an aakaar surface

animaterial

A simulation operating system — not an FEM package. You describe what a thing is and what it must do; animaterial lifts it to a semantic model, chooses how to solve it, and shows you the field.

open-source · local-first · structural today, multiphysics by design

Open the web studio — stage a study Download offline studio (on-system solvers)

Animaterial Studio — the desktop interface — Animaterial Studio — intake, specify, solve, in one local desktop app.

The idea

The central object is the Analysis Model — a feature graph, not the mesh. Animaterial never meshes raw CAD: it lifts geometry to a semantic model of typed features, chooses a solver, and only then discretizes the way that solver wants. A mesh is one option; for mechanisms there may be no mesh at all. Everything — geometry, meshing, solver, material, visualization — is a plugin, which is exactly how thermal / fluid / EM / biomechanics get added later without re-architecting.

1 · Intake & feature graph

Ingest a model (STEP / IGES / STL / glTF…), heal it, and lift it into a graph of typed features — beams, shells, solids, joints, contacts.

2 · Specify

Choose a solver mode early, then attach boundary conditions, loads, connectors, and materials — symbolic expressions welcome.

3 · Solve

Run one of four modes on whatever CPU / RAM / GPU the machine has, and return results as one universal Field ontology.

Four solver modes

One canonical model, four ways to solve it — chosen by what you're asking, each on its own open-source engine.

Kinetoelastic (graph)

mechanisms, compliant & deployable structures — force paths & mechanical advantage

SymPy/SciPy (symbolic) + Taichi-XPBD

Elastica / Cosserat

slender, large-deflection bodies — tapes, cables, continuum robots

PyElastica (+ SOFA)

FEM explicit

impact, drop, fast transients, contact

CalculiX-explicit + Taichi

FEM implicit

statics, modal, buckling, quasi-static

CalculiX (+ Code_Aster, FEniCSx)

Seeing it work

Real studies, run on open-source solvers — from a textbook-validated beam to a snapping compliant mechanism.

Cantilever displacement field — **FEM-implicit, validated.** A cantilever beam (Gmsh → scikit-fem P2): tip deflection 0.1903 mm vs the Euler-Bernoulli closed form 0.1905 mm — a 0.09% agreement across the whole geometry → mesh → solve → post chain.

Force–displacement curve — **FEM-implicit, validated.** A cantilever beam (Gmsh → scikit-fem P2): tip deflection 0.1903 mm vs the Euler-Bernoulli closed form 0.1905 mm — a 0.09% agreement across the whole geometry → mesh → solve → post chain.

Compliant gripper snapping through inversion — **A compliant gripper, snapping through inversion.** A large-deflection CalculiX study — the kind of buckling / snap-through behaviour a mesh-first tool fights, captured here as the feature graph intends.

Gripper stress field — **A compliant gripper, snapping through inversion.** A large-deflection CalculiX study — the kind of buckling / snap-through behaviour a mesh-first tool fights, captured here as the feature graph intends.

What it won't compromise

100% open source

Every dependency carries an OSI-approved licence. No proprietary kernels.

Local-first

Runs on your workstation, scales to its cores / RAM / GPU. Cluster & cloud are an optional backend.

Never mesh raw CAD

Lift to a semantic feature graph first; the mesh is solver-driven, not the starting point.

One model, one result

Every plugin reads & writes the same Study and the same Fields — which is how multiphysics gets added without re-architecting.

Request a research preview

Animaterial is in research preview. Tell us what you'd simulate with it and we'll get you access — the request goes straight to the team.