Research · Representing physical intent

Programmable lattice design

Open-source Python tools for building conformal, graded, and multi-material lattices that can be scripted, simulated, and printed.

The key idea. Commercial lattice tools such as nTop are costly, hard to script for parameter studies, and limit custom unit cells. In OpenVCAD a lattice is a short Python program, with any unit cell, conformal layouts, and graded geometry and materials, so one script can generate, simulate, and print a whole family of designs.

Lattice cells mapped onto a curved surface, cell by cell, so the lattice follows the shape of the part.

Lattices as Python programs

A lattice in OpenVCAD has two parts: a unit cell, which defines what repeats, and a cell map, which defines where and how it repeats. The built-in library covers the common families, and custom cells can be written in Python from vertices and edges, implicit functions, or Voronoi seeds. The result is an ordinary OpenVCAD object, so it can be combined with other geometry, carry material fields, and be compiled for any supported printer.

import pyvcad as pv
import pyvcad_metamaterials as mm

# A 30 mm block of gyroid cells, three on each side
cell_map = mm.rectangular_cell_map(
    (pv.Vec3(-15, -15, -15), pv.Vec3(15, 15, 15)),
    cells=(3, 3, 3),
)
lattice = mm.gyroid(cell_map, wall_thickness=1.2)

Swapping mm.gyroid for mm.octet or a custom cell, or looping over wall thickness for a parameter study, is a one-line change.

Catalogue of rendered unit cells in three rows, triply periodic surfaces, strut lattices, and plate lattices
Built-in unit cells. Triply periodic minimal surfaces (top), strut lattices (middle), and face and plate lattices (bottom).

Conformal lattices

Cells do not have to sit on a rectangular grid. A cell map can follow a CAD face, fill the space between two surfaces, or wrap a triangle mesh, so the lattice bends with the part instead of being clipped out of a block. The plate-lattice helmet liners are a real-world example: a lattice that fills the space between the head and the helmet shell.

Render of a gray diamond strut lattice draped over a saddle-shaped surface
Following a CAD face. A diamond lattice mapped over a freeform saddle.
Render of a curved gray gyroid panel with five round vent holes and a solid rim
Following a mesh. A gyroid on a domed mesh patch, trimmed around five vents and sealed with a rim.
Render of a gray sphere covered in a regular pattern of Schwarz-P lattice openings
Filling a shell. A Schwarz-P lattice through a spherical protective shell.

Custom, stochastic, and mixed cells

Researchers often need cells that no catalogue includes. Graph cells, periodic implicit functions, Voronoi and Delaunay foams, and joins between different cell types are all built from the same few Python interfaces.

Render of an open-cell foam of pentagonal and hexagonal struts
Custom cells. A Weaire–Phelan foam generated from Voronoi seeds in a few lines of Python.
Cutaway render of irregular flat cell walls packed inside a sphere
Stochastic lattices. Graded Voronoi cell walls inside a sphere, shown in cutaway.
Render of an octet strut lattice on one side merging into a Schwarz-P sheet lattice on the other
Mixed topologies. A Schwarz-P sheet lattice joining an octet beam lattice through a shared band.

Grading geometry and materials

Wall thickness, strut radius, and cell size can vary through a lattice, and so can material properties such as stiffness or color. Each of these is a field that can be defined across the whole lattice, restarted inside every cell, or laid out along each strut or wall. Because the fields belong to the cells, they stay in place when the lattice is mapped onto a curved part.

Render of a cylindrical strut lattice colored yellow at the joints and dark blue at the middle of each strut, with a color scale
Graded along each strut. A cylindrical lattice whose struts are stiff at the joints (yellow) and soft at midspan (blue). The pattern stays attached to the struts as they bend around the map.
Cutaway render of a gyroid block colored yellow on its wall surfaces with a thin dark blue core inside each wall, with a color scale
Graded through each wall. A clipped gyroid whose walls have a high-valued skin on both faces (yellow) and a lower-valued core (blue).

Model, simulate, print

The same design can be exported for finite element analysis, with meshes refined where the material changes, and compiled for multi-material printing. That makes it practical to study how a lattice’s material layout changes its behavior before printing it.

Grid of five rows labeled static, linear gradient, Gaussian gradient, sinusoidal gradient, and per-cell gradient, each showing a concentration plot, a green and blue lattice render, a simulation setup, a deformed simulation result, and a photo of a printed pink and white lattice between compression plates
Model, simulate, print. One lattice with five distributions of a soft and a rigid material. Each row shows the OpenVCAD render, the simulation setup, the simulated compression, and the printed lattice. Where the soft material sits controls where the lattice gives way first.

Publications

  1. Implicit Modeling for 3D-Printed Multi-Material Computational Object Design via Python
    Implicit Modeling for 3D-Printed Multi-Material Computational Object Design via Python
    Charles Wade, Devon Beck, and Robert MacCurdy
    SCFProceedings of the 10th ACM Symposium on Computational Fabrication (SCF ’25), 2025
  2. OpenVCAD: An Open Source Volumetric Multi-Material Geometry Compiler
    OpenVCAD: An Open Source Volumetric Multi-Material Geometry Compiler
    Charles Wade, Graham Williams, Sean Connelly, Braden Kopec, and Robert MacCurdy
    Additive Manufacturing 79, 103912, 2024