summaryrefslogtreecommitdiff
path: root/info/digests/texline/no6/bcsep.tex
blob: ca0bf12405f0dd10078271a27ddffd78c3adb0e7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
% BCS Electronic Publishing specialist group conference, 23 Sep 1987
%
% review by David Osborne, Cripps Computing Centre, Univ of Nottingham
% [ JANET: cczdao@uk.ac.nott.vaxa ]		3-5 Oct, 1987
%
\def\quid{{\it\$}}  % credit to Malcolm Clark for this!
\centerline{\bf BCS Electronic Publishing conference}
\centerline{\bf 23rd September} 
\centerline{\bf Storage Technology in EP}
\medskip
\noindent
This was the first meeting of the BCS EP group I had attended,
though the advance programme described it as the fourth full day
meeting of what is now the third year of the group's existence.
The meeting was held at the School of Oriental and African Studies
in London, and was well-attended by representatives of both
commercial and academic organisations.

In the absence of the scheduled speaker from Hatfield Polytechnic,
Ken~Jamieson of Microrite (chairman of the EP group) gave the first
presentation on an Overview of Relevant Storage Technology, setting in
context the theme of the meeting.  He reviewed both the familiar and
unfamiliar media available for storage of text and fonts, leading on to
the new technologies of CD-ROM and WORM which were to be described more
fully later in the day.  His description of exchangeable 15.6Mb hard
disks for PCs enlightened many of those present who, like myself, had
not come across these devices.  Leading on from this, he contrasted the
amount of storage needed for an average A4 page of text, at about 0.5Mb,
with that of raster images which are increasingly demanded in documents
including both text and graphics; eight quarter-page graphic images to
be rendered at 300 dots-per-inch would require about 2Mb of storage. 

Andrew~James, Ventura Product Manager at Rank-Xerox, then led a
discussion on the storage problems experienced by those in the audience.
Andrew has written a review of the DTP (desk-top publishing) market
in the September 1987 issue of ``PC~User'' from his involvement with
Ventura, so his interests naturally centre on the special problems
of PC users.  As he pointed out, most PC publishing users have access
to greatly restricted resources compared with text-processing users
on minis or mainframes.  Shortage of disc space is a common difficulty,
particularly with the growing use of packages using bit-mapped font files
(in the pages of \TeX line, one such immediately springs to mind!)
On a slightly partisan note, he commented that some DTP packages
such as {\it PageMaker\/} exacerbate storage problems by using large,
monolithic document files, whereas others such as {\it Ventura\/}
minimise these by using smaller files collected together.

Having discussed the problems of storage the meeting then turned to
techniques for rapid capture of text in a pair of presentations on OCR
and desktop scanning.  Alan Howard of Kurzweil gave a fascinating
description of the Kurzweil ``Discover'' intelligent scanner.  Intended
for use with a PC, this uses a special processor board designed by
Kurzweil which embodies expert systems software to perform pattern
recognition on the scanned material, which may be text or graphics.  The
text-mode of scanning seems particularly sophisticated, being able to
ignore graphics and ``noise'' (such as handwritten comments, staple
marks and coffee stains), and to scan an A4 page in about 70 seconds
with a 95\% assurance level of character recognition.  The ``format
analysis'' performed by the processor even takes account of the page
layout and indent tabbing.  For output, the conversion phase can produce
a file in formats acceptable to most common word-processing packages as
a mixture of text and markup commands.  In its graphics mode the scanner
can perform contrast enhancement and can handle continuous-tone
photographs, with output in formats suitable for {\it Ventura\/}
and the {\it Xerox Documenter\/}.

It was perhaps unfortunate that a traffic incident delayed the arrival
of Sandra Woodley from Sintrom, since this led to her having to speak after
the Kurzweil demonstration --- a difficult `act' to follow.
She described the Datacopy OCR scanner, which uses a different, less
sophisticated recognition technique and is therefore restricted to
coping with a set of 19 fonts, up to 9 different fonts on one page. 
There is a ``learning'' mode in which the user can teach the scanner in
about 20 minutes to recognise a new font which can then be
added to the machine's library of pre-defined typefaces
(such as Courier 12 and Times Roman 10 \dots not a Computer Modern
to be seen!) At present, the scanner can only cope with text but will
handle both mono- and proportionally-spaced fonts, though kerning and
ligatures can cause some problems.  Again, the scanner is PC-based and
can generate output for a variety of word-processing programs.  Its
ability to scan bound volumes without them having to be unbound is
certain (I nearly said ``bound'') to make it popular with librarians. 

In comparison, the decision of which scanner to buy comes down to
``paying your money and taking your choice'', since the Datacopy costs
\quid 1658 for the flatbed scanner and \quid 742 for its software, while
the much more sophisticated Kurzweil is pitched at \quid 8000 for the
package of scanner, processor board and software. 

We then progressed into more high-tech with a presentation on WORM
(Write Once, Read Many) technology from Geoff Pullen of Microscribe.
Unlike CD-ROM, there is no current world standard on formats for
the optical discs used in WORM, and discs are manufactured for
specific drives.  The discs themselves are made of either glass
or optical polycarbonate and are now guaranteed by the manufacturers
for 30 years, though there was some discussion as to whether this
included integrity of the data recorded on them.
There are many types of drive available, and capacities of 1 gigabyte per
side have been common since 1983.  Three sizes are in use:
14-inch, with 6.8Gb capacity; 12-inch; and 5\frac1/4-inch, offering
115Mb--800Mb per disc.  ``Juke box'' storage for up to 200 12-inch discs
is available.
Image data from WORM discs can be retrieved in standard MS-DOS file formats,
allowing their exchange between media, once retrieved.
An entry-level 5\frac1/4-inch system should cost \quid 16-20000,
while top-level systems are much more expensive at \quid 84000 for
the Racal REOS, and \quid 100000 for the Philips Megadoc.
Despite these high prices, the utility of WORM for secure storage
of large volumes of text is clear.

Contrasted with WORM was the other optical medium, CD-ROM, described in
a talk by a speaker from Saztec Ltd., in the context of their contract
with the British Library for the conversion of the BL's General
Catalogue.  This is a huge task, since the catalogue details 4.2 million
volumes deposited with the library, and occupies 360 volumes which take
up 42 feet of shelf space.  These contain 176,000 pages, representing
about 1 billion characters.  The conversion phase of the contract is
costing about \quid 2M and will occupy about 70 people for 4
years.  The claim that this is an unparalled task was disputed by Lawrie
Newton of OUP, who stated that the computerisation of the Oxford English
Dictionary is of a comparable scale.  The opportunity is being taken of
restructuring the catalogue as part of the conversion to optical disc. 
The CD-ROM discs are similar to the audio discs familiar to
hi-fi enthusiasts and have an unformatted capacity of 600Mb, giving about
550Mb of storage when formatted and indexed.  This means that the
British Library catalogue can be held on merely four discs.  CD-ROM is a
suitable publishing medium for unchanging reference data, such as the BL
catalogue, and can hold both text and graphics, offering the ability for
rapid searching and the downloading of selected text for later
processing on a PC, for example.  The current high costs of mastering
and production mitigate against their use for rapidly changing data, or
where the production run is small, although the hardware to read them is
relatively cheap, at about \quid 200 per drive.  The data preparation for
CD-ROM costs about \quid 9000, with mastering an additional \quid 2000
and replication costs of \quid 12--\quid 18 per disc.  This puts the
cost of a production run of one disc at about \quid 12000, but \quid 25000
for a run of a thousand. 

Rounding off the meeting, Mike Daines of Signus discussed
``Soft Scanning Techniques for fonts'', with particular reference
to Signus' involvement with the {\it Ikarus\/} system.
Signus specialise in the production of digital typefaces for
various manufacturers, both for typesetters and low-resolution devices
such as laser and dot-matrix printers.  Mike noted that ``resolution
independence'' of typefaces can be achieved by storing the font
at a higher resolution than that used by any available device;
this is the method adopted by the {\it Lucida\/} family, by
Bigelow and Holmes (which they described at {\it EP86} in Nottingham).
It is also used by {\it Ikarus\/}, which has a resolution of
15000 dots in {\it x\/} and {\it y\/} for an em-square.
He then noted the advantages of hand digitisation, claiming
a that an effective time of 2.6 minutes per letter could be attained,
once the draft designs had been marked up.  Signus digitise to
an accuracy of \frac1/{100}{\it mm\/} for a letter 10{\it cm\/} high.
Almost all digital formats can be calculated from this
hand-digitised base.  Soft scanning consists of calculating the
intersections of the Ikarus `DI' format (which uses circles, arcs and
straight lines) with the raster pattern of the display device,
producing scan-lines for output.  This method allows later reprocessing
of the fonts to perform any corrections which may be necessary.
Summing up, Mike described what he felt were essential
requirements for digital typesetting:
fonts should have a square raster resolution
(equal in {\it x\/} and {\it y\/}),
a constant database of resolution-independent character definitions
and a common font format;
printers should have a minimum resolution of 600 dots-per-inch,
with fonts downloaded from their host machine rather than printer-resident.
\medskip\noindent
(The next meeting of the EP group is scheduled for November 26th,
to discuss DTP in the printing industry, with speakers including
Pierre Mackay and representatives from OUP and John Wiley).
\medskip
\rightline{\it David Osborne, Nottingham University}