Hi ! I was wondering if anyone tried to implement the Dual Scattering Approximation for Fast Multiple Scattering in Hair paper, this year at Siggraph. I contacted the author about several issues I had (in particular there is a typo in Eq.25 in the integration domain, which is from phi-pi/2 to phi+pi/2), but I have now more and more and I do not want to harrass them. I have now the 3 following problems: - In Eq.6, Omega_f is defined as the front hemisphere, thus depending on the incoming light direction. In this way, the integrals should be switched so that the integration domain can vary with the incident angle. If there is no typo here, then I do not really see what is Omega_f (which should then be a constant integration domain). - In this same equation, I do not see anything which prevents the attenuation to be greater than 1. In particular, even energy conservation (which is expressed with a cos(theta_out) instead of cos(theta_in) like in this formula) can prevent this to be greater than 1. This makes the equation 5 diverge with increasing number of bounces. - Also, at normal incident angles where there should be most forward scattering, my numerical evaluation of the Eq6 decreases to almost 0 (I might have make a mistake in my code... but just to be sure...) Did someone implement this method without troubles ? Thank you very much.