Accepting Light From The Back Of An Object

[LPXO]

Hi,

Im trying to build an illuminance loop in VEX based on user jasons "thin" shader in the codex (http://www.odforce.net/codex/?page=3&sec=all&catlimit=&sorty=&limit=5) to allow my shader to accept light form behind the surface.

So im replicating this line in an illuminance loop:

dclr += - ( Cl * dot(Nf, Ll) * trans_factor * translucency * diff);

Ive attached a screenshot of my illuminance loop.

Im not getting any kind of result from this. Can anyone give me a hint as to what im missing?

Thanks.

[Jason]

Hey there,

The easiest way to to this is:

N --- Normalize +--- Lighting Model ----------------+
                 |                                   |
                 +--- Negate -----  Lighting Model --+ Add --- Output

Basically, add in lighting based on a flipped (negated) surface normal.

[LPXO]

Thanks Jason. Id already tried that but wasn't quite getting the effect i was after. I was hoping to soften the light on the surface of my object (thin membrane) by making it have a subtle diffuse "glow" that appeared to come from behind. Since i wasnt sure the best approach for this I was trying a few different things. Ill play around with using negated normals and see if i can get a more desirable effect. Thanks for the reply.

[Jason]

Sounds like you're talking about Subsurface Scattering - have you tried Mario's SS shaders?

[LPXO]

Yeah, thats exactly what Im talking about except since it's for an assignment I need to write the shader myself! Ive been trying to cheat it and to be honest the best results ive got are rendering off what's behind my membrane as a seprate layer and doing the translucency in Shake.

I have read through Mario's SSS thread and checked out the shaders and it is really cool.

[Jason]

Ah, I see.

You can do a really interesting faking of sub-surface diffuse scattering by baking off the illumination and blurring it and then re-applying it to the surface. There are several caveats for this though: *1* it won't help the thin-film case - since the blurring happens in texture uv space - and *2* it requires nice uv's which approximate real-world distances.

I might be able to find an example of this for you if you like - you can acutally do this at interactive speeds...

This said, it's not going to solve the single-scattering case (thin-film like) for you, but the concept might give you ideas.

[LPXO]

Yeah thats another area I looked into (I think I found that on the Wiki) but didn't make progress on testing it fast enough with my deadline looming so I had to try other methods. Even with the drawbacks I was still interested to see the results. If you have an example somewhere Id love to see it.

Thanks.