any ideas on faking subsurface scattering in an opaque material?

poopipe

hello, 
I need to fudge something like this together on an opaque material. I'm layering materials so I'm forced to use the opaque shading model

I've tried dicking around with fresnel, light direction and normals etc. and had some success but I can't help wondering if somebody else has already solved this. It doesn't have to be accurate, it just has to be plausible.

The initial use case is a big slab of floating ice - the internal structure is easy, it's the transmission of light through the thin bits / edges etc. that I'm interested in.

So.... has anyone done something like this or read a paper about it they can point me at?  

Obscura

So neither foliage or sss shading works? I didn't know that you need to go something other than opaque in order to get sss working. It would be very strange.

poopipe

As far as I can tell you aren't able to mix shading models in layered materials - which doesn't sound unreasonable -  the upshot being that you can't layer a material that has an ss colour channel with a material that doesn't. 
I don't really want to turn sss on for everything (the places I need it are pretty incidental ) and I don't really want to start creating special case materials if I can avoid it.
 
Because I'm a masochist I'm deliberately not using mesh UVs so cannot use baked masks for thickness etc. 
That just made me wonder if I could use distance fields to give me thickness information... 

Obscura

You could try the Henyey Greenstein phase function with low aniso value. It isn't exposed to the function library so search for it in the engine content. You would still need a thickness map though, I think you can't do anything decent without it.

poopipe

That looks like it's worth a pop. As I say, I reckon I can get a thickness value out of mesh distance fields or failing that just fudge something from  bounding volume. 

Thanks very much