Researchers at LIRIS (CNRS/INSA Lyon/Université Lyon 1) have developed a framework for performing geometric computations directly on probabilistic representations of 3D shapes. Using Gaussian Process Implicit Surfaces and the Kac–Rice formula, their approach accounts for noise and missing data without having to arbitrarily reconstruct a single surface. The method provides a principled way to compute geometric quantities such as gradients, divergences, and Laplacians under uncertainty, and was recognized with a Best Paper Honorable Mention at ACM SIGGRAPH 2026.