Open pca006132 opened 5 months ago
It might be worth observing previous attempts at Mesh Offsetting:
My Convex Decomposition PoC (Slow, Unclean Topology, Offset is just Minkowski Sum): https://github.com/elalish/manifold/pull/663
Emmet’s Pure Offsetting PoC (Slow, Rough Seams): https://github.com/elalish/manifold/pull/668
My Pure Offsetting Attempt (Slow, Smooth Seams, Brittle): https://github.com/elalish/manifold/pull/669
Minkowski Sums (Very Slow, Smooth Seams, Robust): https://github.com/elalish/manifold/pull/666
Ideally the offsetting solution should be fast, with smooth seams, and robust to multiple repeated applications. 😄
I’m still actively thinking about my Convex Decomposition algorithm; the next step will be to add merging between the Voronoi cells to simplify the resulting structure when possible 🤔
@hyunjinku they are using voxels for offset, which works but have drawbacks, e.g. losing fine details if the resolution is not high enough.
Hi @zalo @pca006132 are there any plans already available for this project or do we need to come up ourselves with some new ideas for the algorithm ?
there are some ideas, but we don't know how well they will work, and this can be quite challenging to implement
Oh okay I understand
Outline
Implement efficient 3D mesh offset, instead of using minkowski sum with high resolution spheres. (https://github.com/elalish/manifold/issues/192)
Details
3D mesh offset is a useful feature that many users asked for, but is difficult to implement efficiently. Many users use minkowski sum with sphere to perform positive offset, but this can be very slow due to the need for exact convex decomposition.
Our approach will only work for positive offset, negative offset can be implemented by performing additional mesh boolean operations, so this is not an issue. The approach has four phases:
Expected Outcome
A fast 3D mesh decomposition algorithm!
Project Properties
Skills
Difficulty
Size
Additional Information