Paperless Publishing-Colin Haynes - Chapter 14

This is an archive page !!!

 OBS   Back   Forward   Contents   Index   Align Page


 OBS   Back   Forward   Contents   Index   Align Page
Page 324

SCANNING and compression are two technologies that, sooner or later, electronic authors and publishers need to use. It is crazy to waste the time and human effort needed to retype documents existing only in hardcopy when they can be scanned and digitized by computer. Similarly, it makes good sense once a publication exists in virtual form, to exploit the fact that it can be squeezed into far smaller files by a variety of compression techniques.

These technologies are vital in the development of many forms of electronic publishing, but they are not as easy to use as they might at first appear.


Don't be seduced by the obvious attractions of optical character recognition and the somewhat confusing claims made for the technology by those selling hardware and software products. You will be very lucky if you can take that one typed copy of a manuscript, feed it through a scanner and OCR program, and get clean, accurate text on your monitor. In many cases, it might be quicker to retype the text, since only clean documents, well-printed with a limited range of typefaces at particular point sizes, on white paper without fancy formatting, will get close to the claims of 99% OCR accuracy.

If the hardcopy being scanned has complex formatting and includes a range of typefaces or fonts, or a mixture of graphics and text, the OCR program can soon become confused and inaccurate. If the spacing between the letters is less or more than normal, the characters have unusual features, the type used is smaller than 9pt or bigger than 28pt, or the pages were typed using a fabric ribbon, error rates can increase considerably. Particular problems are posed by printing on colored or textured paper, or by that overworked contemporary design device of running text over images, particularly screened photographs.

If the hardcopy has lived an interesting life and acquired stains and smudges, the task of the scanner and OCR software becomes

 OBS   Back   Forward   Contents   Index   Align Page
Page 325

increasingly difficult. Even an actual accuracy rate of over 98.5% yields a mistake in about every line of scanned text, and it is not practical to expect all those errors to be caught by even the best spell checker. Every 1% difference in OCR accuracy makes an enormous impact on the quality of the document, and if you get less than 90% accuracy, the results probably offer no advantage in time or cost over retyping.

If, for example, you are trying to compile an electronic book from a collection of newspaper clippings--perhaps tearsheets of features you have written--the multiple columns, lack of contrast, and probable use of visuals will be just too challenging for most OCR programs. You might be able to improve things a little by making crisp new photocopies, but that extra generation of copying might add as many recognition problems as it resolves.

But it is not all bad news. OCR is improving rapidly, and if you have a suitable hardcopy, the results can be very impressive. There are refinements to OCR programs, such as the AccuPage facilities with Hewlett-Packard scanner software, that can enhance contrast and other important aspects of the image. Before you spend a lot of money on a scanner and software, however, try them first to make sure that they will live up to your expectations.

When image quality is poor and retyping is not a viable proposition, you might consider scanning hardcopies of documents into graphic images and then using a file- or document-management program to publish them. Although document imaging is seen mainly as a solution to the business community's problems in managing its paper, the technology does have a place in certain publishing needs where it is not practical to digitize the text on the hardcopies.

If you are still working on a typewriter and think that you might want to create digital copies at some time in the future, it would be a sensible long-term economy to use a carbon rather than a cloth ribbon. Typing from a cloth ribbon, along with all but the best dot-matrix output, does not fare as well in the OCR process as nice, crisp, laser copies. Similarly, carbon copies and faxes are not handled well by scanners or OCR software.

 OBS   Back   Forward   Contents   Index   Align Page
Page 326


For serious OCR work involving significant quantities of material, a flatbed scanner (Fig. 16-1) is almost invariably easier to use and will give more accurate results than a hand scanner. The flatbeds are coming down in price as they increase in proficiency, and for many independent authors it might make more sense to invest in a scanner/laser printer combination than a copier, particularly if there is a low-cost copying service nearby. Some office supply stores use their copying facilities as a loss leader to pull in customers.

An alternative to either hand or flatbed scanners is the type with roller mechanisms that take up very little desk space and are more portable than the flatbeds, while tending to be easier to use and more

 OBS   Back   Forward   Contents   Index   Align Page
Page 327

accurate than the handhelds. However, the roller scanners cannot cope with bound books.

