Benchmark for digital reproductions of
monographs and serials
As endorsed by the DLF
Original prepared on July 30, 2001
Revised on January 25, 2002
Contents
- Introduction
- What is a digital master?
- Benchmark digital master for a digitally
reformatted monograph and serial
- Notes
1. Introduction
This document defines a minimum benchmark for digital
reproductions of monographs and serials. The case for such a
benchmark is made in an
article by Greenstein and George that is available RLG's
DigiNews.
Work on the benchmark grew out of the DLF's investigation into
the need for and functional specification of a service through
which libraries could register information about the monographs
and serials that they had digitally reformatted (see http://www.diglib.org/collections/reg/reg.htm).
Although the registry is not exclusive (it will record
information about digitally reformatted materials as well as
those that are born digital, and about masters that meet agreed
benchmarks as well as those that do not), it provides an
important opportunity to identify and build consensus around
minimum characteristics that might be expected of certain kinds
of digital objects.
The benchmark defined in this document refers to digital
objects that are created by reformatting printed monographs and
serials, and that are intended as faithful reproductions of the
underlying source documents.
Companion documents may be developed defining benchmarks for
other digital masters - for example, those that may apply to
serial publications or books that are born digital.
The benchmark has been prepared and endorsed by the DLF (a
report on that work is
available from the DLF's website).
2. What is a digital
master?
Digital masters are digital object that are optimally
formatted and described with a view to their quality
(functionality and use value), persistence (long-term access),
and interoperability (e.g. across platforms and software
environments).
This document benchmarks digital masters for digitally
reformatted monographs and serials. This type of digital master
has all of the qualities described above. In addition, it
is intended as a faithful rendering of the underlying source
document, for example, with respect to its completeness, the
quality of the image (including its tonality and color), and the
ability to reproduce pages in their correct (that is their
original) sequence. As a faithful rendering, a digital master
will support production of a printed page facsimile that is a
legible facsimile when produced in the same size as the original
(that is, 1:1).
The benchmark acknowledges that what ultimately constitutes
legibility and fidelity is a subjective decision. In part for
this reason, the benchmark refers minimally to file formats,
compression, and metadata. It does not provide production-level
guidance, for example on how to deal with missing pages, to
"clean up" foxing or blemishes on the digital master, or to
select an appropriate dpi for fonts or source pages of different
size. Such guidance, it is hoped, will evolve through experience
and may be attached as companion documentation to this
benchmark.
3. Benchmark digital master
for a digitally reformatted monograph and serial
A digital master for a reformatted monograph or serial must
include digital page images.
At least one version of the digital master's page images will
have or exceed the following minimum level characteristics.
Black and white (may include simple line drawings,
de-screened halftones) |
Grayscale |
Color |
600 dpi, 1-bit or bitonal TIFF images[1].
Images must be sized and saved at 1:1 scale to the dimensions
of the original page.
Images must be saved uncompressed or with lossless compression
(e.g. ITU-T6, LZW, CPC). Where images are compressed they must be
made available in the Group-4 format. The images may be dithered
up from 400 optical dpi 1-bit images
|
300 dpi, 8-bit grayscale uncompressed TIFF, or lossless
compressed image (e.g. JPEG2000).
Images must be sized and saved at 1:1 scale to the dimensions
of the original page.
The dpi specification will relate directly to the font-size
and page dimensions of the original source document, and to local
definitions of legibility and fidelity. In many cases, 400 dpi
will be preferred. Where larger pages are concerned (for example,
those exceeding 7 inches in the long dimension), the lower dpi
specification may be required).
|
300 dpi, 24-bit color uncompressed TIFF, or lossless
compressed images (e.g. JPEG2000).
Allowed color spaces include RGB, sRGB, PhotoYCC, YCC, CIELab,
and CMYK, with RGB and YCC being recommended as preferred for
digital masters. Images must be sized and saved at 1:1 scale to
the dimensions of the original page.
The dpi specification will relate directly to the font-size
and page dimensions of the original source document, and to local
definitions of legibility and fidelity. It may also relate to the
perceived artifactual value of the source object or the extent to
which its physical characteristics such as foxing, etc., are
perceived of as conveying some important information or
meaning.
|
Digital masters for digitally reformatted monographs and
serials must have descriptive, structural and administrative
metadata, and the metadata must be made available in
well-documented formats.
Structural metadata should be sufficiently detailed to allow
reconstruction of the sequence of the original artifact. A
minimum list of structural metadata elements is being developed
as a companion to this benchmark.
Preservation digital masters may include machine-readable
(keyboard or OCR) text. That text may be corrected or
uncorrected. If it is corrected, the accuracy level will be
specified (e.g. as 99.995%). Such text may be encoded (at any
level, e.g. as specified in TEI Text Encoding in
Libraries. Guidelines for Best Encoding Practices. Version 1.0,
July 30, 1999)
4. Notes
1. 600 dpi will capture roman scripts
down to 6-point type with the microfilm Q1 equivalent of 8.
Smaller text, scripts with fine lines and small dots and other
diacritics (like italics, Arabic, etc.) need higher resolution to
be captured completely.
return to top >>
|