Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=350977

pssst... "Shadow Cams". They define the scene area from which each of the shadow maps is rendered. Often, 90% of the area rendered to your shadow map (and hence 90% of the pixels in your shadow map) is not included in your actual image rendered. Zooming and dollying these cams to cover only the part of the universe in your final render can yield the same effect as increasing the size of the shadow map without gobbling up the additional memory and render time. Manipulating the Shadow Cams also allows you to control other factors affecting the shadow's density & diffusion and edge feathering. Here is a post here with some sample images: http://www.renderosity.com/messages.ez?Form.ShowMessage=350977