[Patch] Specular Highlight

[Ian]

Not yet, but it's interesting to note that ocean hunter has a very similar shading problem, but in reverse. In supermodel it there is shading, on the real hw it's gone completely, and the model is only lit by ambient. And this only seemed to happen when a scaling matrix was applied to the model.

[Ian]

Well .. i checked out the size of the model normals and they all came out unit length

Code: Select all

   "fsViewNormal   = gl_NormalMatrix  *gl_Normal;\n"
   "float len = length(fsViewNormal);\n"
   "if(len>1.1) {fsColor = vec4(1,0,0,1);}\n"
   "else if(len<0.9) {fsColor = vec4(0,1,0,1);}\n"
   "else  {fsColor = vec4(0,0,1,1);}\n"

Just sets the colour based upon normal length
>1 = red
<1 = green
1~ = blue

[Ian]

Well I tried just simply clamping the ambient to 0.75



This is the Wii version running under Dolphin



The Wii version no doubt renders with triangles, where as the model3 was using quads with quadratic interpolation, so the shading is a bit different in each version. But it could well be the ambient is clamped at something like 0.75. Not sure if that has any negative effects in other games. Don't recall any other games using an ambient of 1.

[Bart]

This is maddening and I'm amazed at the progress you guys have made

Why would there be an arbitrary clamp like this? It doesn't seem to make sense... Is there anything about the viewport these models are drawn in that might suggest a different clamp value?

[Ian]

I've no idea >_< As always it's possible the logic is wrong .. but i can't see any other reason for this.
Ambient of 1 is a bit of an oddball because it essentially disables regular lighting completely
Will test other games when I have some time to see if anything else set an ambient of 1

[Ian]

What did you think about the lost world bullets?

Actually it seems there's a little highlight on the right half. I get this shade in Supermodel, in line with the arcade:

lw2.jpeg

Well I finally had a chance to look at this. The lost world bullets shouldn't have specular at all because the light shines directly towards you. We are lighting the back faces! Should be simple enough to do a dot product between the light angle/normal to check we aren't lighting the back faces of objects.

Need to have a closer look at Harry's specular code because he clearly got a lot further with this problem than me.

[Bart]

Regarding the 0.75 clamp -- what if the highest bit of the ambient light value is not used at all, but the rest of the bits form a value that ranges from 0 to 0.5? For example, instead of:

Code: Select all

lightingParams[4] = (float) ((vpnode[0x24]>>8)&0xFF) * (1.0f/255.0f);

What if it's:

Code: Select all

lightingParams[4] = (float) ((vpnode[0x24]>>8)&0x7F) * (1.0f/255.0f);

Does this look really wrong? At any rate, I don't think the Scud Race debug menu corroborates it but who knows if even that is to be trusted.

Another thought is -- is it possible that some part of the lighting calculation is actually unclamped and *wraps*? Seems unlikely if floating point values are used internally but if any of the math is done in fixed point integral format, it could be a remote possibility.

[Ian]

I had actually thought maybe only 7 bits could be used? Quite possible. I'll dig around into the SDK to see if I can find anything

[Ian]

This is the SDK function

void __thiscall PRO_Viewport::SetSunIlluminationData(PRO_Viewport *this, float ambient_contribution, float sun_contribution)
{
PRO_Viewport *v3; // esi@1
unsigned __int8 v5; // c0@1
unsigned __int8 v6; // c3@1
double v7; // st7@2
double v8; // st7@5
float v9; // [sp+8h] [bp-Ch]@1
float v10; // [sp+8h] [bp-Ch]@9
float v11; // [sp+Ch] [bp-8h]@1
float v12; // [sp+10h] [bp-4h]@1

v11 = ambient_contribution;
v3 = this;
v12 = sun_contribution;
v9 = sun_contribution + ambient_contribution;
if ( !(v5 | v6) )
{
v7 = ambient_contribution;
v7 = v7 / v9;
v11 = v7;
v8 = sun_contribution;
v8 = v8 / v9;
v12 = v8;
}
v10 = v11;
if ( v11 >= 1.0 )
v10 = 0.99900001;
*((float *)v3 + 65) = v12;
*((_BYTE *)v3 + 82) = (signed __int64)(v10 * 256.0);
PRO_Pingpong_Data_Block::NotifyUpdate(v3);
}

So it's definitely using just 8 bits. Code is also confusing. It seems to force normalise the sunparamaters, so they always add up to one. Unless I am reading it incorrectly.

The manual says this

The default provides an ambient lighting of 0.5 and a sun intensity of 0.5. These values should not add to more than 1.0 or be negative.

[Ian]

What did you think about the lost world bullets?

Actually it seems there's a little highlight on the right half. I get this shade in Supermodel, in line with the arcade:

Well I finally figured out why there is almost no specular on the model3 whilst in ours we had specular. It's because the model normals are not unit length. They are like 0.2 in length instead of 1. So I guess only receive 20% of the usual brightness.
I think it's safe to say we have probably figured out all the corner cases with regards to specular? Especially now we understand what LA machine guns is doing