ShadowPass Node

This is EZ, I think :)

1) It's a node. Nothing scary, just a node.

2) It ONLY can send its output directly into an input into the Main Node. Nowhere else.

3) It receives exactly TWO inputs of the same type. Nothing scary there either.

4) The first input will be called "MAIN PASS". How cool is that?

5) The second input will be called "SHADOW PASS". -- I'm running out of lame comments.

6) All the logic requires to compute the "MAIN PASS" input will be put into... the main pass.

7) All the logic required to compute the "SHADOW PASS" input will be put into the shadow pass.

So, for example, assuming this node is connected to the alpha clip input of the Main Node... then the main pass will do:

float MAIN_PASS = blah * blah * blah; //<-- all the logic of the first input.

clip(MAIN_PASS);

And the shadow pass would have something like...

float SHADOW_PASS = bleh * bleh * bleh; //<-- all the logic of the second input.

clip(SHADOW_PASS);

So assuming someone, like me, wants to dither the alpha clip of the main pass but not do the same on the shadow pass.. this node would like to say hello. :)

The only slightly complex thing would be if the user has any GET/SET nodes in their tree. And say, they do a SET at the end of their main pass logic, and then do a GET on that variable at the beginning of their shadowpass logic lol. But if they do that, then they are just stupid. Or smart, how should I know.

But either way, it shouldn't prevent shader forge from... creating the shader. It would simply have to include all their main pass logic necessary to compute that GET into the shadow pass, plus all their shadow pass logic. I assume shader forge already knows how to figure out what logic is required to compute ANY variable in the tree... so you should basically already have all the code necessary to create this new node?