Alpha Invariance: On Inverse Scaling Between Distance and Volume Density in Neural Radiance Fields
This addresses a technical issue in 3D scene reconstruction for computer vision and graphics researchers, but it is incremental as it builds on existing NeRF models.
The paper tackles the problem of scale-ambiguity in neural radiance fields, where volumetric densities vary inversely with scene size, and proposes a method to improve alpha invariance, resulting in more robust behavior compared to existing heuristics.
Scale-ambiguity in 3D scene dimensions leads to magnitude-ambiguity of volumetric densities in neural radiance fields, i.e., the densities double when scene size is halved, and vice versa. We call this property alpha invariance. For NeRFs to better maintain alpha invariance, we recommend 1) parameterizing both distance and volume densities in log space, and 2) a discretization-agnostic initialization strategy to guarantee high ray transmittance. We revisit a few popular radiance field models and find that these systems use various heuristics to deal with issues arising from scene scaling. We test their behaviors and show our recipe to be more robust.