22. ALBEDO CORRECTION FOR RENDERING DOWNSAMPLED HETEROGENEOUS MEDIA
Name: Lifan Wu
Grad Year: 2020
Volumetric micro-appearance models have provided remarkably high-quality renderings, but involve data intensive high-resolution 3D volumes, which usually take several gigabytes of storage. When an object is only viewed from a normal distance, the highest level of detail offered by these models is usually unnecessary. It enables us to use a downsampled version of data, but traditional linear downsampling of model parameters can yield poor accuracy. We present a new approach to optimize single-scattering albedo at downsampled resolutions, enabling accurate lower-resolution rendering of heterogeneous and anisotropic media. The optimized albedo generalizes well to novel configurations, and the resulting data sets offer several orders of magnitude storage savings.
Industry Application Area(s)
Software, Analytics | Graphics, Vision