Falconius describes part of the problem. The real problem is how the image is sampled. A quick internet search on "image resampling" and "moire patterns" and "aliasing" will turn up a lot of discussions on the subject. The (overly) short version is that if you make an image smaller by just picking a pixel at a multiple of the image size (e.g. shrinking an image to 1/3 its size by picking every third pixel), then any pattern that happened to be in the original image at every third pixel will be made more apparent. A grid, for example, might drop out lines and a pattern that is composed of every other sample will show as a larger checkerboard pattern. Very broadly speaking, shrinking an image involves using all pixels in the source image to produce the destination image (typically, a blur as big as your shrink ratio is a good starting point); enlarging an image involves trying to create detail at the right scale to avoid artifacts like blockiness and ringing. http://www.cartographersguild.com/tu...explained.html has some good discussions on the subject.