Skeletal-Anchored Dual Harmonics for Structured 3D Modeling

Skeletal-Anchored Dual Harmonics (SADH) is a new representation for structured 3D modeling that couples local surface geometry with an internal meso-skeleton. Instead of treating a shape as a cloud of independent primitives, it builds compact surface patches rooted on anchors optimized inside the object volume.

Each patch uses a dual-channel spherical harmonic formulation: one channel captures local radial geometry, while the other defines adaptive patch support through a generalized viewing cone. The paper’s main claim is that this gives anisotropic, direction-aware detail without falling back to isotropic primitives like medial spheres or Gaussian kernels.

The pipeline starts from unorganized point clouds and jointly optimizes surface geometry, anchor positions, patch orientations, and structural connectivity in stages. A geodesic anchor graph is used to preserve neighborhood relationships, which is important if you care about keeping the resulting structure coherent rather than just visually plausible.

For game devs, the practical angle is reconstruction and editing of complex shapes from sparse or messy data. If this approach holds up outside the paper, it could be useful for scanning-to-asset workflows, shape compression, and tools that need both surface fidelity and an interpretable internal structure. The paper reports accurate reconstruction and compact skeletal organization across a range of geometries, but it’s still an arXiv submission, so the real question is how robust and fast it is in production-like cases.