Desktop publishing: a guide for skeptics

Desktop publishing: A guide for skeptics You already hate what it's called, and you suspect you may hate what it its, but desktop publishing (DTP) is now such a broad category of products and principles that it's worth examining the subject to see what it offers the professional magazine publisher. Before you decide that DTP is just a tool for aimless design noodling by people who are supposed to be getting newsletter out to their customers, note that the major concept here is decentralized typesetting. At its heart, DTP puts the calculating tasks for composition in a microcomputer, and long before it dresses up this facility with icons--and, in some cases, eliminates some rigourous typesetting controls--DTP software makes your computer a legitimate front end for a typesetting system. From this, and its corollaries, publishers can benefit a great deal.

This is an article for skeptics. It will take you through some of the conventional objections to DTP, and acquant you with the components of a professional production system that utilizes current products. The discussion will help you focus your purchasing decision on functional aspects so that you can better imagine accomplishing your work on a DTP sytem.

For the sake of simplicity, I'll draw product examples only from PC-class applications, since many publishers with an investment in computers own MS-DOS machines from IBM or the legion of the compatible manufacturers. DTP was born on the Apple Macintosh, and you will hear valid arguments claiming that Macs are still the best hardware platform. And clearly, your first buying decision must address the PC versus Mac choice. I'm not going to develop all the nuances of that comparison here, but I'll mention that the three truly high-end composition software packages you may want to consider are all PC-based.

Putting DTP in context

Slowly but surely, the computer assumed its rightful place in a typesetting system. About 15 years ago, it was particularly chic to refer to equipment as a "computerized typesetter" because there actually ws the possibility that it was otherwise. But enough nostalgia; computers are useful for many things, but they seem absolutely required for the grueling task of counting character widths, inserting hyphenation points, and justifying columns.

The front end/back end distinction began to creep into typesetting equipment descriptions years ago, but the front and the back were tightly married, communicating with proprietary software and even proprietary font widths. The first hint of decentralized typesetting was the suggestion that we buy off-line terminals, some dumb, some counting. The photo unit was seen as but one piece in a typesetting system, and the terminal an independent element.

Before you worry that DTP is too good to be true, note that it boils down to random microcomputers serving as front ends in a world of shared font and counting software. This is possible because we've been freed from compliance with proprietary H & J software and unique photo unit drive instructions.

Type width information can be shared because type software can be stored digitally and scaled to size by a device-independent page description language (PDL). The primary PDL in the marketplace is Adobe's PostScript, although other PDLs are poised to compete with it. And characters are created not by a motor that whirls a film master into place and causes a lamp to flash and expose the typesetting paper, but are stored as bit maps or vectors and exposed by lasers or CRT tubes, keeping the type digital until the last possible moment.

Setting the record straight

If DTP earned its questionable reputation for any good reason, it's that monkeying around with letters was made to appear to be a substitute for setting type. Frankly, if you gave people with no training and lots of free time a Comp Edit, they'd set bad type too, but they'd also get tired quickly. DTP software, like PageMaker and Ventura, doesn't exhaust them as rapidly, not least because the screen itself shows the end results, treating the user to interactive WYSIWYG (what-you-see-is-what-you-get) displays. And so we're seeing disturbing things lately-like captions set in Mistral and six-point rules roaming like thugs all over the page. It should be clear that the tool isn't at fault here. In the long run, these design faux pas will slowly become less common. Isn't it likely that there were a good number of bad drivers when the Model T first appeard? Which isn't to say that design skills come as easily as driving, but visual literacy can be expected to increase. One consequence will be the return of the tools to those best equipped to use them, and a better appreciation of their work.

As a publisher with a trained art staff, you're not worried about this problem. But you may wonder if you can obtain adequate typographic quality from a PC-based composition system. The first big knock against DTP output was the low resolution of the office laser printers used to make plain paper proofs. At 300 dots per inch (dpi), these printers certainly don't yield printable, magazine-quality type.

But the laser printers themselves have nothing to do with it, for the simple and heartening reason that the type is dwelling in a digital world, ready to be scaled up at any resolution. With PostScript in charge of the printer's imaging system, the restraint on quality is not the original type data but the printing device itself. The two PostScript-driven Linotronic typesetters, at 1270 and 2540 dpi, produce very real type.

The final brickbat tossed at DTP is a quality question that is a little more complicated. Sure we can get high resolution output so the type itself looks sharp, but what about the quality of the composition program itself? Does DTP offer nothing but loose lines, juvenile hyphenation programs, and puny controls over spaceband settings?

