An end-to-end geometry encoding format for aerodynamically optimizing and directly manufacturing folded surfaces.
NetFold takes complex 3D aerodynamic shapes, evaluates their fluid dynamics and manufacturing cost, and losslessly unwraps them into a flat 2D net—exporting machine-readable SVGs ready for industrial laser cutting and sheet-metal bending.
It solves the fundamental geometric constraint of Gauss's Theorema Egregium by intelligently slicing compound curves into developable flat patches.
[3D CAD Mesh] ──► [Aero Engine] ──► [NetFold] ──► [Laser Cutter SVG]
- Aero Engine (Physics): Uses a 3D Source Panel Method to calculate pressure distributions, drag, and downforce over the car body.
- Structural Penalty (Weld Length): NetFold calculates the exact linear distance of welding required to physically manufacture the aerodynamic shape, ensuring the Aero Engine never outputs an unbuildable design.
- Lossless Unwrapping: Uses BFS traversal and SAT overlap detection to recursively unwrap and split the 3D geometry into flat 2D patches (islands).
- Machine-Readable Export: Generates completely automated SVG cut files mapped to standard industrial protocols.
NetFold completely automates the translation from abstract 3D math into real-world shop floor instructions.
- Planar Panel Fusion: Automatically detects flat structural regions (like wings or sidepods) and fuses their geometry in the SVG. The 3D physics engine keeps its highly-dense triangle mesh for accurate fluid dynamics, while the laser cutter receives one solid, pristine sheet-metal polygon.
- Physical Bending Math: Calculates the true physical bending angle (
180° - dihedral) from a flat metal sheet, eliminating translation errors for fabricators. - Smooth Curve Filtering: Smart text-scaling filters out manufacturing labels for extremely shallow bends (
< 5°), allowing aerodynamic surfaces to be cleanly slip-rolled without overlapping label instructions. - Multi-Island Sheet Freedom: Complex non-convex geometry (like Formula Student nosecones) are dynamically split into separate, non-overlapping SVG files, giving teams the freedom to laser cut different patches from different sheets of metal.
NetFold serializes geometry into highly compressed binary (.nfb) or human-readable JSON (.netfold).
| Mesh | Triangles | Islands | Weld Length | OBJ Size | Binary Size |
|---|---|---|---|---|---|
| Cube | 12 | 1 | 7.8 units | 0.4 KB | 1.8 KB |
| Icosahedron | 20 | 1 | 6.8 units | 0.6 KB | 2.2 KB |
| Geodesic sphere | 320 | 1 | 23.7 units | 12.9 KB | 12.4 KB |
| FS Nose cone | 720 | 1+ | 29.8 units | 21.0 KB | 27.8 KB |
Binary format is comparable to raw OBJ size while encoding full fold angle and stitch topology.
from core import encode_mesh_multi_island
from core.export_svg import export_svg
from core.aero_3d import run_3d_panel_method
# 1. Calculate Aerodynamics
# pressure_score = run_3d_panel_method("nose_cone.obj")
# 2. Encode Geometry & Calculate Manufacturing Cost
nf = encode_mesh_multi_island("nose_cone.obj", name="nose_cone")
weld_penalty = nf.calculate_weld_length()
print(f"Required welding: {weld_penalty:.2f} meters")
# 3. Export to Laser Cutter
saved_files = export_svg(nf, "nose_cone_pattern.svg")
print(f"Generated CNC patterns: {saved_files}")- Export to SVG / DXF for CNC cutting
- Advanced Fabrication Bending Math
- Adaptive Planar Fusion
- Interactve web viewer (Three.js)
- Support for open meshes (with boundary)
- Optimal island layout (minimise bounding box area)
- Python package (
pip install netfold)
Amulya — Built to bridge the gap between theoretical aerodynamic simulation and real-world sheet metal fabrication.