Scanners: An Update

Halvor Moorshead takes a look at recent developments in scanning technology.

In the July/August 1998 issue of Family Chronicle, we published an article called Scanning Tips and Tricks. Since this time a number of things have changed. First, a lot more people have got scanners — today they are almost as common as printers. Second, the resolution of most scanners being marketed is 600dpi (dots per inch) compared to the 300dpi of a year ago. Thirdly, the quality of image produced has improved noticeably.

The Epson Perfection 636 is one of the best new scanners available for the home market. Resolution is 600dpi, it is far faster than earlier models and produces excellent images.

Resolution
The resolution of a scanner can be specified in a number of ways. The most impressive sounding rating is the Maximum Interpolated Resolution where figures of 9,600dpi are common. Interpolated resolution figures are not very helpful in determining the quality of the scanner as these figures depend on software, not on the scanner.

A second measure is the Maximum Hardware Resolution — this will usually be shown as two different figures — for example 600 x 1200dpi or 600 x 2400dpi. It is easy and inexpensive for scanner manufacturers to produce resolutions of a high order in one dimension. The scanner's sensor (part of the moving bar which produces the image) has a maximum resolution depending on the model — currently most of them are 600dpi. However, if the bar is moved at half the normal speed across the image, it can produce a resolution in one dimension of twice (or more) that of the sensor.

The optical resolution is the number you should look for — it is the figure (almost always the lowest of the different ones quoted) determined by the sensor.

The significantly increased resolution of the scanners now available is less useful than it would seem. The most common photographic print is 6 x 4in. Scanning one of these in color at 600 x 600dpi produces an image of over 25 megabytes. Now we all know that modern computers have got large capacity hard drives and you can now save data on Zip drives or make your own CDs. Even so, 25 megabytes — is a huge file and few applications will need this.

The same photograph scanned at 300dpi is about 6 megabytes and at 200dpi, 2.7megabytes. If you are interested, Family Chronicle images are supplied to the printer at 200dpi and the printing process cuts this to 133dpi.

The high resolution is useful when you want to copy a small portion of an image and blow it up but this will only apply if the original has ultra sharp focus and resolution. Look closely at a photograph scanned at 600dpi and you can easily see the grain of the photograph's emulsion.

Compression
In our previous article, we mentioned how useful compression can be for saving images. It is worth repeating. You can reduce the file size of your images by 90 percent or more by saving the files in the JPEG format. When you open a JPEG file, it is automatically expanded back to its original size. The drawback of compression is that there is a slight loss in quality and this is compounded every time you open the image, make a change and save again. For this reason, JPEG files should not be used if you are retouching the image (which may require numerous openings and closings of the file) but for most purposes you can safely save your images as JPEGs.

Optical Character Recognition (OCR)
Every scanner I have tried has come bundled with OCR software. This enables you to scan printed and typed material and convert it to text files. Many people try OCR, notice loads of mistakes and give up. The accuracy of OCR depends on the software and the scanner — and both have improved considerably in the last couple of years. The OCR software supplied with your scanner is almost always a "lite" version of a sophisticated software package that the publishers are trying to sell you. I have used Caere's OmniPage Limited Edition and Xerox's Textbridge Classic (both "lite" versions). These basic packages are perfectly useable for straightforward work. Currently it costs $99 to upgrade OmniPage Limited Edition to Omnipage Pro and $49 to upgrade Textbridge Classic to Textbridge Pro. The "Pro" versions are definitely better than the simple packages supplied with the scanners as they learn, can retain formatting, allow you to copy tables and more.

Scanning images provides almost instant gratification but this does not apply to OCR; this is probably why it is not pursued more often. Most of us have far less use for OCR than for scanning images. However in some circumstances it can be an enormous time saver. For example, I inherited the only known copy of a small book published in 1939 about the life of my great grandfather shortly after his death. I have made several dozen photocopies of this for his descendants but these are unlikely to survive for many generations. I have now transferred the whole book to text, carefully scanned the photographs from the book and have added others from family albums. From this I have made several CDs to distribute to his descendants. It by far the largest job I have tackled using OCR. At first it was slow and boring but as I grew more familiar, I grew better and better — the final 60 pages took me only as long as the first 10. Not many of you will have such an ideal usage but when you become familiar with the operation you can transfer typed notes, family histories and other material to text.

