A graphics dev took a swing at encoding rendering signals - radiance, irradiance, depth, AO, BRDFs - using tight MLPs in HLSL. They benchmarked size, storage, and runtime cost.
Turns out, MLPs beat L2 spherical harmonics for packing radiance. But they stumble on irradiance and specular BRDFs. Bring in Rusinkiewicz reparameterization, and the MLPs find their footing.
