illustration systems -- current technology oveview By Donald Jenner (or whomever) Any editor will tell you, a story is not just a column of words. Effective copy needs equally efective illustration. This is no less true for "desktop" publishing than for pages made the old-fashioned way. New tools bring the kind of high end illustration effective design demands within the purview of even the most modest DTP system. A year or so back, a capable computer-based illustration system started at $20,000, and could easily cost three times that. Today, the entry point is around $5,000, and a do-everything system need not pass the $15,000 point. The change is attributable to the combination of commodity-priced hardware with Microsoft's Windows version 3. Cheap hardware and better software spells out a new strategy for better price/performance in production graphics. Windows version 3 (Win3) solves several problems that have tended to put off potential users, or make the PC-family systems expensive when set up for graphics. First, Win3 is a nice, easy-to-use environment. It features the point-and-shoot capability that has won friends for Apple's Macintosh systems. In fact, it improves on the Apple interface in several ways -- not surprising, since Microsoft's system software geniuses have learned from previous designs, including their own. An example: With a Mac, when a pull-down menu opens, only by holding the mouse button continuously will it stay opened; let go of the button and the menu snaps shut. This can be disconcerting; it has led alternative-mouse makers to add a special button to correct the problem. In Windows, that menu stays open until either a command is selected, or the user clicks on another part of the screen. Another example: Windows is really intended to support several open applications on screen at once -- even on machines with limited memory; try that on a Mac, and it either doesn't work, or you get the dreaded bomb message -- and unlike Windows, that means not a clean termination, with little loss, but a complete system restart. Win3 really is friendly. Second, Win3 makes software installation a great deal simpler. One great objection to the use of PC-family, DOS-based systems has been the seeming difficulty of setting up DOS. There is a good reason for that complexity; those objectionable CONFIG.SYS and AUTOEXEC.BAT files together constitute a user profile that makes it possible to customize a powerful system from a wide array of off-the-shelf parts. But some folks just don't want to be bothered. The Windows installation process can do most of that automatically, or it can allow different degrees of control, to match a given user's ability and interest in controlling the process. Win3 offers an advance over earlier versions, by allowing post-installation modifications from within Windows, with simple step-by-step dialogue-box guidance. Third, Win3 breaks through the DOS memory limitation. This was the greatest problem PC-family machines posed to graphics users. Graphics-computing is memory-intensive; the data is substantial, the programs are big, and things run slowly at best if there is not a lot of memory to use. Win3 provides strategies for using up to 16 megabytes of memory for both 286- and 386-based systems; on 386 systems, it can swap in unused hard disk space as a supplement to real memory in the system. The net result is more complex pictures at greater speed. What really makes Windows interesting -- if you're a graphics user -- is that is it supports an array of first-class graphics applications. The basic tools for illustration are represented by three top-notch applications: Micrografx Designer and its predecessors (formerly billed as "the cure for Mac-envy") created the whole Windows graphics marketplace. Computer Support Corporation brought out Arts & Letters Composer, followed by Arts & Letters Editor; they built substantial expertise derived from Diagraph into a fully professional illustration system. Corel Systems moved from sophisticated hardware into sophisticated graphics software and didn't miss a beat -- in the process Corel Draw became the darling of many professional illustrators; Corel 2.0 is presently the program of choice for many professional illustrators. These design tools are complementary, as much as they are competing products. Most serious designers use two or three, playing off the strengths of each for the particular job at hand. Each of these programs is fairly modest in price (around $600-$700 list, quantity-one -- with substantial discounts in the cheap-software stores); having them all is not a difficult business decision. For this you get a lot: Not dozens, but hundreds of fonts -- and the capability of modifying all of them into still more. Clipart images -- a favorite productivity shortcut -- that fill a separate book in some cases. These are production tools. Where does this leave paint programs? Most