The best solution for many individual scanning needs is one of the services that converts paper documents into electronic documents. These are springing up in cities where there is sufficient demand. If there is not one near you, you can mail or ship your texts to a distant service, such as the Larry Whitaker Association (619-259-7911). Expect to pay about 50 cents for an average-sized image, and $1.50 to scan a standard letter-size page of text until competition is brisk enough to bring down prices.


Once your image is in a digital form and you have manipulated it on screen to your satisfaction, you might want to capture it to incorporate in your publication, or to print out as a hardcopy. Screen captures are widely used in the printed documentation accompanying software.

You will rarely get satisfactory results using the print screen command from your keyboard; screen capturing usually requires a dedicated program. I recommend Hotshot Graphics from SymSoft (702-832-4300). It is versatile, easy to use, and gives you the capacity to manipulate images that can be useful when preparing and marketing e-books.


Compression will be the key to the development of many forms of electronic publishing. While it is comparatively simple and reliable to compress and decompress text files to make them conveniently portable, the process becomes far more complex and unreliable when working with graphics. The potential problems increase by leaps and bounds with the need to compress the animation sequences and full-

 OBS   Back   Forward   Contents   Index   Align Page
Page 328

motion video that are becoming important elements of electronic publishing.


Video compression technology

Even the massive storage capacity of a CD might hold only 30 to 60 minutes of digital video, compared to hundreds of text-only books. Distributing video or more than a few still images on even high-density floppies is impossible. Video files also take up impossible amounts of hard disk space on typical personal computers. Compress those files to get more into a smaller space, and you can run into all kinds of system compatibility and monitor resolution conflicts when they are decompressed by your readers.

If you want to build video or large numbers of graphical images into your electronic publications, you will nevertheless be compelled to use compression techniques, probably those built into your multimedia authoring programs. Compression techniques vary in their efficiency and reliability. Beware particularly those that limit the marketability of your publications by requiring very specific systems on which they can be played back.

Lossy compression methods achieve far smaller files than the lossless compression techniques intended to guarantee that no data is lost when the file is decompressed by the user. The lossy methods work by eliminating some of the data in the original, uncompressed images. For example, the lossy compression program might eliminate some of the variations in brightness for certain colors that would not be obvious to the viewer's eyes. Or, when there are repetitions in a sequence, such as a vehicle moving across the screen, the lossy compression program will try to make intelligent guesses at which parts of the picture need not be changed for every frame.

In image compression, the tiny pixels that are the basic elements displayed on your monitor are grouped into blocks before compression. These blocks are, in effect, an averaging of the color and light characteristics of the typical 64 individual pixels within the block. When decompressed, the image might be distorted and very different from the original, with results that cannot be guaranteed if

 OBS   Back   Forward   Contents   Index   Align Page
Page 329

your publication is going to be played on a wide range of computer systems.

The multimedia market is expected to grow to $25 billion a year by 1998, a five-fold increase since 1992, and much of that growth will be made possible by continuing developments enabling publishers to squeeze still and moving images, particularly video, into their titles. It is a technology which, even if you don't begin to understand it, you must still take into consideration in your planning.


Fractal compression

Although all forms of image compression are improving by leaps and bounds, fractal transform image compression seems to have inherent advantages that give it an edge. Indeed, one of the pioneer researchers in this field, Taylor Kramer, chief technical officer of Total Multimedia (TMM), maintains, "Fractal video compression is the enabling technology that will unleash a torrent of creative products for educators, magazine and book publishers, electronics companies, and movie and music producers."

"We envision a future in which electronic publications will be routinely enlivened with video and multimedia illustrations now that fractal compression is available for DOS, Windows, and Macintosh systems," he says. "The publisher just needs to compress a video sequence once, and it will play back reliably on all these systems."

One pioneering electronic publishing use of fractal compression is Microsoft's Encarta multimedia encyclopedia, in which thousands of color images are squeezed onto a single CD-ROM. You can experience this compression technology in works from other leading electronic publishers as well, such as Compton's New Media, Grolier, and Software Toolworks. The costs are comparatively high, but you can now get fractal compression software packages from TMM (805-371-0500) for under $1,000, with a license fee or royalty negotiated on any commercial titles in which you use it.

