Vol. X, No. 1

Spring, 1997

Raster And Vector Images - An Important Distinction

S. C. Jones

Many pages of this Newsletter have been devoted to explaining the differences between raster and vector images and the consequences that flow from those differences. The problems and issues involved were brought home to me again when I was preparing images for the Web from "AHREO - An Addition to AutoCad for Field Archaeologists" in our February, 1997, Newsletter. Trying to transfer the drawings to the Web provided an excellent example for discussion. The images supplied were computer-generated, but we had only the paper output, not the computer files. For the printed Newsletter, we simply pasted the drawings in place and let the printer do the rest. (Sadly, we pasted one of them upside-down in its place, see "An Apology".)

Putting the drawings on the Web was a different matter. They had to be turned into image files; so they were scanned as line art at a high resolution, and we expected to use the resulting images on the Web. Scanning, however, produces raster images, and therein lies a serious problem. Although our eyes see lines on the paper as continuous, a scanner does not. It sees only discrete dots of black or white. (It may look for levels of gray or colors, but for line art the system treats all dots as black or white.) The individual dots are truly discrete to the computer - isolated and individual. All lines and curves are interpretations of the human eye and brain, created by combining dots. Of course, if the dots are very close together, they may be combined easily; if farther apart, not so easily.

I scanned the figures at 600 dpi, many times the 72 dpi resolution that is considered normal for a monitor(1); so the scanner treated the paper as if it consisted of discrete squares, each 1/600 of an inch on a side. Each such square was inspected, and the result - a black or white reading - was passed to the computer. The computer then constructed a new version of the drawing. The original drawing, of course, consisted of lines on paper; the new one consisted of dots on a screen and could be rendered as dots on paper. Since we wanted a drawing to put on the Web, though, it was the dots on the screen that I examined.

When I examined the image on screen, I could select the level of magnification/minification for viewing. (I was using Adobe PhotoShop® to deal with the scan, but any drawing program would have treated the scan in the same way.) Examining sections of the scanned images at full size - one pixel on screen for each dot in the scanned image - verified that scanning at this degree of resolution did indeed yield all the details visible on the paper version. The image, though, was so large that it was not very useful. I had to scroll around to see it, because it was the width of more than seven computer screens. Such an image would be equally useless on the Web, not to mention terribly time-consuming to transmit. When I reduced the size of the image so it would fit on the computer screen, however, the image was barely legible. The reason for that lies in the use of dots to represent lines.

The computer does not recognize the dots as parts of lines; as I said earlier, it sees only discrete dots. When the size of the image is reduced, many of the dots must be ignored, but there is no relationship between the presence/absence of a line and the choice of which dots to ignore, because the computer does not understand that lines are involved. The retained dots comprise the entire visible image, but they no longer adequately represent the lines on the printed page. Too many black dots have been ignored. This version of the drawing is useless.

There are two ways to reduce the size of the drawing, one is temporary, for viewing the drawing on screen; the other is permanent, making a new file based on fewer dots. Both methods, not surprisingly, produced the same results - dismal. In both cases, the individual lines of the original drawing were reduced to meaningless dots, which could not be perceived by the human eye/brain to be continuous lines.

I then tried scanning the image directly at 80 dpi to see if the resulting image would retain the sense of lines of the original. It did not. The level of detail was virtually the same as in the image which had been reduced from 600 dpi.

The solution is to create an image that is understood by the computer to consist of lines, not dots. We have two programs that can interpret dots and create lines from them, but the complexity of these drawings is such that trying to do so on these drawings seemed pointless. So these drawings will not make it to the Web in useful form. That is one of the little, practical difficulties we have yet to conquer with our current technology.

For other Newsletter articles concerning pottery profiles and capacity calculations or the use of electronic media in the humanities, consult the subject index.

Next Article: Combe Capelle Data Coming to Archaeological Data Archive

Table of Contents for the Spring, 1997 issue of the CSA Newsletter (Vol. X, no. 1)

Master Index Table of Contents for all CSA Newsletter issues on the Web

Return to CSA Home Page

(1) Monitors vary in output resolution, but 72 dpi was established as a standard for some purposes by the Macintosh. Current monitors may use higher levels of resolution; a few use lower levels. An image of a specific size, stored at 72 dpi, will appear somewhat smaller on a monitor using a higher resolution. Return to body of article.