On 10 Dec 2015, at 14:39, Jackson Smith <[log in to unmask]> wrote:

Hi Saad,

I'm sorry to have had to make you spell it out so plainly. But this has helped my understanding a great deal. Since you flip the angle if the cosine is less than 0, the widest possible spread of new directions is achieved by setting the curvature threshold to 0, no? And that will effectively sample the set of angles bounded within the half of a sphere where the cosine is positive?

Best wishes,
Jackson