Fix wrong factor in EEVEE multi-GGX average color

The original paper is assuming `f90 == 1`. However, it is just an
estimation, and the difference is really subtle.
This commit is contained in:
Weizhen Huang 2023-08-29 13:53:24 +02:00
parent 567e3b9b19
commit 3876133ef5
2 changed files with 2 additions and 2 deletions

View File

@ -89,7 +89,7 @@ vec3 F_brdf_multi_scatter(vec3 f0, vec3 f90, vec2 lut)
float Ess = lut.x + lut.y;
float Ems = 1.0 - Ess;
vec3 Favg = f0 + (1.0 - f0) / 21.0;
vec3 Favg = f0 + (f90 - f0) / 21.0;
/* The original paper uses `FssEss * radiance + Fms*Ems * irradiance`, but
* "A Journey Through Implementing Multiscattering BRDFs and Area Lights" by Steve McAuley

View File

@ -295,7 +295,7 @@ vec3 F_brdf_multi_scatter(vec3 f0, vec3 f90, vec2 lut)
float Ess = lut.x + lut.y;
float Ems = 1.0 - Ess;
vec3 Favg = f0 + (1.0 - f0) / 21.0;
vec3 Favg = f0 + (f90 - f0) / 21.0;
/* The original paper uses `FssEss * radiance + Fms*Ems * irradiance`, but
* "A Journey Through Implementing Multiscattering BRDFs and Area Lights" by Steve McAuley