Confusion over Microfacet-based BRDFs and Normal Distribution Functions | pressku.com

Trending 3 months ago

Through reference various sources online, I've go a spot confused. I'll concisely outline what I deliberation I do understand:

My knowing is that microfacet-based BRDFs presume location is ever a cleanable reflection, but nan existent microscopic normal changes according to immoderate Normal Distribution Function (NDF), resulting successful a non-perfect macroscopic reflection. When ray tracing, you sample this NDF to make a "true" normal, and past you tin compute a bounce ray utilizing your incident ray, and a specular reflection pinch your "true" normal.

My disorder comes from a fewer things:

  1. What is nan logic that immoderate microfacet-based BRDFs are diffuse only (such arsenic Oren-Nayar) aliases specular only (such arsenic Cook-Torrance)? My presumption would beryllium that if a roughness worth of 0 were used, each of nan normals should align and frankincense you would get a purely specular reflection. Similarly, if a roughness worth of 1 was used, nan normals would beryllium each scattered, and you'd get a much diffuse reflection. Yet, Wikipedia states for nan Oren-Nayar model: "In nan lawsuit of $\sigma = 0$ (i.e., each facets successful nan aforesaid plane), and frankincense nan Oren-Nayar exemplary simplifies to nan Lambertian model" If each of nan microfacets align to nan aforesaid plane, really could nan computed reflection still beryllium diffuse? What happens erstwhile I sample a bounce ray, arsenic surely if nan normals are each aligned, nan sampled ray should beryllium specular?

  2. If Oren-Nayar is simply a BRDF, and Cook-Torrance is simply a BRDF, but you request some to afloat correspond a azygous material's diffuse and specular components, is nan operation of them besides a BRDF? Is combining elemental BRDFs that seizure only a azygous facet of reflection really you create a afloat PBR BRDF? (Looking astatine nan OpenPBR white-paper, it seems that a azygous worldly is defined arsenic a mixing of aggregate simpler BRDFs)

  3. If you require aggregate BRDFs to picture a azygous material, do they stock nan aforesaid NDF? When way tracing, really do you sample a bounce ray? If you are ever simply performing a specular reflection connected immoderate sampled normal, does this mean nan BRDF cannot beryllium value sampled?

I deliberation I americium getting confused by nan basal quality betwixt a BRDF and an NDF, and nan roles they play.

More
close