https://github.com/stevenygd/PointFlow

Key what operation/network component is invertible for 3d shapes?

The normalizing flow has been generalized from a discrete sequence to a continuous transformation [16, 5] by defining the transformation f using a continuous-time dynamic ∂y(t) ∂t = f(y(t), t), where f is a neural network that has an unrestricted architecture.

A black-box ordinary differential equation (ODE) solver can been applied to solve the problem.

PointFlow: 3D Point Cloud Generation with Continuous Normalizing Flows, iccv19