As stated before all perception based metrics are distance dependent. If someone asks us to see if there is any difference between two similar printed pictures, we will move the pictures back and forth in order to see some difference. This is true if the differences are not significant and obvious at first sight, of course. Actually, what we are doing is moving the maximum frequency on the contrast sensitivity curve, and trying to match it with a frequency where difference appears. There are lots of situations where the viewing range is limited and known in advance. For example, if some rendered images are to be presented in a classroom, the viewing distance range is known, it is between the viewing distance of the first row, and the viewing distance of the last row. If images are rendered with some progressive method, the rendering can be stopped much earlier than if images are intended to be observed from 50 cm. On the other hand, if some blurring operator is shown, the lecturer should apply quite a large blurring matrix if he or she wants all those present at the lecture to see the blurring effect. Even in conferences and university lectures, images that can not be distinguished are frequently presented as examples for differences. If the perceptual metric is taken into account such situations can be avoided. The next example could be a display device in a shopping window. Sometimes a monitor is placed in a shopping window and some animation is running. Obviously, the distance between the monitor and the window-glass is the minimum viewing distance, and the distance between the sidewalk border and the monitor is the maximum distance. If this were taken into account during the rendering of animation frames, the rendering time could be shortened significantly. There are numerous other similar examples, that can save us a lot of extra rendering time. If the image difference in a distance range should be computed, the algorithm would be a little bit different. It is then possible to compute the same number of rectangles of various sizes, and to apply weighting functions according to the particular viewing distances.