, the Library produced three service images, one for the initial, "contact sheet" inline display in the browser, a second larger reference view, and the third compressed service image for printing on a desktop printer. The browser images were sized to fit most screen displays, while the printing image was sized to fit on a sheet of typing paper. The master image can be used to print the drawing at its original size. The browser images were produced in a process like that used to produce the "page turner" images of the data pages, described above:
- transforming to grayscale,
- blurring,
- reducing in size, and
- applying an image enhancement algorithm to suppress the speckling that sometimes results from the addition of tones and reduction in size, and
- sharpening.
The printing images were produced by reducing the spatial resolution of the archival images and recompressing with the ITU Group IV algorithm.
Specifications for the measured drawings
- Master images
- Spatial resolution: 400 dpi
- Tonal (pixel-depth) resolution: 1 bit-per-pixel
- File format: Tagged Image File Format (TIFF) ver. 6.0
- Compression: ITU Group IV
- Inline service images: Thumbnail and reference
- Spatial resolution: Thumbnail — about 220x175 pixels; Reference — about 640x480 pixels
- Tonal (pixel-depth) resolution: Thumbnail — 8 bits-per-pixel; Reference — 4 bits-per-pixel
- File format: GIF (Graphics Interchange Format) format
- Compressed service images
- Spatial resolution: 200 dpi; scaled to about 4800x3800 pixels
- Tonal (pixel-depth) resolution: 1 bit-per-pixel
- File format: Tagged Image File Format (TIFF) ver. 6.0
- Compression: ITU Group IV