In Ward original paper
(http://radsite.lbl.gov/radiance/papers/sg88/paper.html) the
error_reuse_rotate is
error_reuse_rotate = sqrt(fabs(1.0 - N(P).N(Pi) ));
while in povray it's
error_reuse_rotate = 2.0 * sqrt(fabs(1.0 - N(P).N(Pi) ));

Is there a reason for this factor 2 ? If it's an error it makes povray take
more gather locations than necessary for a given error_bound (I think there
is also a small influence on the weighting of points averaged)

M

Post a reply to this message