Hmm … a quick search for: subsurface scattering single scatter gives me this link
Path Traced Subsurface Scattering using Anisotropic Phase … - Pixar paper:
An important distinction of diffusion models is that they only model multiple scattering. We generally
distinguish between scatter events at different path depths: Zero-scatter events are photons that enter
a surface and immediately reach an exit point without having scattered in the medium itself. Single scatter events do the same with a single interaction inside the volume. Multiple scatter events have two or more scatter events before exiting through the surface of the object. Diffusion models only capture the multiple scattering contribution, leaving zero and single scatter events to be treated explicitly as separate BxDF lobes.
I don’ beliece so because:
Pixar distinguish between zero, one and multiple-time scatter events in the computation of SSS because in the tutorial they say:
RenderMan gives the user lots of flexibility by providing 4 different subsurface models, plus 1 single scatter model.
then this single scatter model just stops if it goes multi… so it is simpler/faster… but they also say:
The use of Single Scatter is very important for thin surfaces, but might make surfaces appear too gummy, so we’ve decided to modulate the effect manually with the same DMFP map used in the Subsurface model.
They seem to somehow got the info when it is zero, one or mutli-event/scatter…
So they do compute full SSS but use the info for 0, 1, n scatter events to make a better result… but then also control it manually… This somekinf of puzzles me:
- stopping SSS computaion if more then one events occurs would make it faster/simpler but
- using the info of the numeber of scatter events requires a full computation…
So this doesn’t make it faster but somekind of better (or use it only on thin surfaces)? But also need a manual adjustment ( → time).
Hmm blender (cycles) doesn’t (yet) seem to give your access for the info which kind of scatter event it detects while computing the SSS…
I’m not sure if such an info can be produced by comparing different values of scale or radius of a Subsurfac Scattering node. Something like: scatters already with almost same value with smaler radius so does it mutliple times… (this of course wouldn’t make it faster… more the contrary)
On the other hand: more (multi) scatter events would add value to the Transmitted Scatter Distribution (BTDF) and so top the Specular Transmission (see WikiPedia: BSDF) and so the suggestion of @piranha4D for the video using translucency or rather AO to got this info seems plausible.