Not necessarily. Although many DTP softwware packages can take on their established rivals feature for feature, there are also some pretenders to the throne mixed in--and this tends to distort the potential. But there are several programs that a pass a typographic sophistication test on every point but providing hung punctuation. The exam includes automatic and manual kerning, min/max/optimum spaceband setting, sophisticated hyphenation by dictionary or algorithm, extreme positional control of baselines, and many other fine points. I expect hung punctuation within the hour: There are no real ceilings on this, as the recent appearance of kerning as a standard DTP feature has proven.

DTP is a kingdom with fuzzy boundaries, and the publisher wishing to make a serious investigation of it will quickly collide with products that can't be taken seriously for production work. The situation is exacerbated by vendors slapping the term "desktop publishing" onto anything from word processing packages to solar calculators. (You can copy fit with 'em!)

I'm writing this with the aid of WordPerfect, and if I want to bend the term a bit, I've got a desktop publishing system. WordPerfect, for example, lets me define multiple columns and place footnotes with conditional instructions about their relationship to the main text. Because WordPerfect has drivers for laser printers like Apple's Laserwriter, I can print my text in Times Roman as easily as in Courier. Am I publishing yet?

It's my hope that most of the world still believes I'm not. The ability to summon typefaces at the output stage should not be confused with typesetting. And it's typesetting that seems to be the watershed between powerful text control and dTP. Furter, it's typesetting that all the casual desktop aficionados, in their hearts, really lust after--for newsletters or price lists or self-published novels. How can we break the news to them that typesetting isn't substituting an Avant-Garde "g" for a Prestige Elite one? To make matters more confusing, when the typeface is "in" the output device and something like Helvetica comes knocking on your door the minute you plug in a Laserjet, it's easy to assume that typesetting resides in a machine. As a magazine publisher, you can easily recognize the distinction, but you are doomed to find yourself reading about products and programs that blur it all over again.

Examples of PC software that essentially control the appearance of printed documents are Fancy Font, LaserWare and Printix. These packages are valuable tools for some office applications, but they meet virtually none of the criteria for composition, and there are no magazine production applications for such products.

At the other extreme are application packages that not only turn your PC into a typesetter's front end, but remind you all the time that they're doing so. These are the code-based DTP products, such as DeskSet, Magna Type and Superpage. They're different from mainstream DTP software on three counts: They cost noticeably more, they blur text entry and page formatting a bit by asking editors to code text (although they don't actually require it), and they drive not only PostScript devices, but conventional phototypesetters--from Compugraphic, Varityper, Linotype and the like. Indeed, they're also different in that they provide true typographic sophistication, but when all is said and done, some of the popular DTP packages give them a run for their money on basic typesetting duties.

Perched in the middle are the products you hear the most about: PageMaker and Ventura for the PC; and PageMaker, Ready Set Go, Mac-publisher, Quark Xpress, Scoop and others for the Macintosh. The jump between PageMaker and MagnaType is financially step, and the marketplace seems particularly ripe for products with high-end functionality at a lower price.

Software and hardware

There's a tendency to define a DTP application through a single software package, like PageMaker or Ventura. These are indeed the heart of a DTP system, but in planning a system for magazine production, you'll want to integrate several other items.

There are six aspects of the page creation process that electronic systems now address, and despite the impulse to obtain and integrate all six, publishers are wise to consider these applications as discrete. Although turnkey systems are now available, assuring that hardware and software are properly matched, the potential user need not assume that combinations are excessively difficult to construct independently. The components of a DTP system can be clarified as facilities for text editing, text entry, graphics input and editing, composition and page make-up, screen display of page make-up, and output.

* Text editing: For DTP, text editing includes both the usual chores of copy editing, fact checking and editing for style, and the new job of preparing files for page processing and composition. The file management tasks for users of a DTP system can be a bit daunting, and purchasers are advised to contemplate their operational approach in detail. In general, you'll want to develop a system for identifying the constituent parts of an article and acquainting the page layout artist with the relationship of text call outs, illustrations and sidebars. Ventura, for example, offers the capability of linking an illustration with a section of text, so that repagination takes into account context relationships. This is a very valuable feature, but your file preparation system must give the designer instructions on what to link to what. When all is said and done, plain old pieces of paper are very necessary: At minimum, editors must identify the elements of an article, perhaps down to the format level, and may even want to rough out the relation of illustrations to text in a thumbnail. So much for the all-electronic office.

