Scanning 101

by Al Massey

What a scanner does and what to look for when buying one.

I suppose Chicken Little was right after all, “the sky IS falling,” at least where scanners are concerned.

A quick “scan” of mail-order catalogs shows that a little over a year ago you had to pay upwards of $1,000 to get a “good” scanner and around $500 to get an “acceptable” one. Today, it is possible to get a really good scanner for under $300 and an accept able one for under $100. If I could offer you just one bit of advice before you rush out and grab one of these little darlings it would be: “please make sure you have the horsepower before you spend your hard-earned dollars.”

Unless you enjoy watching grass grow don’t attempt to scan unless you have a Pentium processor or equivalent with at least 32 megs of RAM. Now I know you will see scanners offered that claim to run on a 486 and 8 megs but I DON’T RECOMMEND doing it. I ha ve spent countless hours scanning with 486 machines and minimum amounts of RAM and believe me, slow does not begin to describe the experience.

It is interesting to note that as scanner prices drop, convenience, ease of use and quality are on the rise. The need to reduce clutter, copy important documents, maintain web pages, scan photos, produce flyers and newsletters is making the scanner as im portant as the keyboard or mouse in getting information into your computer.

Scanners basically fall into two categories, sheet-fed and flatbed. A sheet-fed scanner grabs the item you scan and moves it across the scanning head, while with a flatbed scanner you lift up a hood similar to a photocopier and place the object flat agai nst a glass plate. A lamp then moves in small increments under the glass while an optical system bounces the light into the scanner’s photo sensors building a page one line at a time.

In judging which you should choose, bear in mind that although the sheet-fed types take up less space with more than one model incorporating a keyboard and are cheaper, you are bound by those same limitations. The items you scan must be flexible enough t o pass smoothly through the rollers so you are limited in what you may scan. You can scan single pages, not magazine articles or books or thick objects. I have found that one or two sheet-fed scanners do address this situation by having two settings, one for these objects and another for thicker objects such as photographs or cardboard or business cards that allow them to pass straight through without curling around the scanner head. While this approach works fairly well it is still not up to par when c omparing the results with flatbed scanners. If you look close at a sheet-fed scan on thick objects you will typically see some jagged edges. Flatbeds don’t have to deal with paper stretching or rollers that wear out so they are generally a bit more prec ise in scanning the object.

At the low end it is a matter of trade-off. Flatbeds typically cost a bit more and take up considerably more room. On the plus side flatbed scanners, even those on the low end of the totem pole, offer at least 300x600 dpi while this is not always the cas e with sheet-feds. Flatbeds can also scan just about anything you can place on the bed, books, magazines, photos, or the family dog.

When you pick a scanner and have decided between a sheet-fed or flatbed the next thing is to look carefully at the specs, particularly the resolution. At the core, scanners work by turning your typed pages, photos, and drawings into rows of tightly packe d dots. These dots are stored as bits in a file called a bitmapped image file. The tighter the dots are packed, the greater the resolution (resolution is most often expressed in dots per inch or dpi). The higher the resolution the better.

Beware! Scanner vendors, like everyone in the computer industry, are always looking for an edge and sometimes print confusing resolution specs. The scanners we are concentrating on here (under $500) will top out at around 600 dpi, with most being in the 300-400 dpi range, so if you see a $300 scanner advertising 1200 dpi or greater chances are they are clouding the issue by offering what is called “interpolated resolution.” Interpolation is achieved by adding additional or “virtual” pixels between the r eal samples and estimating their color based on adjacent samples. In other words “your guess is as good as mine.” Interpolated resolution provides more pixels, so you can enlarge a scanned image in your page layout or graphics program without getting tha t characteristic grainy appearance when the individual pixels become visible to the naked eye. This really becomes apparent on photographs that tend to get “fuzzier” as you crank up the interpolated resolution.

How much resolution do you really need? It depends on what you plan to do. Here are some general rules.

A quick guide to dpi depends on your need. If your main concern is faxing you can make do with a 200 dpi black and white scanner. For copying, you can get by with a 200 dpi scanner or one cranked to 300 dpi or 400 dpi via interpolation. For the best resu lts, make sure the scanner’s resolution matches your printer’s resolution. For document management, your scanner’s resolution should equal the resolution of your viewing monitor (or printer if you plan to provide quality printouts). A 200 dpi scan provid es a workable, though not superb, printout on a 300 dpi or 600 dpi printer. A 300 dpi or 400 dpi scan will look better, but remember, it will also eat up 1MB to 2MB of disk space.

Next you will want to focus on the scanner’s ability to capture grayscale and color images. Here you will want to be concerned with something called “color or bit depth,” because you don’t want to come up short and purchase a scanner that can’t produce t he kind of color file you need.

First is black and white, also called binary or bilevel. Because they need less circuitry and only simple sensors, these are the cheapest and also produce the smallest image files.

Grayscale boosts bit depth from two to eight bits per pixel. With two-bit scanning, you can capture four colors (typically, four shades of gray). If you opt for grayscale, make sure it is at least eight bits per pixel so the scanner can capture 256 level s of gray.

Color offers at least 24 bits. A 24-bit color scan can scan more than 16 million colors. Color scanners that cost less than $300 typically capture eight bits of color information for an image’s red, green, and blue color values. They then combine them in to a 24-bit color file. Needless to say, 24-bit color images require a lot of RAM and disk storage.

Next you need to concern yourself with the best way to connect this device to your PC. The greater the color depth the more critical speed becomes.

At the top or most expensive end scanners are all connected by SCSI interfaces, while only a select few of the lower end models connect with a SCSI interface. Most often under $300 units connect through the PC parallel or serial port.

SCSI is the speed demon here, theoretically transferring data at up to 5MB per second. This is something to seriously consider if you are doing a lot of 24-bit or higher images. In addition to a separate SCSI card you will need to steel yourself with the joys of configuring the card to play nice with your operating system. This can be a black art worthy of Merlin.

Most of the lower end scanners I have tested make use of the parallel port and have been VERY simple to set up.

Make sure the unit you select has pass-through parallel port connections allowing you to make use of the scanner and printer on the same port. Unlike SCSI, you won’t be able to scan and print at the same time but you will be able to switch back and forth effortlessly.

If you are operating a Pentium with an enhanced parallel port you should experience a transfer rate approaching 2MB per second.

Serial ports transfer data at around 56kb to 115kb and do not offer pass-through connections so they really should be of secondary consideration.

Next month I will deal with how you massage or manipulate the data once you it into your PC. We will talk about document management and OCR (Optical Character Recognition) software and provide an evaluation of the more popular packages from the low, midd le and high end.

Al Massey is a HAL-PC member who can be contacted at almas@hal-pc.org.


E-mail me at webmaster@hal-pc.org with any comments you have and tell me what you want to see here.

Back to the Magazine Home Page