That's odd. I haven't needed to scale up the parallax factor in mine. I basically just do the following:
float2 baseUV = uv0 * baseTiling + baseOffset; float2 parallaxOffset = pom(height, heightmap, baseUV, viewDirTangent); baseUV += parallaxOffset; parallaxOffset /= baseTiling; float2 detailUV = (uv0 + parallaxOffset) * detailTiling + detailOffset;
Am I missing something?
For clarification, I am using the following POM implementation:
A Closer Look at Parallax Occlusion Mapping