Geometry, AI & 3D Printing
Go from mesh fundamentals to printable solids, PyTorch3D augmentation, tokenization, diffusion, model routing, texture preservation, and production QA.
Research notes for artists and engineers who want to understand not only which mesh tool to click, but what the tool is solving and how to build better workflows around it.
Go from mesh fundamentals to printable solids, PyTorch3D augmentation, tokenization, diffusion, model routing, texture preservation, and production QA.
Representation contracts, three masters, measurable checkpoints, manufacturing manifests, and choosing GLB, STL, or 3MF for the job.
Gradients, divergence, cotangent Laplacians, mass matrices, curvature flow, Poisson reconstruction, and harmonic coordinates—built from visual intuition upward.
Watertightness, manifoldness, wall thickness, overhangs, orientation, scale, tolerances, and a useful Trimesh inspection script.
Why Laplacian eigenvectors behave like a shape’s natural harmonics—and how that changes smoothing, compression, segmentation, and search.
Safe transforms, sampling, geometric corruption, supervision contracts, deterministic batches, and surface-aware losses—with code.
Coordinate quantization, face ordering, adjacency-aware sequences, MeshGPT, MeshAnything V2, and why token budgets shape topology.
Fixed volumetric scaffolds, signed-distance parameters, diffusion training, marching tetrahedra, conditioning, and production limits.
TRELLIS.2, Hunyuan3D 2.1, mesh tokens, DMTets, candidate filtering, and a practical production router.
A practical chain for turning generated concepts into editable, painted geometry while keeping UVs, material intent, and production provenance intact.
A practical comparison of PBR image-to-3D reconstruction, local RTX 5090 deployment, API pricing, and print-ready workflows.