Scanning

Many of us who are interested in genealogy have inherited old photographs of family members. They were usually kept in a family album and a family matriarch or patriarch would identify each individual. If you have photos that you can't identify, check out our article Who are These People. Now genealogy programs allow the inclusion of those old photos into a document that can be reproduced and distributed to all interested family members. So how do we get the old photos into the computer? We use a scanner.

Scanners have become very reasonable in price and many people now have them in their homes. Many more will acquire one in he next year. And while it's hard not to get some useful output from scanners, knowing certain basic principles can save time, disk space and transmission time. Following are some principles of basic scanning and some creative things that can be done with scanners and image processing programs.

Types of Scanners

Scanners generally fall into about 4 categories;

1. Flat bed scanners; Most people will acquire a flatbed scanner that operates similar to a copy machine. You raise the cover and place the document to be scanned face down on the glass surface. Depending on the software you use, you cause the scanner to scan the document, usually first in a preview mode after which you can adjust some parameters prior to conducting a final scan.
2. Hand-held scanners: These seem to have fallen into dis-favor with the advent of lower priced flat beds. They were dragged across the item to be scanned, covering only part of the image and a second or third scan was necessary to complete the full scan. Skewing, misalignment and the need to "stitch" larger images together make these less than favorable for quality work.
3. Sheet-fed scanners: These scanners require the image to be scanned to be passed into the machine on rollers so that only flexible sheet documents can be scanned. They were cheaper than flat beds at one time but the restriction that thick documents and books can't be scanned limited their usefulness.
4. Film and slide scanners: These come in two flavors; attachments to flat bed scanners and stand-alone slide scanners. The attachments scan at the basic resolution of the flat bed scanner which generally isn't enough to allow significant enlargement of the scanned image. Stand-alone slide and film scanners have resolutions of in the range of 3000 dots per inch which allows significant room for enlargement.

Resolution and Image Size

Scanners have a limited number of sensors with which to capture the image and thus have to "sample" the image. The more samples that can be obtained, the closer the image comes to being a continuous tone image as a photograph. The number of sensors across the scanner determines the "width" resolution while the number of discrete stops in the "length" direction determines the length resolution. Generally these are 600 dpi width by 600 dpi length where dpi is the number of "dots per inch" or samples. For a full color image, each "dot' or sample consists of red, green and blue information. 24 bit scanners (which have been the standard until recently) have 8 bits of each of the 3 primary colors. More recently the number of bits has risen to 30 and as high as 36. With 24 bit scanners, each color is represented by 8 bits which provide a range of 256 shades. The total color gamut is then 256 times 256 times 256 or over 6 million colors. Note that 8 bits equals one Byte and computer memory and disk storage are measured in Bytes. (We capitalize it just to enforce the definitions of bits as b and Bytes as B. Mb is megabits and MB is megabytes. One is 8 times the other.)

A full size 8 in by 10 inch image scanned at 600x600 resolution is 3x8X600x10x600 or 86,400,000 Bytes of data. That takes a lot of disk storage and would take in excess of 4 hours to send to someone over the Internet using a 56k modem. (Modems are measured in bits per second not Bytes per second). Corrected 2/10/2004

Images generally fall into 3 categories: color, grayscale and line art. The discussion above assumed color images; grayscale includes only black in various shades of lightness that we call grey. A 24 bit scanner would scan greyscale as 256 shades of grey running from white to black. Such an image would require only 1/3 of the storage space and transmission time of a color image. The last category is line art such as a drawing which generally includes only black and white and needs only one bit of data. It requires 1/8 of the storage space and transmission time of a greyscale image.

Matching the Scan to the End Usage

Before you scan anything, take a few minutes to plan on how you are going to used the scanned output. Perhaps the end usages will fall into one of the following cases.

1. Computer (including e-mail and Web): Computer screens are generally accepted to be 72 pixels per inch.(Here a pixel corresponds to a scanned "dot". The term means "picture element".) It's appropriate to scan at the resolution of the screen so scanning at 72 dpi would produce an 8x10 inch image of about 415,000 Bytes. Since most scanners use 300 or 600 as the base dpi, scanning at a sub-multiple makes sense so 75 dpi is commonly used. To save more storage space and transmission time the image can be scanned at half size or by some other reduction factor. If an 8x10 image is reduced to 4x5 the image sized is reduced by a factor of 4 (2 in each dimension) so the above image would be about 104,000 Bytes. (Other ways of reducing image size will be discussed later.)
2. Printing: A general rule of thumb is to scan an item for printing at 1/4 of the printer resolution for color images and greyscale and at the full resolution of the printer for line art. Thus, a 600 dpi printer requires only a 150 dpi scan to produce quality output. Some printer manufactures argue that a 600 dpi printer printing in three colors is actually an 3x600 or 1800 dpi printer and and scans should be at higher resolutions. Try scanning an image at 600 dpi, making a copy at 300 dpi and a copy at 150 dpi and then print them and see if you can tell the difference. Choose the resolution that satisfies you. Note that the paper used in the printer makes a significant difference: most printers allow 4 settings for paper including plain paper, quality ink jet paper, photo quality paper and photo paper. Be sure to match your printer's setting to the paper you are using. Printing at higher resolutions on plain paper will not gain much and printing at lower resolutions on photo paper wastes ink, may smudge and doesn't do justice to the image. Note: with printers having resolutions of 1440 and 2880 and higher DPI, scanning at 1/4 the resolution of the printer no longer applies. Try printing at 150 DPI; if that doesn't satisfy you go to 300 DPI. You should not need to go higher.
3. Printing fabrics and tee shirts: The resolution won't get any better than the thread spacing on the fabric so don't waste a lot of storage space those images. Perhaps 75 dpi is all you need.
4. Fax: If your scanner has a fax output, use that. Faxes have to be scanned at different width than length resolutions.

