This article serves to illustrate the effect pixel size on the scanner can have on the resultant image. In each pair of images, below, the same subject is being scanned. The same area of the subject is being scanned, and the output of the scan has the same number of pixels (16).
To create these illustrations, I started out with an image of random noise. To make the picture somewhat interesting, I increased the saturation in one corner and decreased it in the other. I did a similar thing with lightness. The overall effect is that of a random image at very high magnification.
Next I made three images by adding black lines to the original image. The black lines represent the portion of the subject not seen by the scanner because the scanning area of each pixel does not cover the area.
Subject viewable by scanner | Result of scan | Comment |
In this "scan" each pixel "sees" a relatively large part of the subject. |
Subject viewable by scanner | Result of scan | Comment |
In this "scan" each pixel "sees" a smaller portion of the subject. |
Subject viewable by scanner | Result of scan | Comment |
In this "scan" each pixel "sees" only a very small portion of the subject. Note that although the overall picture is "noisier" it more closely resembles the character of the original image. |
The thing to notice is that in the first scan, the resultant image more closely represents an average of the overall image. The image scanned has relatively little noise. The second scan, since the pixels see less of the subject, is noisier. The image resulting from a scanner with pixels that see a smaller area represents less of a statistical average (for each pixel) and more of a random representative.