Theoretical computer graphics - Ray casting algorithm

I was studying for a Computer Graphics exam using previous iterations and I came across this question

Comment the following:

"The ray-casting algorithm is fine for calculating projected shadows, yet is computationally very expensive when running a second time for visibility calculations".

Can anyone clarify why it is "computationally very expensive when running a second time for visibility calcu开发者_如何转开发lations"? Sorry for repeating the quote.

Thanks all.

My guess is that it refers to the fact that shadows can easily (more or less) calculated from the vertices (corners) of the silhouette while visibility requires the checking of all possible rays of light that could come bouncing from the object (not just its edges) into the camera.

So the number of rays for shadows is much smaller than the number of rays to determine visibility.

Does that make sense? I could draw a picture...