The choice of a word processing package can be made with some independence from the DTP composition package. Editors throughout the magazine can use different packages, feeding manuscript through ASCII or, if available, through the DTP software's wp conversion facility. The presence of extensive formatting controls in the word processing package itself is a moot point.

* Text entry: Manuscripts can be created and edited on word processing software, or retyped on the computer. If the manuscript itself can originate electronically, most word processing software offers authors all they need. Capabilities like pseudo-justification and multiple columns are actually meaningless here, since these operations must be turned over to the composition software that works on the text at a later stage. Although DTP packages all have some limits on the word processing formats they'll import directly, all of them are happy with plain old ASCII. So any word processor that offers ASCII conversion will do the job of bringin text over.

A modem an conversion software are the second part of text entry. These two text capture products are appropriate for publishers who wish to receive manuscripts in electronic form. A modem allows a properly equipped contributor to file his story over the phone. To steer clear of pesky communications software and telephones charges, he could just as well ship his story on a floppy via overnight courier. But what if he writes in Multimate and you edit in WordPerfect?

Enter text conversion software. Although you can always fall back on ASCII as an intermediate file structure, you may want to preserve the spacing and attributes of the manuscript that ASCII can't translate. Several software products meet your needs. Software Bridge, XWord, Word for Word, PC-Switch, and R-Doc/X are examples of programs that convert the unique formats of a range of word processing applications. Because each of these treats a specific set of word processing formats, your buying decision may be based primarily on the availability of the particular conversion you need to make.

The third aspect of text entry is the scanner. Chances are, you'll be dealing with some manuscripts that aren't electronic. You have the option of using a scanner for Optical Character Recognition (OCR) to convert marks on paper into ASCII blips. The scanner scene is heating up; you'll have several to choose from, most of which combine graphics scanning with OCr. Plan on spending $2,000 to $3,000 for a scanner, and plan on a bit of disappointtment, too: OCR is easily foiled by such villains as photocopes, alien typewriters, and dot matrix print. Your text file will certainly include some misreads, and it can be efficient to pass the file through a spelling checker to locate errors. Even after that, you'll still need to proofread, so don't think of scanning as the perfect text capture system.

Do you like tough choices? Yes, most scanners now offer both a graphics and OCR mode, but these applications are a bit more at odds with each other operationally for a single device to serve both properly. It comes down to what you'll scan: If OCR is your primary need, you'll want a scanner with a sheet-feeding bin that can gobble up manuscripts unattended. If graphics scanning is mot important, you may well want a flatbed scanner with a hinged top to accommodate bound originals or odd-size items that can't travel through a feeder.

* Graphics input and editing: The scanner is also the tool for hauling graphic images into your system electronically. With graphics paint software, you can create images or edit those retrieved by a scanner or imported from another graphics file format, such as the PIC chart files from Lotus 1-2-3.

Principally, you'll want graphics to appear in your page make-up package either as finished line art that a PDL like PostScript can print, or as the electronic equivalent of FPO stats, showing a halftone image for design and cropping purposes, but not as reproducible art. Yes, you can feed a scanner a continuous tone image and ask it to render the tonal information as gray scaled pixels (called dithering), but no, the halftone process as we know it has not come to inexpensive digital equipment. The scanner can resolve tonal information as pixels, but the 300 dpi limit restricts the outcome from rivaling true halftones. These scanned halftones can be placed during page make-up, but they're of interest chiefly as an aid to designing a page.

It's in the graphics area where PostScript does the most to muscle aside conventional typessetter output. The PostScript PDL is capable of taking graphc information, from line to screen values to page placement, and printing this as easily as it renders a typeface. For PostScript, it's all the same sort of data. If you hanker after DTP chiefly to incorporate line art on your pages, you'll want to consider output to PostScript-equipped laser printers and the PostScript-driven Linotronic typesetters. If fine typography or the need to utilize a conventional type-setter has nudged you toward MagnaType, Superpage or DeskSet, you should be aware that you can't call upon a Compugraphic to draw pictures for you. The high-end packages, however, are happy to drive PostScript equipment, and accept graphics files for page make-up.

Well--accept some of them, that is. Each software package has some restrictions on the graphic file types it can read and scale on the pagE. PageMaker, for example, is the most broad minded, accepting many file types and also providing the intellingence to scale the images for optimum printer output. Plan to spend some time sleuthing the compatible file structures to connect your graphics pakage, scanned images and DTP software.

