[boulos]

I can’t seem to find the PDF of Brent’s siggraph sketch in 2006 (Shadow map bias cones), but our follow up work [1] hopefully makes the connection between distance and r^2 clear: these techniques are all computing some sort of solid angle and filter width from that.

As Raph is mentioning in a side thread, there are lots of assumptions! (Some of them similar to the randomized assumptions behind alpha, generally).

[1] http://graphics.stanford.edu/~boulos/papers/prefilter_rt08.p...