Standards and Technical Specifications > Digital Image Standards

 WDL Digital Image Standards

Digital image files of original items that have been selected for digitization and/or inclusion in the WDL should be provided in an uncompressed TIFF format. 

File Formats for Delivery of Digitally Converted Materials to the World Digital Library

Type of original - library item	Type of reproduction	Format	Resolution or other notes
Textual: printed books, pamphlets, broadsides, ephemera, manuscripts (text and text with grayscale illustrations)	Grayscale	TIFF	300 DPI; crop to .6 to .75 cm beyond text block (preferred) or .6 to .75 cm beyond page-edge, if has artifactual value; 8-bit grayscale,
Textual: printed books, pamphlets, broadsides, ephemera (text with color illustrations)	Color	TIFF	300 DPI; crop to .6 to .75 cm beyond text block (preferred) or .6 to .75 cm beyond page-edge, if has artifactual value; 24-bit color RGB
Textual: newspapers, periodical pages	Grayscale or color (depending on the original)	TIFF	300 DPI; crop just beyond  the text block; 8-bit grayscale or 24-bit color RGB
Pictorial: photographs, lithographs, daguerreotypes, transparencies, photographs	Grayscale or color (depending on original)	TIFF	300 DPI; crop to .6 to .75 cm beyond image edge; 8-bit grayscale, 24-bit color RGB; if smaller than 10 by 12.5 cm, 600 DPI.
Cartographic: maps, atlases	Color	TIFF	300 DPI; crop to .6 to .75 cm beyond image edge; 24-bit color RGB

Use of 300 DPI assumes the goal is to provide basic access to the intellectual content of the materials.  Subject to further discussion, it may be appropriate to use 400 or 600 DPI for certain materials in order to support successful optical character recognition (OCR) or enhanced access to artifactual features of the original items.  In the event that a requirement stated above exceeds the capacity of the equipment provided, the files should be the maximum possible with the equipment.

Image Files to be Provided

Please supply all images in the TIFF 6 format, uncompressed. 

• Pictorial materials (photographs, prints, etc.)
• Photographs or prints larger than 10 x 12.5 centimeters may be scanned at 300 DPI; smaller materials should be scanned at a higher DPI.
• Bitonal or grayscale images will be scanned at a bit depth of 8 or 16-bit grayscale
• Color images will be scanned at a bit depth of 24 color RGB
• Maps and atlases
• Maps and atlases shall be scanned at a capture resolution of 300 PPI, 24-bit RGB color.  Maps and atlases should not be sized to a particular height or width but delivered at the capture resolution.
• For all image files
• No sharpening or enhancement, other than correction of the color.
• Image files shall have minimum post-production processing.
• “Intel” TIFF file format.
• Original capture should be: color in 36- bits or more per pixel; black and white (grayscale) at 12 or more bits per pixel.
• Administrative data to be provided
• For all image files, it is useful to provide a report of capture device used and standard settings for that device.  This information should be supplied as part of the larger XML schema.
• TIFF files are able to carry administrative data in the TIFF header.  Files to be provided for the project should contain the TIFF header values listed here: http://memory.loc.gov/ammem/about/standardsTable4.pdf.

Basic Criteria for Assessing the Suitability of Digital Images

LC will advise the contributor on suitability of digital images created for the project.  If the contributor supplies digital image files that are not considered to be suitable for posting on the Web Site as determined by LC specialists, LC will suggest ways that the files can be changed.  The contributor will then use reasonable efforts to make the identified changes and re-deliver the digital image files to LC.

Cropping

Whenever possible, the entire original sheet or page should be presented. In no event should the image of the actual document be cropped. (Researchers using digital documents often wish to be reassured that the entire document has been captured. This is especially desirable for unbound manuscript documents. A "border zone" between .6 and .75 cm beyond the edge of a scanned document is desired whenever possible.  For newspapers and periodical pages, a much narrower “border zone” is desirable, just outside the edge of the page.)

Skewing

Images from unbound materials should not be skewed.  (For bound materials, the tightness of the bindings may result in slightly skewed images.  Ideally, such images should still be skewed no more than 5 degrees, but in some cases, even this limit is not attainable. In these cases a note in the scanning metadata should be made that the image was scanned using best efforts, and note the reason for and extent of the skewing.)

Bitonal Images: General Guidelines

• Images should be neither too dark nor too light. (Indicators of failure in this area are such things as lost lines in type, blocked-up parts of letters, and washed out areas. These qualities of the images sometimes can be evaluated by adjusting the image in the software to exaggerate the effect and thus confirm brightness or darkness; this can help in seeing the relative values. If in doubt, it is useful to examine at 1:1. A printing crosscheck also may help diagnosis.)
• Images should be clear, sharp, and free of blur. (The best checking includes 1:1 display. A printing crosscheck may help diagnosis.)
• Images should be free of excessive dots, spots, and similar imperfections.
• Images should be cropped and segmented according to requirements, and the segments named correctly.
• Images should have the proper rotation. Generally speaking, the top of the text or illustration should be the top of the image.

The following guidelines should be useful for testing print-ability (assumes HP LaserJet):

1. The printer driver resolution setting should be checked and set at the highest possible setting (600DPI, if available).
2. If five or more images of a batch have already been printed for crosschecking and have printed satisfactorily, no more need be printed. 
3. It is necessary to print both a full image scaled to page and a cropped part (if needed) that will print at 1:1.

Grayscale and Color Images

Images should be neither too dark nor too light and have the correct distribution of tones. (Indicators of failure are such things as very dark [black] areas, and washed out highlights.)

(It may be useful to look at the histogram to be sure the full range of values is present [for grayscale: 0 through 255].  For a printed document, there ought to be hump at the light end representing the paper and a hump at the dark end representing the strokes. The left side represents dark values; the right side represents light values. Some checks can be done in advance of thumbnail viewing. Adobe Photoshop software includes a sophisticated histogram that presents numerical information about the range of tones. It may also be useful to adjust contrast, brightness, or gamma to get a sense of what the image might look like if adjusted. Both ThumbsPlus and Photoshop include contrast, brightness, and gamma adjustments.)

The color of the images should be accurate. (This is a very difficult question, prone to subjective judgments. It may be useful to have the original item at hand to determine whether it looks accurate when compared to the original.)

• Images should be clear, sharp, and free of blur. (The best checking includes 1:1 display. A printing crosscheck may help diagnosis.)
• Images should be free of "excessive" dots, spots, and so forth.
• Images should be cropped and segmented according to requirements, and the segments named correctly.
• Images should have the proper rotation. Generally speaking, the top of the text or illustration should be the top of the image.
• The same guidelines for test print-ability as those listed above for bitonal images may be used.

Resolution and color bit depth figures are minimum values.  DPI figures are one-to-one relative to the original, allowing printing at the same value (PPI).

 Top