PyAMorph (formerly pySdf)

Overview

PyAMorph (the successor to pySdf, developed as part of a Research Internship at IIT Madras) is a library for 2D and 3D signed distance functions: analytic and image-derived geometry, stochastic microstructure generation, morphometric characterization, and FFT homogenization of effective properties — NumPy/CuPy transparent, so the same code runs on CPU or GPU.

• Geometry & CSG — 40 2D shape classes and 6 exact 3D primitives with boolean operators; CSG-tree/JSON export; zonal levelsets with material labeling; vectorized fast-sweeping reinitialization.
• Microstructure generators — fibrous media, TPMS metamaterials (gyroid, Schwarz P/D, Neovius), spinodal structures, packed spheres, foams, polycrystals, woven fabrics, and propellant grains with exact burnback.
• Effective properties — FFT-based conductivity, elasticity, tortuosity, and Stokes permeability homogenization.
• Characterization — per-object morphometry, structure-tensor orientation fields, two-point correlation, chord-length distributions.
• Data ingestion — Chan-Vese segmentation for image/volume → SDF, a curated CT-volume registry, and STL → SDF conversion.
• Selected for oral presentation at INCAM 2026 (IIT Kanpur), with the work to be published in a Scopus-indexed journal.

Live demo: pyamorph.vercel.app