* Composition: The distinction between front ends and back ends that came to conventional typesetting with the arrival of off-line terminals freed us to think of composition as a set of calculations completely separate from the process of making an image of the type. Off-line terminals could be either counting or noncounting, intelligent or dumb. In DTP, the microcomputer will indeed be intelligent and will perform all the labors necessary to position and size the type. If the system uses a PostScript device for output, the composed file carries a long set of commands in PostScript language. If the software can drive a conventional typesetter, as DeskSet, MagnaType and Superpage can, the instructions direct the imaging process just as input from that machine's own terminal would. The typesetter, as a DTP output device, becomes the dumb element in the chain, because all the hyphenation and justification decisions are made on the file before it reaches the typesetting machine.

When sorting out various DTP packages, professional publishers are seeking products that provide the level of typographic nuance and quality they require. Features like kerning and tracking, now widely available, tend to be required. The ability to define minimum, maximum and optimum word spacing is essential for fine typography. Further, this definition should be made in a relative size, as a percentage of a font's set size or as increments of an em.

A DTP package's hyphenation system is of major significance. Dictionary-based hyphenation is accurate but slow; algorithm-based hyphenation can be speedy but may err on the side of either missed opportunities or bad breaks. For example, an algorithm that prevents "batt-ing" might also disallow "bunt-ing" if it analyzes simply the "t" to "ing" relationship. The most sophisticated hyphenation programs dwell in MagnaType, which sports five levels of H & J, plus foreign language hyphenation. Superpage II offers a 93,000-word dictionary plus optional algorithms; PageMaker provides a 110,000-word dictionary; and Ventura gives you a choice of two algorithms (one fast and one exhaustive) and foreign language algorithms as well. DeskSet uses hyphenation algorithms plus exception dictionaries. Publishers will also want to evaluate the H & J program's control over successive hyphenated lines, and the minimum characters before and after a break.

* Page make-up: It takes an act of will to divorce page make-up from composition when examining the interactive WYSIWYG packages like PageMaker and Ventura. Ventura, for example, composes a text file the moment you import it by reviewing the text for all potential hyphenation points and bringing it in with, at minimum, a default text type style. Then, with the text on the screen, the user proceeds to move and resize it, finishing the design of the page. It's page make-up the user sees and works on; composition take place invisibly.

We can roughly classify all DTP packages by three types, which treat the product's page make-up paradigm. The first, exemplified by PageMaker, is page-oriented WYSIWYG. The central conceit here is that the user will position all text and art, piece by piece, gaining support for his decisions by a screen display that shows size and position relationships fully. This approach parallels traditional paste-up very directly, and assumes you want the computer merely to help you with what you've always done before. It's simple to learn, because you don't go about your work any differently. And it's either thoughtful or witless, depending on your point of view, because it waits for you to place each and every element. You'll perform many repetitive tasks with a page-oriented program, but you'll always have rather natural control over the results.

The second model, epitomized by Ventura, might be called document-oriented WYSIWYG. The two main principles here are that type specifications should work in sweeping, global ways and are best defined as discrete style tags, and that your decisions about page make-up aren't particularly localized. The style tag concept allows editors or page layout artists to identify elements of copy in particular categories. The software maintains a roster of type specs matching that category's tag, and applies it whenever the tag name is encountered. An editor doesn't need to fill in a long line of type parameter code, but can simply call out the tag name, in brackets. Further, the tags can not only be redefined during page make-up, but can be assigned for the first time there, allowing the layout artist to worry about such details as flush left captions in nine point. Finally, the document model assumes that you're using the computer to keep track of a welter of details for you, such as footnote numbering, and that it will paginate the document based on your specs, without any assistance from you. After you add art and change text positions, it will obediently repaginate everything. The assumption here is that the user can define typographic appearance, at least initially, without regard to page placement: The choice of type and leading will have everything to do with the classification of text elements, and nothing to do with filling a particular place on the page.

The third paradigm, represented by DeskSet, MagnaType and Superpage, is the direct evolution of conventional typesetting, and could be called code-based batch processing. No matter how long you try to keep it from the user, typesetting requires codes, and these applications are honest about it: An editor or his trusty page make-up assistant has to type out instructions on font size, style, leading and so forth. To simplify this routin, Magna and Superpage allow you to designate a slew of commands under a single format code, which can be named logically. Computer users will recognize this instantly as a macro capability. Editors no longer need to hammer out a long line of Atex-style code, but can instead say[uf22] to Magna, which will use format 22 in response. Superpage allows writers to name tags with logical l abels and place these in word processor text. DeskSet offers eight user-defined keys that can hold a format string, but a typical magazine page would require many more strings.

