This diagram brings up and interesting question. How important is it that the center of a graphics window, graphics objects, etc. fall exactly on a pixel rather than between two pixels. In many cases it does not matter.

However, in some cases it might. I worked on a project where space images (planet surfaces, moon surfaces, ...) where examined and sizes and distances where estimated. (A cursor of exactly 21 pixels was used.) This means the cursor's center was a pixel and could be placed exactly on a pixel in an image.

Theses graphics window cursors came in several shapes and sizes (crosshairs, circles, rectangles, etc.) and were selected by the user to suit their purpose.