dacs.doc electric

In the previous section, we discussed three ways to get a photographic image into digital form — film scanning, document scanning and digital camera. Now that we have the image in the computer, how do we save, store, and recover it? We use compression.

When a 35 mm slide is scanned into the computer, it can have an image of 28 megabytes (9.33 megapixels). This is a big file to store on a disk drive or send over the internet. One might conclude that we have 28 megabytes of information, but that is not true. We have 28 megabytes of data and maybe 2 megabytes of information. How can this be?

In 1949, Claude Shannon, together with Warren Weaver, published a book that contained a formula for measuring the quantity of information in a communication. This was the first time that someone viewed information as a measurable quantity. Looking at a Kodachrome slide, one can now measure how much information it contains. With that measurement, one can then determine how much computer memory would be needed to store just that information in digital form.

Theoretically, the amount of information on a slide is infinite, but there is a practical limit. That limit is the amount of information needed to reproduce the slide in a form that will be acceptable to a viewer. Weaver showed that information was dependent upon probability and predictability. Using his theory, the amount of information contained in a pixel is affected by what we know about the pixels that surround it.

Enough theory!

The practical side of this is that in 1986 work was started to establish a standard method for representing photographic images in digital form. The group was ultimately called the "Joint Photographic Experts Group (JPEG), under the auspices of the International Organization for Standards (ISO), and the result is known as the "jpeg" standard. Much of the work was done at IBM Research in Yorktown Heights by William Pennebaker and Joan Mitchell. Files that use that standard have the file extension .jpg or .jpeg. The conversion of a 28 megabyte "full" image into a jpeg image is called "compression."

Jpeg is not the only way to compress a photographic image, but it is the most widely used. The method does not deal with pixels individually, but with square sections of an image that are 8 by 8 pixels, or 64 pixels each. The standard allows for different levels of compression, ranging from very poor quality to very good quality, depending upon settings used by the compression program which, in turn, results from the users' needs and desires.

How does jpeg work? I have a 600 page book that clearly describes it. You only need to know that the data in the 8 x 8 pixel square is represented by mathematical formulae with constants. If a constant is calculated to be 4.3765344787, for example, it can be "rounded" to 4.377 in one case, 4.38 in another case, and maybe just 4 in still another. As the constants are rounded, the amount of storage required to store the information is reduced, and the file size is reduced accordingly. Of course, 4.38 is not the same as 4.3765344787, so some information is "lost" in the translation. What is important is whether the difference will be noticeable to viewer. Because some information is lost, jpeg is referred to as a "lossy" algorithm.

It is typical for a jpeg file to be one tenth to one twentieth the size of the original uncompressed file. A 28 megabyte file can be compressed into a two megabyte file with little noticeable difference in image quality, and one megabyte with acceptable quality. If compressed to 100 kilobytes you would definitely notice a difference.

We don't need to go into many alternative ways of compressing an image, but some of the more popular ones include "gif", a format patented by Compuserve, and "stn" (Sting) images created by the Genuine Fractals program. Gif images work best with drawings and images using spot color, like a color cartoon. Still, very decent renderings of photographs are now available through advances made in the gif compression methods. Sting images use another patented approach which claims to be better when small images are to be rendered at a larger size. You will be the judge.

Uncompressed images may have file extensions of .tif or .tiff if they are of the "Tagged Image File Format." This format allows for some negligible compression. It is a useful format because it is understood by almost any photo editing program. A similar uncompressed format used by Microsoft is .bmp, or simply, bit-map. There may be a hundred different image formats out there, each for a particular purpose, but those that I have mentioned cover most users' needs.

A digital camera will usually store its images directly in a jpeg compressed format. (Some may offer uncompressed formats as well.) That is why a 5 megapixel image will only take 1 megabyte of storage on your memory card. However, when you bring that image into your computer for editing purposes, it will be "uncompressed" into its original size. The five megapixel image which uses 1 megabyte on your card will become a 15 megabyte file in your computer. I know of no photo editors that can work directly with a compressed image, although I have heard mention of their development from time to time. Their time has passed, anyway, since memory is so cheap that it is easy to get enough memory on a computer to handle almost any situation.

The advantages of image compression are obvious. You can store 600 1-megabyte (5 megapixel) images on a regular CD. You could only store 40 uncompressed images. And the time needed to send an image over the internet makes sending uncompressed images virtually impossible.
The methods for storing, transmitting and retrieving images are the same as for any other kind of file. The risks are the same. Magnetic media, such as floppies, lose their data over time. CD's are a better choice for long term storage, as most computers now come with CD writers. However, there will be new standards. The long term risk is whether, in fifty years, your grandchildren will be able to find a CD reader when they want to look at their baby pictures. Oh well, they probably won't want to look at them anyway. But if they do, you may be better off printing them.

All along, we have been discussing only single "still" images. You are probably guessing by now that the same reasoning applies to motion pictures as well. Of course you are right. There, the standards group is called "Motion Picture Experts Group" and their standard is called mpeg. Makes sense. Interestingly, motion pictures can be compressed further than still images because successive frames of a film are very similar. In addition to visual information, motion pictures also require a soundtrack, hence the same information-theory reasoning applies to the compression of the soundtrack, and is called MP3 —which is an acronym created from an acronym, with the number 3 referring to the expected level of sound quality. It all fits together. And MP3 is now flying high in its own right for compressing music. Just as with photos, a 20 to 1 compression ratio is possible here as well.