IBM researcher Benoit Mandelbrot conceived fractals in 1975 as a way of mathematically defining irregular shapes. Fractals are actually

 OBS   Back   Forward   Contents   Index   Align Page
Page 330

coded mathematical instructions that enable your originating computer to display a picture. As such, they are much easier to compress into small portable files that will fit on a disc than are the digital elements that make up the picture. When these instructions reach their destination, the computer mathematically expands them again to enable it to display your pictures on whatever monitor it is using.

It is really not the picture that it is being compressed, but the digitized instructions for creating the picture. Compare it to mailing a friend the recipe for mixing and baking a cake, rather than having to make and send the cake itself.

The process differs completely from the physical visual steps in enlarging or reducing a photograph. There is no messing with the pixels themselves in fractal compression, so the process tends to be much faster and more accurate than with other methods. Image files might shrink to one-hundredth of their original size and decompress quickly to give a faithful image on a wide range of monitors, irrespective of their resolution specifications. The decompression process is invisible to the reader, but for complex images, it might take a few seconds. That delay you experience when running multimedia, then, is not just your computer seeking the relevant information from the disk, but decompressing it also.

The implications of fractal compression are enormous for the videotape as well as print media. TMM's Taylor Kramer predicts that cost and durability alone must make movies migrate from tape onto CD-ROM because, thanks to fractal compression, the expense of duplicating a show onto a CD-ROM will be one-tenth the cost of putting it on tape, and the disk will last 100 times longer. Movies going onto CDs will stimulate demand for playback hardware, as happened with video. That's good news for e-book and multimedia publishers because it will increase our markets.

If it were not for compression technology, some of the most interesting applications of electronic publishing would not be possible. The educational sector provides many examples. One of the most interesting is a history course created in the Hueneme School District in California. Using TMM's fractal compression technology, the

 OBS   Back   Forward   Contents   Index   Align Page
Page 331

children interact in English or Spanish with a multimedia program on the Lewis and Clarke Expedition that includes extracts from the diaries and clips from the movie Dances with Wolves.

Over 7,000 students at nine elementary and two junior high schools in the Hueneme district have been exposed to these new multimedia forms of electronic textbooks, and the results have been very encouraging. Test scores have improved to some of the best in California, despite the fact that Hueneme spends less per pupil than the state average. Attendance and discipline problems have become significantly lower, also, since multimedia entered the Hueneme classrooms. This multimedia system runs on a variety of networked PCs, including some older and less powerful systems that can only cope with sophisticated full-motion video because of fractal compression.


Zipping your texts

For electronic publishing that does not involve video, the most universal file compression system is PKWare's PKZIP to compress, and PKUNZIP to decompress. You will find copies of these shareware programs wherever shareware is distributed. By using this popular system, your files will be acceptable to bulletin board sysops and on-line services everywhere. If you are running Windows, another shareware program, WinZip, is much easier to use. The excellent WinZip costs $29 from Nico Mak Computing Inc., P.O. Box 919, Bristol, CT 06011-0919. An evaluation copy, together with the LHA self-extracting archive ideal for e-books, is included with the Writer's Electronic Publishing Kit offered to readers in the coupon at the end of this book.

You must license these compression programs to use them for your publications; there are full details in their documentation files. Several other shareware compression programs do not require licensing fees--and some of these are very good. However, choose carefully because not all are as acceptable on-line, nor as reliable. One of the safest choices is called LHA, widely available on bulletin boards and in shareware collections as freeware, which does not require licensing.

 OBS   Back   Forward   Contents   Index   Align Page
Page 332

In certain publishing situations, you might be able to use the compression facilities available in MS-DOS 6 onwards, but only with high-density disks and for distribution to systems running DoubleSpace.


The last thing you want is for the electronic publications you distribute to provide a hiding place for computer viruses, those destructive programs that attach themselves to program and data files as they move from system to system. There are hundreds of different viruses now circulating, some of them very virulent, destructive, and difficult to detect.

Scan your disks and on-line downloads coming into your system. Also, make sure that any disks that you distribute are clean. Labeling them as having been scanned and virus-free can be a help to sales. The shareware versions of Viruscan and VirusShield are well proven anti-viral protection. MS-DOS 6 and later also feature competent scanning software.

 OBS   Back   Forward   Contents   Index   Align Page