Many-Worlds Inverse Rendering
This addresses a bottleneck in inverse rendering for computer graphics and vision applications, but appears incremental as it builds on prior methods for visibility sampling.
The paper tackles the problem of discontinuous visibility changes in physically-based inverse rendering by introducing a many-worlds representation that differentiates volumetric perturbations of surfaces, leading to a simpler and more efficient Monte Carlo algorithm with rapid convergence in iteration count and cost per iteration.
Discontinuous visibility changes remain a major bottleneck when optimizing surfaces within a physically-based inverse renderer. Many previous works have proposed sophisticated algorithms and data structures to sample visibility silhouettes more efficiently. Our work presents another solution: instead of differentiating a tentative surface locally, we differentiate a volumetric perturbation of a surface. We refer this as a many-worlds representation because it models a non-interacting superposition of conflicting explanations (worlds) of the input dataset. Each world is optically isolated from others, leading to a new transport law that distinguishes our method from prior work based on exponential random media. The resulting Monte Carlo algorithm is simpler and more efficient than prior methods. We demonstrate that our method promotes rapid convergence, both in terms of the total iteration count and the cost per iteration.