Making operational decisions

From a practical point of view, you'll want to make some operational decisions about how much the editor has to code, and to what extent the page layout artist should get into the act. Editors don't have to fill the file with instructions, but thsoe typographic choices that are context based are best encoded when content is being examined. All these packages are at pains to make the coding task painless, and potential users should not be intimidated by the need for editors to embed at least some codes.

The batch element in here is an elusive characteristic. The most zippy interactive WYSIWYG package has to call time out occasionally to recalculate the page's type, but the halt is measured in seconds, and you're returned to the page itself. Batch processing is the computer's chance to go off by itself and follow a long set of instructions you've written for it. Technically, you write batch instructions without seeing the results until the batch file is run. This is the opposite of interactive processing in which, delayed or not, your choices are mirrored on the screen.

Code-based programs always have an aura of batch processing around them, but hey make up for this by providing a page preview display, which you can toggle on and off. MagnaType is specifically of this persuasion, and Superpage may win the hairsplitting award for callint its display interactive: The screen is a true WYSIWYG display at all times, but editorial changes are made in a little window and don't affect the page until you accept the code revision. Now, please note, it's a good hair to split--Superpage may have hit the best combination of batch power and visual revision, sparing the user from lines of writhing text that squirm to refustify as you insert three dropped characters.

Whatever the page make-up paradigm, the user labors with it as an interface to procedures he previously accomplished in the three-dimensional world.

Aiding him in the transition is themouse, which is designed to make movement around the screen natural and instinctive. Some page make-up packages don't require a mouse for control of operations, and MagnaType doesn't support one, but for the msot part, layout artists will benefit from using a mouse.

Mourse use is about as intuitive as those Apple computer ads imply. Far from being a goofy gimmick, the mouse is actually a very sophisticated cursor control device that decodes XY coordinates effortlessly. Compared to hammering on the left arrow key for five seconds, zipping the mouse around the screen is a pleasure; the movement always seems to mirror your search for a location, rather than a cursor's slow journey to a spot you relentlessly chart for it.

* Display: All the packages discussed here can use conventional display monitors (and MagnaType can use only those), but large screen, high resolution displays will be of interest to some publishers. Resolution is measured in pixels per inch, and smart shoppers must perform several calculations to determine what a vendor's spec sheet really means. A report on the number of pixels the monitor provides must be placed in the context of the screen dimensions: a 1600 x 1200 pixel display on a 15-inch monitor is a much tighter resolution than the same pixel count on a 19-inch monitor.

Selecting an aspect ratio

Choosing a high resolution monitor means selecting an aspect ratio. Deep (or portrait) monitors show a single page in great glory, while wide(or landscape) monitors offer the best view of facing pages. Prices range from $1,800 to $3,0000 for high resolution monochrome monitors.

* Output: although we've all seen some fractional ads created on laser printers and sent proudly to the production department as camera ready, we k now that m a gazine applications require high resolution output. I'm not embarrassed to say that I didn't know what the dot per inch resolution of a phototypesetter was until those 300 dpi printers appeared to make me realize it was a great deal more than that. It turns out that 1270 dpi is the minimum for clean resolution, and 2540 dpi is desirable. These are the resolution ranges of the Linotronic 100 and 300 typesetters, devices that accept PostScript instructions.

The three high-end packages discussed here will drive not only PostScript equipment, but typesetters from Autologic, Compugraphic, Varityper and others. The route to high resolution is not a hard one to travel, and you're not specifically confined to PostScript equipment, either.

Laser printers make splendid proofing devices. they use plain paper, make very little noise, require no plumbing, and are happy to print your memos as well. But if one is to be used in your DTP system, it will need to be PostScript-equipped. You'll want to feed it some font software and as much RAM as you can afford, since the fonts and page descriptions consume memory in very large chunks.

As of this writing, a PostScript laser printer would run you about $5,000. The main reason these compact little devices cost more than a Yugo is that they contain a full bore processor chip, at least a megabyte of RAM, and the PostScript printer controller itselF. They're computers in their own right, able to perform myriad calculations on PostScript instructions and print a picture of the results.

COPYRIGHT 1988 Copyright by Media Central Inc., A PRIMEDIA Company. All rights reserved.
COPYRIGHT 2004 Gale Group