I've done some preliminary work on the live SDF operations on mesh primitives recently.
It shows promising results and potentials, while also reveals some bottlenecks and unaddressed issues. Some new algorithms are under development, and some tradeoffs are made between performance and quality.
SDF can be always converted to watertight meshes, but not vice versa. Meshes with self-intersection, open holes, T-junction/dangling edges (non-manifold geometry) can cause difficulties in determining inside and outside of the surface (the "Sign" of the "Signed" Distance Field). Therefore, the current experiments are only applied to meshes with proper topologies.
This work belongs to the second milestone on the roadmap, and the current major task is still to finish the first milestone: optimizing and polishing the current live basic primitives and improving the rendering functionalities. There may be some prototype versions to explore the usages and workflows of this feature, but they won't be included in beta versions in the near future.

moat
2025-05-04 19:36:30 +0000 UTCEd Ferrari
2025-04-28 00:57:58 +0000 UTC