Point Based Modelling and Simulation - Sampling
A low-discrepancy, blue-noise point set represents a continuous geometric object well, minimising discretisation artefacts. In this project we consider low-discrepancy, blue-noise sampling algorithms and theory for triangulated 2-manifolds. Results include efficient sampling algorithms that can operate in real-time scenarios and enable complete control of local sampling density and anisotropy.
We have considered this for applications in rendering, modelling and simulation: anisotropic sample placement applied to non-photorealistic stroke-based rendering; fast, robust mesh sampling for interactive sampling and rendering; improved CVT and ODT generation for meshing and simulation.