I've noticed in SRTT and SRIV you use 2 different formats for the normal map texture.
You have obviously swizzled the normal vectors and thrown away the z vector (as it can be derived using the Pythagoras' theorem).
Do you use 2 different shaders for each normal map type or do you use the so called Capcom normal map hack?
i.e.
Option 1 (2 different shaders):
Shader 1 (low quality normals using DXT1):
x=r, y=g, z=1-sqrt(r*r+g*g)
Shader 2 (high quality normals using DXT5):
x=a, y=g, z=1-sqrt(a*a+g*g)
or Option 2 (The Capcom hack):
x=r, y=g, z=1-sqrt(r*r*a*a+g*g)
(For low quality DXT1 normal r=x vector and a=1.0. For high quality DTX5 r=1.0 and a=x vector).
If you use the Capcom hack then the 2 normal quality methods will be interchangeable.
You have obviously swizzled the normal vectors and thrown away the z vector (as it can be derived using the Pythagoras' theorem).
Do you use 2 different shaders for each normal map type or do you use the so called Capcom normal map hack?
i.e.
Option 1 (2 different shaders):
Shader 1 (low quality normals using DXT1):
x=r, y=g, z=1-sqrt(r*r+g*g)
Shader 2 (high quality normals using DXT5):
x=a, y=g, z=1-sqrt(a*a+g*g)
or Option 2 (The Capcom hack):
x=r, y=g, z=1-sqrt(r*r*a*a+g*g)
(For low quality DXT1 normal r=x vector and a=1.0. For high quality DTX5 r=1.0 and a=x vector).
If you use the Capcom hack then the 2 normal quality methods will be interchangeable.
Last edited: