Algorithmic Constraints

Our basic goal is to get the perceptual benefits of soft shadows without significantly increasing the runtime compared to Whitted-style shadow ray testing. This goal is achievable if some accuracy is sacrificed. However, to be both convincing and fast, approximate shadows must have three basic characteristics:

• Only one sample should be used per pixel/light.
• Shadow penumbra width should behave in a believable way, starting at zero at the occluder and increasing linearly with distance from the occluder.
• The algorithm should be visually smooth for both static and dynamic scenes.

It is generally accepted that it is hard for observers to tell the difference between shadows cast by differently shaped lights. For this reason we assume roughly spherical lights. We do a rough calculation at each illuminated point of what fraction s of the light is visible, and attenuate the unshadowed illumination by s. Thus our goal is to estimate s in way that is efficient and is consistent with the three requirements above.