Material Vops To Have Lighting Model Vops Inside?

[old school]

Pretty much all of the surface VOPs in the Material folder in a surface VOP network contain Lighting Model VOPs. Fine if you use one of these as the basis for the entire material that you are building.

The downfall is if you start to mix two or more material VOPs together with say a Color Mix VOP. You are effectively calling the illuminance loops on all the lights illuminanting the surface two times. Mix in a third material and you increase this to three times. Add more materials and mix in at your peril. Start to build better illuminance loop VOPs that generate deep rasters (AOV's) with intelligent ligth message passing and it get's real complicated real fast.

I have seen some Slim shader networks that contain several illuminance loops unbenounced to the slim artist several times in the past. Isn't this a recipe for disaster due to incresed render times and pipeline complexity or am I way off in my assumption.

The big question is:

Should we be mixing color fed in to a single lighting model VOP or be mixing the result of two or more ighting models?

I am sure some of you will see where this is leading.

Is this an issue that a VOP shader writer artist needs to worry about?

[Wolfwood]

It comes down to flexibility/ease vs. speed. (As always)

Most shaders have at least two illuminance loops already (diffuse() & specular()). The diffuse and specular calculations can easily be done in the same loop and it is a little bit faster.

If I was doing a short film at home where I only have one or two computers and I was the only one who had to worry about the shader code I would probably use the multiple illuminance approach until I was really happy with the look of everything, then I would go in and do a major optimize pass. Like get rid of some parameters, combine illuminance loops etc. But I would only do that if it was just me and I was my own director.

Otherwise lots of mixing different functions with illuminance loops like a specular() diffuse() clay() etc etc in a studio environment isn't the most efficient way when it comes to CPU time but its very efficient when it comes to user time.

surface
singleLoop()
{
vector Nn = normalize(N);
vector In = -normalize(I);
vector C = 0;
illuminance(P,Nn,1.5708) {
   vector Ln = normalize(L);
   vector H = normalize(Ln+In);
   C += Cl*(dot(Nn,Ln) + pow(max(0,dot(Nn,H)),1/.1));
}

Cf = C;

}

Render Time: 1:06.15u 1.02s 1:07.72r
Function              Excl.   Incl.       Calls  Instructions   Instr.   Local
                       Secs    Secs                           Per Call Storage
------------------------------------------------------------------------------
opdef:/Shop/singleLoop    9.07   10.57  29,404,617   582,837,913    19.82  20,556
opdef:/Shop/v_asadlight    1.50    1.50  29,404,617   264,641,553     9.00   3,096
------------------------------------------------------------------------------
     Invocations Instruction
     259,690,261 set
     112,667,176 normalize
     110,191,530 add
      56,333,588 mul
      56,333,588 nlight
      53,857,942 dot
      29,404,617 blight
      29,404,617 div
      29,404,617 length
      29,404,617 neg
      26,928,971 jump
      26,928,971 max
      26,928,971 pow

surface
twoLoop()
{
vector Nn = normalize(N);
vector In = -normalize(I);
vector C = 0;
illuminance(P,Nn,1.5708) {
   vector Ln = normalize(L);
   C += Cl * dot(Nn,Ln);
}

illuminance(P,Nn,1.5708) {
   Ln = normalize(L);
   vector H = normalize(Ln+In);
   C += Cl * pow(max(0,dot(Nn,H)),1/.1);
}


Cf = C;

}

Render Time: 1:07.07u 1.00s 1:08.31r
Function              Excl.   Incl.       Calls  Instructions   Instr.   Local
                       Secs    Secs                           Per Call Storage
------------------------------------------------------------------------------
opdef:/Shop/twoLoop   10.51   12.04  29,404,617   776,292,002    26.40  19,528
opdef:/Shop/v_asadlight    1.52    1.52  29,404,617   264,641,553     9.00   3,096
------------------------------------------------------------------------------
     Invocations Instruction
     286,619,232 set
     139,596,147 normalize
     112,667,176 nlight
     110,191,530 add
      83,262,559 mul
      58,809,234 blight
      53,857,942 dot
      53,857,942 jump
      29,404,617 div
      29,404,617 length
      29,404,617 neg
      26,928,971 max
      26,928,971 pow