It's incredible that the ambient occlusion in this image was only done by using pixel depth buffers. It seems that just knowing the depth of the nearest object doesn't tell us enough information about where to cast shadows, but it works pretty well. Why would we ever need to actually test for ambient occlusion with ray-tracing, or use an occlusion texture/map, when we can capture the effect pretty well in screen-space?
It's incredible that the ambient occlusion in this image was only done by using pixel depth buffers. It seems that just knowing the depth of the nearest object doesn't tell us enough information about where to cast shadows, but it works pretty well. Why would we ever need to actually test for ambient occlusion with ray-tracing, or use an occlusion texture/map, when we can capture the effect pretty well in screen-space?