The visualization of large volumes of information requires mechanisms to reduce the amount of information to be displayed. In this paper, we propose a technique to combine the usage of filters (like the degree-of-interest function) with different, scale-dependent representations. This eliminates the need to create new representations or to use graphical attributes to show the degree of interest of each object. This approach is specially useful when the number of variables is large or the information density is too high. The method can be complemented with zoom mechanisms and distortion of the position of the objects in space, taking advantage of the reduction of the area occupied by those objects with less interest to increase the area available for those with a higher degree of interest, normally located near the focus.