I'm playing around with the nifty AXYZ sub-surface scattering solution and am not sure I'm understanding part of it. The SSS multi VOP node has a Pointcloud Sample parameter that the documentation describes as specifying the number of point cloud samples to use from within the Scattering Distance. I also inferred that increasing the value will give smoother results, (as opposed to different ones).

In my tests, I'm finding that increasing the samples is, in effect, increasing the radius as well, thus changing the overall character of the subsurface lighting. I'm digging around in the vex code for clues as to what's going on, but am posting this in the hopes that someone will know off-hand what is going on.

Attached are: a hip file, a render with samples set to 500 and 5000. It's the latter one that stood out as odd to me.

thanks, R

Well, I think I have part of the answer. So far as I'm understanding things, the behaviour I'm getting is because of a call to pcfilter at the end of vop_multiSSS defined in voplib.h.

Essentially it is doing a blur on the results of the vop_ssIntegMulti function called before it, but doing that blur based exclusively on the Pointcloud Sample parameter, ignoring any sort of scattering distance.

In other words, if I make the number equal to the number of points in the point cloud, I will get something almost uniform (depending on the filtering kernel, which is unknown to me). A test I just ran seems to support that theory.

Anyway, I think I understand it better now.

In case you haven't noticed, there's a somewhat technical companion thread at odforce here:
http://forums.odforce.net/index.php?showtopic=2262 [forums.odforce.net]