If the original material shows lots of errors when processed, you might try a trick that has worked for me. Try scanning the material into image processing software (PhotoDeluxe, PhotoImpact or whatever comes with your scanner) and try to improve the image in this software by increasing the contrast, deleting unnecessary text, illustrations and odd marks and then saving the image. This can then be imported into the OCR software. It will not always help, and may in fact be worse, but in some cases it has improved accuracy considerably. I wish I could be more specific on the steps you take but it will depend on the nature of the original. Obviously better images available for the OCR software will result in better output.

OCR will not handle even the neatest handwriting at present and this ability may be decades away.

These three images of the author from the same photograph demonstrate clearly the huge improvement in image quality and the major fall in prices that has taken place over the past five years.
Image from a professional scanner that sold for $7,000 less than five years ago. It still produces good images but these are always rather red: this however can easily be compensated for in photoediting software.	The image here was produced by a scanner that was the best rated in a test of products aimed at the consumer. It sold for $300 three years ago. At the time this produced better images than several other scanners to which it was compared.	The image from an Epson 636 (one of three scanners selected as Editor’s Choice by PC Magazine). It retails for about $300. Unless the printing processes has introduced color-shifts, it can be seen that this image is far better than the other two.

Image Quality
The quality of images from the new scanners is significantly better than those of even a couple of years ago. We show three images, each from a different scanner. Until this issue is printed, we won't know how each compares as there is a small "shift" in all images as a result of the printing process. The difference in quality on the computer's monitor is very significant.

Slides
Adapters for slides are available for many scanners. Few of us take slide pictures any more but we will often have large collections of pictures from the 1950s and 1960s when this was popular. A scanner is a good way to copy and distribute these. I scanned several hundred slides and copied them as JPEG files to a CD about a year ago using a dedicated slide scanner. However, I am not happy with the quality. Slides inherently have far more information on them than prints but this does not seem to have been captured in the transfer even though the equipment I used was from a reputable company and was designed principally for slide scanning. I am going to keep my original slides handy until I am sure that I can transfer with little or no loss of quality.

The Whole Picture
For those of you did not see our previous article, we will briefly repeat other points:

1. Scanners allow you to control the brightness and contrast of your images. Normally you should do your manipulation in your image editing software, not by changing the scanner’s defaults.

2. The default setting on your scanner will usually be color. If you scan black and white images in color you will often not see the difference — but your file will be three times the size that it needs to be. There is no advantage at all in saving black and white as color.

3. All the photo editing software I have seen (which includes many but not all packages) have a facility to improve the scanned image with only a single mouse click. This sets the brightness and contrast ranges to what the computer decides are optimum settings and then sharpens the image. I almost always use this feature. It does not always help. If it does not, you can select UNDO. In a few cases, especially with old, faded photographs, the improvements can be remarkable.

4. Photographs in newspapers and magazines are made up of a number of small dots. These can present problems when scanning, especially when the pitch of the dots on the original and the resolution of the scanner are nearly, but not exactly, the same. Interesting but undesirable moire patterns can result. Some scanning software allows you to "descreen" such images during the scan; others want you to remove the screen after the scanning.

5. Let's face it, even with compressed files, pictures take up a lot of hard drive space. I keep mine on CD using a CDR (Recordable CD). Although I have the ability to copy onto rewritable CDs, I much prefer the permanent medium (recordable CDs cannot be overwritten). CDRs cost about $250 and the disks about $1.50; not cheap but much cheaper than a decent camera and film processing. Once on CD, the images are easy to copy, if you have the equipment, with absolutely no loss (one of the advantages of digital technology). One CD will hold about 1,000 quality images if they are stored in compressed format.

6. It is not fair to judge a scanner by the quality of a printed image. Printers have improved continuously but are still the weak link in the chain. There are high quality printers which do a reasonable job but they are not cheap. Some photographic and office stores will now output your digital images on quality equipment at a reasonable charge.

This article originally appeared in the November/December 1999 issue of Family Chronicle.