MSoften

MSoften is a mesh smoothing technique used in digital geometry processing to reduce surface roughness on polygonal meshes while preserving salient geometric features. It is designed for post-processing data from 3D scanners, CAD exports, and other mesh sources where noise and micro-roughness impede downstream use such as visualization, measurement, or simulation.

The concept emerged in the early 2010s within the computer graphics and computational geometry communities as an alternative to standard Laplacian smoothing. Various research groups and software developers released implementations that combined feature preservation with controlled smoothing. The term MSoften reflects the method’s emphasis on softening noise without erasing sharp edges.

MSoften typically applies an iterative, weighted update to vertex positions. It uses a curvature-aware weighting scheme and an edge-length metric to steer vertex displacement, coupling a diffusion-like term with a ridge retention component. Some implementations offer optional bilateral filtering or adaptive neighborhood selection to further protect corners and creases. Output meshes remain manifold if input is manifold and usually retain original topology.

Applications: MSoften is used in reverse engineering, 3D scanning cleanup, medical surface visualization, and game or film asset preparation. It commonly supports common formats such as OBJ, PLY, and STL and is integrated into several open-source suites and commercial modeling tools.

Limitations and considerations: The balance between smoothing strength and feature preservation requires tuning. Over-smoothing can blur important detail, while under-smoothing may leave noise. Computational cost scales with mesh size, though many implementations optimize with parallel processing or GPU acceleration.

Related techniques include Laplacian smoothing, Taubin smoothing, and bilateral mesh smoothing.