This is what I’m talking about:
// We have high frequency directional data but low frequency spatial data in the envmap.
// We have high frequency spatial data but low frequency directional data in the lightmap.
// So, we combine the two for the best of both. This is done by removing the low spatial frequencies from the envmap and replacing them with the lightmap data.
// This is only done with luma so as to not get odd color shifting.
// Note: make sure this matches the lightmap mixing done for translucency (BasePassPixelShader.usf)
//Sample.rgb *= IndirectIrradiance;
I just changed it using the source, but I’m wondering why the devs think this is a good idea…
For example, let’s say you a cube under a bridge where very little indirect lighting hits it which will result in a dark lightmap.
However that cube should reflect the sky in full brightness even though it doesn’t get a lot of indirect lighting.
Another thing is that the docs say: “The cubemap reflection is mixed together with the lightmap indirect specular based on how rough the material is”
However, the code does not take roughness into account…