View more

View more

View more

### Image of the Day Submit

IOTD | Top Screenshots

### The latest, straight to your Inbox.

Subscribe to GameDev.net Direct to receive the latest updates and exclusive content.

# UE4 IBL glsl

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

3 replies to this topic

### #1cifa  Members

Posted 14 July 2014 - 04:02 PM

************** UPDATE **********************

I had a stupid issue due to the resolution of the texture, not the code below works as it should.

***********************************************

Hi there!

Now I'm still trying to get the envBRDF LUT (roughness / NdotV) to render properly as shown in the paper:

What I have now is:

the usual Hammersley functions:


bits = (bits << 16u) | (bits >> 16u);
bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u);
bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u);
bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u);
bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u);
return float(bits) * 2.3283064365386963e-10; // / 0x100000000
}
// http://holger.dammertz.org/stuff/notes_HammersleyOnHemisphere.html
vec2 Hammersley(uint i, uint N) {
}



The ImportanceSampleGGX practically as reported in the paper:

vec3 importanceSampleGGX(vec2 sample, float m, vec3 N){
float phi = sample.x * 2.0f * PI;
float cosTheta = sqrt( (1.0f - sample.y) / 1.0f + (m*m - 1.0f) * sample.y );
float sinTheta = sqrt(1.0f - cosTheta*cosTheta);

vec3 vector = vec3(
sinTheta * cos(phi),
sinTheta * sin(phi),
cosTheta
);

return vector;
vec3 up = abs(N.z) < 0.999 ? vec3(0.0,0.0,1.0) : vec3(1.0,0.0,0.0);
vec3 tangentX = normalize(cross(up,N));
vec3 tangentY = normalize(cross(N,tangentX));
// Project
return tangentX * vector.x + tangentY * vector.y + N * vector.z;
}



The integrateBRDF

 vec2 IntegrateBRDF( float Roughness, float NoV ){
vec3 V;
V.x = sqrt( 1.0f - NoV * NoV ); // sin
V.y = 0.0;
V.z = NoV;
// cos
float A = 0.0;
float B = 0.0;

for( uint i = 0u; i < NUMBER_OF_SAMPLES; i++ )
{
vec2 Xi = Hammersley( i, NUMBER_OF_SAMPLES );
vec3 H = ImportanceSampleGGX( Xi, Roughness, vec3(0.0,0.0,1.0) );
vec3 L = 2.0 * dot( V, H ) * H - V;
float NoL = max( L.z, 0.0 );
float NoH = max( H.z, 0.0 );
float VoH = max( dot( V, H ), 0.0 );

if( NoL > 0 )
{
float G = G_Smith( Roughness, NoV, NoL );
float G_Vis = G * VoH / (NoH * NoV);
float Fc = pow( 1.0 - VoH, 5.0 );
A += (1.0 - Fc) * G_Vis;
B += Fc * G_Vis;
}
}
return vec2(  (A)/float(NUMBER_OF_SAMPLES), (B)/float(NUMBER_OF_SAMPLES)) ;
}



As you can see they're almost a copy-paste from the reference, but what I got is completely different:

The roughness is the uv.x and uv.y of a quad that are calculated in the VS as:

	vTextureCoordinate = (aVertexPosition.xy+vec2(1,1))/2.0;



Am I missin something very stupid?

Edited by cifa, 15 July 2014 - 06:27 AM.

### #2lipsryme  Members

Posted 14 July 2014 - 05:13 PM

He is using Disney's roughness which is squared so that would be m*m inside the importanceSampleGGX function.

My HLSL code:

float3 ImportanceSampleGGX(float2 Xi, float Roughness, float3 N)
{
float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph]

// Compute distribution direction
float Phi = 2 * PI * Xi.x;
float CosTheta = sqrt((1 - Xi.y) / (1 + (a*a - 1) * Xi.y));
float SinTheta = sqrt(1 - CosTheta * CosTheta);

// Convert to spherical direction
float3 H;
H.x = SinTheta * cos(Phi);
H.y = SinTheta * sin(Phi);
H.z = CosTheta;

float3 UpVector = abs(N.z) < 0.999 ? float3(0, 0, 1) : float3(1, 0, 0);
float3 TangentX = normalize(cross(UpVector, N));
float3 TangentY = cross(N, TangentX);

// Tangent to world space
return TangentX * H.x + TangentY * H.y + N * H.z;
}


For completeness sake here's how I do G_Smith:

// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html
float GGX(float NdotV, float a)
{
float k = a / 2;
return NdotV / (NdotV * (1.0f - k) + k);
}

// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html
float G_Smith(float a, float nDotV, float nDotL)
{
return GGX(nDotL, a * a) * GGX(nDotV, a * a);
}


Not sure about the TexCoord code you are using but you just need to render a regular post process quad and simply pass through the TexCoords as they are and then use them as inputs to the integrateBRDF function like so:

return IntegrateBRDF(input.TexCoord.x, input.TexCoord.y);


Probably unrelated but I'm using an R16G16_FLOAT texture like Brian Karis is suggesting.

Edited by lipsryme, 14 July 2014 - 05:19 PM.

### #3REF_Cracker  Members

Posted 02 August 2014 - 11:58 PM

Would you guys be able to share some details on how you're filtering your IBL cubemaps?
What roughness are you using at each mip?
What format do you use?

How do you put it all together?

Thanks!

Check out my project @ www.exitearth.com

### #4lipsryme  Members

Posted 09 August 2014 - 01:13 PM

Hi REF_Cracker. Sorry for the late answer I just happened to stumble upon the thread again (hadn't been following it).

The format I'm using is R16G16B16A16_FLOAT for each cube map face.

I calculate the roughness in the shader like this:

float Roughness = (CubeLOD) / (CubeLODCount);


Which would give you values like these:

0 (256x256)

0.166667 (128x128)

0.333333 (64x64)

0.5 (32x32)

0.666667 (16x16)

0.833333 (8x8)

1 (4x4)

CubeLOD is starting at 256x256 and going down 6 levels (so 7 in total / original + 6 mip levels) to 4x4 for the roughest one.

Maybe you should use less mips and make it end at 16x16 or so haven't done much testing here.

Edited by lipsryme, 09 August 2014 - 01:14 PM.

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.