Hi Matthieu,

I can think of two possibilities right away that might be causing you an issue. First, are you shrinking the target size along with your detector. If you have a large target sphere for a small target object, you might be aiming rays roughly towards the target but still missing the actual detector in the middle.

Second, and probably more likely, when you use importance sampling the system has to scale the intensity downward for each ray to make up for the fact that a larger than normal number of rays are going towards this small selected region. As it happens, this can mean that you are triggering the minimum ray intensity threshold as set in the System Explorer. Try using a lower value and see if the rays manage to find the detector again.

If neither of these are the case, do you have a representative sample system that you’d be willing to share with the community?

yes, I was reducing the size of the sphere as well.

It was indeed an issue with the relative intensity.

Would setting this relative intensity to 1/(number of source rays * number of scattered rays) be a general good rule of thumbs ?

Matt

I don’t know that there is a specific rule of thumb that has been worked out, but you would want to consider the number of scattered rays and the cross sectional area of your target. The number of source rays isn’t really an issue here because that setting is relative to the starting power of each ray. So if you initially have 1 Watt and a million rays, that relative measures is starting with the one millionth of a Watt. 

The other major factor besides number of scatters is the cross sectional area of your target. As it shrinks, the scattered rays used in importance sampling are reduced further in power, and this will go roughly as the square of the radius reduction. So if you lose your rays reducing the target radius by an order of magnitude, you’ll probably need to change the threshold by two orders of magnitude.