Image Formats

Your scanner can probably save your scanned output in a multitude of image formats. What are they, what should you use and how do you use them? We won't cover them all here but we will discuss the most popular.

1. Tagged Image File Format: This is a .TIFF image but with the original DOS 3 letter limitation on extensions it has the extension of .TIF. It is a bit mapped image format in that each pixel in the image is represented by a corresponding Byte or group of Bytes in the data. It is common to both Windows and MacIntosh platforms.
2. BMP: This is another bit mapped image that is used by windows in wallpaper and is the image format used and generated by the standard Windows Paintbrush program.
3. GIF: Created by Compuserve using a format that Unisys claims is proprietary, this is a common image format used on the internet. It uses lossy compression to reduce the image size and can only contain 256 shades of colors. Much of the net's animation is through animated GIFs. It is not good for photographs because of the limited color palette and all images on the same page must use the same 256 shade palette.
4. Joint Photography Experts Group: This is .JPEG or .JPG and is a compressed file format. Compresssion is lossy in that some of the detail is lost, starting with the very tiny changes that generally are not noticable. The amount of compression is selectable so that images can be reduced to a storage size that is a good compromise between loss and image size. This is the common format for putting photographs on the Net.

Working with Images

Scanned images retain all of the faults and blemishes of the image from which they were scanned. Many old photos lack contrast and may have faded significantly. Some may be scratched, have writing on the face of the picture or other insults that make the image less than it was when your ancestors prized it as the greatest technological advance they had ever seen. But all is not lost. Here are some things that can be done:

Most scanners come with some form of imaging program that can manipulate, correct and enhance the scanned image. Generally that program is Photoshop LE. But whatever it is or whether you use another program, the process is generally the same.

If you have an image that you are going to work with, scan it in .TIF format. You may eventually want it in .JPG format but every time you open and re-save a .JPG image, it gets re-compressed with the attendant loss of data. Keep it as a .TIF until you are ready to save it for the final time. Once you save your .JPG image and are satisfied with it, you can dump the .TIF.

Some programs simplify the process so as to keep the operator out. Others offer many more options and put the operator in charge. For the simple programs like Adobe Photo Deluxe, you go through a process called "fix photo" which tries to make the photo into what the program thinks it should be. It's simple, usually impoves the photo but doesn't handle all of the problems associated with antique photos.

Programs like Photoshop LE, Picture Publisher, or Microsoft Photo Editor allow more control of the process. The first step is to correct the contrast and brightness. You can do it manually in all of the programs but they also allow you to try an automatic correction. In Picture Publisher it's called "Stretch Detail" on the effects menu; in Microsoft Photo Editor it's on the effects menu and you can use "balance" to do it manually or "auto balance" to do it automatically. If you don't like what you get you can go to edit and click on "undo". So try as much as you can and "undo" what you don't like.

If the colors are off you can use the color balance in any of the programs but if we're dealing with old photos they probably aren't in color. Some may look that way because they have yellowed with age. You can correct for this in scanning by scanning them as grayscale and not as color.

Once you have the color, brightness and contrast the way you want them, it's time to clean up scratches and other marks. Some small marks can be "smudged" out. Most programs have a "smudge" tool but that is generally unsatisfactory for areas large enough to show the details of the smudge. It's better to use the "clone" tool and copy an identical section of the image to the area to be corrected. That process includes the texture of the affected area where as "smudge" tends to destroy it. Areas that have been damaged such that significant detail is lost require more artistic work. If the left eye is lost, copy the right eye and rotate it 180 deg. Darken or lighten it to match it's surroundings. Draw in what you have to but unless you're very artistic, that is not a very satisfactory approach. If the damage is near the edge of the photo, consider cropping the image as long as significant detail is not being lost. Use an oval or circular crop or mask and use a textured or gradient fill outside the mask area. A crop would eliminate everything outside of the crop boundary while a fill would paint texture or a gradient color range outside of the mask area.

If you have a grayscale photo and want to jazz it up, you can color it by masking certain areas and using a color fill. In order to have the fill look like its not one solid color pasted on, you have to choose a transparency that lets the underlying texture show through. Most programs have "layers" or "objects" that let you manipulate them before you "combine" them into a single image. The transparency of each "layer" can be adjusted until the image looks right.

Return to Pastfinders Home Page