STL Generation for Network Data

Convert node-and-edge networks into 3D-printable STL meshes. The project supports planar extrusion for 2D networks, cylindrical beams for 2D or 3D networks, variable beam thickness, node diameters, multiple materials, and HTML previews.

Quick Start

git clone https://github.com/DMREF-networks/STL_generation_3d.git
cd STL_generation_3d
python -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt

On Windows, activate the environment with:

.\.venv\Scripts\activate

Generate the included 2D example with the planar method:

python generate_2d_stl.py

When asked for the input directory, enter sample_delaunay_2d.

Use cylindrical beams instead with:

python generate_3d_stl.py

Geometry Choices

Network dimensionality and meshing method are separate choices:

• Network dimensionality describes the input node positions: (x, y) or (x, y, z).
• Meshing method describes the solid geometry placed around those nodes and edges.

Planar method · 2D network	Cylindrical method · 2D network	Cylindrical method · 3D network

Planar Method

The planar method is an extrusion process:

1. Each edge becomes a rectangle and each connected node becomes a disc in the XY plane.
2. The overlapping 2D shapes are merged into one outline.
3. That outline is extruded along Z to the requested height.

The result is a 3D STL, but the source network must be flat. Use generate_2d_stl.py for this method.

Cylindrical Method

The cylindrical method creates a cylinder for each edge and a sphere at each connected node. It accepts either kind of input:

• With 2D positions, the beam centerlines remain in one plane.
• With 3D positions, the network extends through X, Y, and Z.

Use generate_3d_stl.py for this method.

Input positions	Available methods
2D (x, y)	Planar or cylindrical
3D (x, y, z)	Cylindrical only

Input Files

Every network needs:

• node positions (xy)
• connectivity (adj)
• optional per-node diameters (node_diameters)

The interactive scripts expect a directory containing a matched pair, not an individual file:

network_xy.csv
network_adj.csv

NumPy files use the same pattern:

network_xy.npy
network_adj.npy

Suffixes are allowed as long as they match, such as network_xy_0.1.npy and network_adj_0.1.npy.

Position arrays may have two columns (x, y) or three columns (x, y, z). Two-column positions are placed at z = 0 automatically.

Connectivity may be a square adjacency matrix or an edge list. Config-driven edge lists use one of these explicit layouts:

edge_list_interpretation	Columns
legacy	source, target[, ignored_length]
thickness	source, target, thickness
material	source, target, material_code
thickness_material	source, target, thickness, material_code

CSV, NumPy .npy, and explicit-schema pickle .pkl inputs are supported by the config-driven workflow.

Ways to Run

Browser UI

python material_stl_ui.py

The browser UI provides file pickers and controls for geometry, thickness, node sizing, material assignment, and output location. Click Generate STLs to create the meshes. Each run also saves a reusable JSON config next to its outputs.

If a file picker is unavailable, paste the path into the field.

Simple Interactive Scripts

python generate_2d_stl.py  # planar extrusion; 2D input only
python generate_3d_stl.py  # cylindrical beams; 2D or 3D input

Both scripts ask for the input type, input directory, beam diameter, model side length, and whether connectivity values control beam thickness. The planar script also asks for extrusion depth. STL and HTML files are written to the current directory.

The older combined prompt remains available as:

python npyToSTLScript.py

Reusable JSON Configs

Use configs for repeatable jobs, edge lists, node diameters, or multiple materials:

python config_to_stl.py sample_configs/multimaterial_test.json

Config paths are resolved relative to the JSON file. See sample_configs/multimaterial_test.json for a compact example.

STL does not reliably store material metadata. Multi-material jobs therefore write one STL per material in the same coordinate frame, ready to import together into a slicer.

To build a config without running Python, open material_config_builder.html in a browser. The builder downloads JSON; run that config later with config_to_stl.py.

Important Controls

• Beam diameter sets the default edge width or diameter in millimeters.
• Model side length scales the longest coordinate span to the requested size.
• Variable thickness applies connectivity weights as edge_diameter = beam_diameter × edge_value.
• Extrusion height sets the Z thickness of planar output.
• Node diameters optionally replace automatic junction sizing with one absolute diameter per node.
• Junction material can be fixed, separated for mixed-material nodes, or assigned by the dominant connected material.

By default, a junction matches the thickest beam touching that node. Isolated nodes do not produce junction geometry.

Examples

• sample_delaunay_2d/ — small input for the interactive scripts.
• sample_configs/multimaterial_test.json — adjacency-matrix material assignment.
• sample_configs/delaunay_centroidal_stiffness_practice/ — edge material codes and node-diameter practice.
• sample_configs/voronoi_random_material_demo/ — deterministic two-material Voronoi network.
• sample_configs/huppi_periodic_a05_demo/ — periodic Gabriel, Delaunay, Delaunay-centroidal, and Voronoi examples.

Keep personal inputs and generated files in local_configs/ and local_output/. Both directories are ignored by Git.

Checks and Cleanup

Run the regression tests with:

python -m unittest discover

Remove generated caches and older byproducts with ./clean.sh on Linux/macOS or clean.bat on Windows.