certainly not in the dust, but perhaps with a changing role. Paint is time-consuming, and very rarely a good production technique. Actually, the trend seems to be to use bitmap images, but to handle them differently. The change is driven by the advent of modestly priced color scanning. Until recently, that was the province of Sharp and Howtek (the latter company remarketing the Sharp product under its own label). Now, Microtek and Epson have entered the arena with exceptionally able products. These flatbed scanners offer letter-size scanning in color, grey-scale and monochrome, and can support up to 600 dot-per-inch resolutions for around $2,000 list. Just to make life interesting, several companies are following Sharp's lead into the world of low-cost color scanning, offering color "hand scanners." Sharp brought its product to market about a year ago; the JX-100 is essentially a scaled down flatbed scanner; you lay it on top of the picture to be scanned -- images up to 4" x 6" are accomodated. Omron and Niscan are set to enter this market, setting the price point at around $750 list, with a maximum of 400 dot per inch resolution. The new products eliminate the need for an expensive GPIB interface adapter, using a parallel I/O port -- which many folks already have in their systems serving as an unused printer port. They offer enhanced performance at prices readily competing with grey-scale-only scanners. What makes these scanners particularly interesting is the software that drives them. The initial competitors are Media Cybernetics, under its Halo trademark, Astral Development's Picture Publisher Plus and Computer Presentations, Inc.'s ColorLab. Testing the three programs on the bench, the Computer Presentations product proved the most interesting of the three. First, unlike Haloscan, which could not control our parallel I/O port, ColorLab chugged along merrily on all of the test systems, from a lowly 8mHz AT-class machine to a mighty IBM 386 box. Second, it sports a more intuitive interface than Picture Publisher Plus: Tell the software to scan, and it loads a prescan window. Scan the whole page or a portion, with whatever level of precision is needed. Select the final scan area by adjusting an onscreen marquee with the usual handles, and scan at the selected resolution and color setting. ColorLab has another nifty feature -- a software sibling called ImagePrep. Think of ImagePrep as ColorLab without the scanner control elements and you have the idea. All the image enhancement, file format control, screen-capture facility and so on is available in ImagePrep, in more or less the same intuitive package as in ColorLab. This has some interesting implications in a serious graphics studio. First, no one is going to have a scanner at each workstation; these are costly peripherals and they aren't used often enough to justify the expense of more than one for a half-dozen or so seats. So images are going to be scanned in one place, but the artist using the image may be someplace else. With ImagePrep, that is no problem. The second implication is even more interesting. A vast amount of bitmap art has been produced over the years. It is in all sorts of formats -- Truevision Targa images, all kinds of images stored in CompuServe's GIF format (which has become a kind of standard), and in the ubiquitous color TIFF, PostScript and PCX formats. Microsoft is adding still another -- BMP -- that has both Windows and OS2/Presentation Manager variants. ColorLab and ImagePrep read and write these formats. Suppose these images were published on CD-ROMs (chosen because these files can easily run upwards of a megabyte); this could be a new kind of resource for graphics artists. All the illustration software currently available can merge bitmaps and vector art (and expect the arrival of programs like SuperMac's vector/paint software for Windows), and the result of the combination is truly spectacular. What develops is a new strategy for graphics users -- whether in independent studios or a corporate setting. Tactically, it produced a new shopping list, when last the studio/testbench was rebuilt. The basic hardware fact is that the 386 is here to stay -- at least for the foreseeable future. Intel has a lot invested in that architecture, and the size of its current and near-term installed base represents marketing momentum. The successor versions (the 486 is but the first of these) will speed things up, and make for more powerful computing, but the software will be of the same ilk as that we see now. This is Intel's line, and it is consistent with much hard evidence. Most of the time, the biggest, fastest 386-type machine is not needed for illustration graphics. The increase in facility affored by a 33mHz 386DX over that yielded by a 20mHz 386SX is marginal in real-time terms. In fact, the darling of our testbench is IBM's original Model 80, sporting a relatively slow 386 processor. It does have a 387 floating point coprocessor, but that is not there for illustration graphics; all that software uses integer math. There are better places to spend extra money. Windows loves memory, and uses up to 16 megabytes effectively in lots of ways. RAM chips are cheap; think of four megabytes as a practical minimum. Let Windows set up a hard disk cache and a RAM drive for "temp" files, and leave about two and a half megabytes for programs. Another good place for your capital expenditure is the hard disk system. Windows fills itself out by using hard disk space as virtual memory. The faster the drive, the better that works. Windows applications are large, and frequently access libraries of all sorts; again, this goes better with a fast drive. Applications are enormous these days (not infrequently, a minimal installation of a graphics program eats up four megabytes); a graphics setup without at least 120 megabytes of disk storage will be constrained. Most important, of course, is the display system. A professional system has to be able to display at least 256 colors (that is, 8-bit color), with at least 800 x 600 pixel resolution. Higher resolution is better -- especially since the larger monitor that entails can sit a bit further away on the desk. Our design systems are equipped with IBM's 8514/A adapter and matching monitor, and with NEC's Multisync Graphics Engine and 3D monitor. Both these systems off-load much of the display work to a separate graphics coprocessor -- almost essential when dealing with lots of pixels in lots of colors. Without the coprocessor, screen updating is painfully slow. For folks who need the very best, the answer is "true color," a 24-bit display system, with a potential palette of 16.7 million colors. Raster Ops has this today, for both AT-buss (sometimes called the "ISA" buss) and PS/2 Microchannel machines. Expect more of these soon, including one from Hercules; some sources have reported this board will come to market early in '91, with a pricetag below $1,200 -- that is, around the same price as IBM's new 8-bit XGA adpater.... What about older AT-class systems, based on the Intel 286 processor? Don't toss them out. They run Windows very nicely, and if they have a little muscle (say, a 12mHz processor or better), make good backup or extra-seat machines. Even the slower ones do very nicely as scanner and printer servers. Our testing system uses an 8mHz Tandon Targa box to control Epson's ES-300C scanner; we gave it a bigger hard disk capacity and a fancy display system (Headland's VRAM VGA tied to an NEC Multisync monitor) and regularly use it to process very detailed scans under ColorLab software. It is slower than the heavy-duty iron, but we're usually off doing something else during the slow, automatic parts of the process. The same system also supports our Hewlett-Packard LaserJet III. The color printer hangs off one of the design stations. Needless to say, networking is part of the survival tactics around here. We use Artisoft's Lantastic software and proprietary two-megabit-per-second adapters. The speed of ethernet is attractive (and Lantastic can run on ethernet), but the proprietary adapters are perfectly adequate (easily as fast as our hard disks...). Lantastic requires very little memory in any case, and still less when used with the Artisoft proprietary adapters. The network runs over spare pairs in the telephone loop -- no pulling cable. The system is easy to use -- no one complains about it. It even supports voice (with an additional board); we use it as an intercom. Windows recognizes it without difficulty. Last but not least, you get all this and more for around $200 to $250 per station (street price). It is nearly ideal, in short. What don't we have here? First, there is no film recorder. In fact, for print-oriented work, film is the best way to eliminate problems. But a good film recorder is a big-ticket item and about as fussy a piece of equipment as there is. It is simpler to send the finished design (proofed on the local color printer) as a PostScript file to a service bureau. The cost is modest enough to make that practical, and the service is usually faster than necessary. Some service bureaux (such as Kinetic Presentations in Louisville) have added a $90,000 Canon color laser printer to their offerings; this produces printshop quality four-color paper copies at about $1.25 apiece in small quantities. This can change the nature of lowly desktop publishing. The point to be taken from all this is simple: We are doing professional illustration for professional publication -- they are selling and people are asking us for more. If the entire studio went up in smoke, all the stations and peripherals could be replaced for less than one seat would have cost two years ago. Capital investment is less; insurance is dramatically lower -- and there is no sacrifice of capability when it comes to getting a good job done. ###