Cycles: Small tweaks for Henyey Greenstein closure code.

* Avoid duplicative fabs(g) check in sample code.
* Avoid dot product in eval code.

Helps like ~1% when Scatter Anisotropy is 0.

1 files changed, 9 insertions, 7 deletions

diff --git a/intern/cycles/kernel/closure/volume.h b/intern/cycles/kernel/closure/volume.h
index 9f519e2e2b1..1cf5e17b8f7 100644
--- a/intern/cycles/kernel/closure/volume.h
+++ b/intern/cycles/kernel/closure/volume.h
@@ -26,9 +26,6 @@ CCL_NAMESPACE_BEGIN
  * uniform sphere. g=0 uniform diffuse-like, g=1 close to sharp single ray. */
 ccl_device float single_peaked_henyey_greenstein(float cos_theta, float g)
 {
-	if(fabsf(g) < 1e-3f)
-		return M_1_PI_F * 0.25f;
-	
 	return ((1.0f - g * g) / safe_powf(1.0f + g * g - 2.0f * g * cos_theta, 1.5f)) * (M_1_PI_F * 0.25f);
 };
 
@@ -47,9 +44,13 @@ ccl_device float3 volume_henyey_greenstein_eval_phase(const ShaderClosure *sc, c
 	float g = sc->data0;
 
 	/* note that I points towards the viewer */
-	float cos_theta = dot(-I, omega_in);
-
-	*pdf = single_peaked_henyey_greenstein(cos_theta, g);
+	if(fabsf(g) < 1e-3f) {
+		*pdf = M_1_PI_F * 0.25f;
+	}
+	else {
+		float cos_theta = dot(-I, omega_in);
+		*pdf = single_peaked_henyey_greenstein(cos_theta, g);
+	}
 
 	return make_float3(*pdf, *pdf, *pdf);
 }
@@ -63,10 +64,12 @@ ccl_device int volume_henyey_greenstein_sample(const ShaderClosure *sc, float3 I
 	/* match pdf for small g */
 	if(fabsf(g) < 1e-3f) {
 		cos_theta = (1.0f - 2.0f * randu);
+		*pdf = M_1_PI_F * 0.25f;
 	}
 	else {
 		float k = (1.0f - g * g) / (1.0f - g + 2.0f * g * randu);
 		cos_theta = (1.0f + g * g - k * k) / (2.0f * g);
+		*pdf = single_peaked_henyey_greenstein(cos_theta, g);
 	}
 
 	float sin_theta = safe_sqrtf(1.0f - cos_theta * cos_theta);
@@ -80,7 +83,6 @@ ccl_device int volume_henyey_greenstein_sample(const ShaderClosure *sc, float3 I
 	make_orthonormals(-I, &T, &B);
 	*omega_in = sin_theta * cos_phi * T + sin_theta * sin_phi * B + cos_theta * (-I);
 
-	*pdf = single_peaked_henyey_greenstein(cos_theta, g);
 	*eval = make_float3(*pdf, *pdf, *pdf); /* perfect importance sampling */
 
 #ifdef __RAY_DIFFERENTIALS__