I think the way SSS is implemented right now ( lightmap caclulation etc. ) has the side effect that you loose the information of the shadow pass. That means that if you want to isolate only the shadow pass in a render layer to adjust the shadows you won’t get them.
Below is a very simple example showing the effect. The plane is rendered with SSS toggled on/off for the two output viewers.
The only solution I could come up with, is to render the shadow pass into a file with all sss materials turned off. Then I could use this map as input afterwards. The problem with this is, that if you change the lighting, you would have to do this toggling all over again.
Quick fix: Create a new scene with your objects linked. Any change in their location/geometry will be reflected in the new scene. In this new scene change all the materials to a non SSS type. In your comp you can load the shadow from the new scene.
it kinda makes sense, since SSS allows the light to penetrate the surface of the object, and not strictly bounce off the front. Perhaps adjusting the front setting…because I wonder - do ANY SSS materials show shadows, and if so, then it must be a plane material SSS setting issue.
I not sure if any SSS material shows shadows in the shadow pass, but I doubt it. I might be wrong, but as far as I have understood the implemented algorithm, prior to the actual SSS pass a full lighting pass is performed, with all the diffuse and shadow calculation taking place, resulting in a lightmap for the mesh. So the all the light distribution is collapsed and ‘baked’ onto the mesh and using this the actual SSS calculation is done
.
So this would effectively kill the shadow pass.
cart you just render the sss object on a separate layer, have a duplicate mesh with no sss on the normal render layer. then compose the sss over the normal render layer?
