Skip to content

Installation¤

COMPAS Model requires Python 3.12 or newer. The project is continuously tested with Python 3.12, 3.13, and 3.14.

Minimal¤

pip install compas_model

Viewer¤

pip install "compas_model[viz]"

The viewer is optional and is not required for model construction, geometry algorithms, collision detection, contact detection, or headless tests.

Precise mesh collision¤

Precise mesh-surface collision uses the optional Coal backend. Coal is available from PyPI, so it can be installed through the project extra:

pip install "compas_model[collision]"

PyPI wheel availability depends on the Python version and operating system. A source installation additionally requires Coal's native build dependencies. For a predictable compiled installation across supported platforms, use conda-forge instead:

conda install -c conda-forge coal

See the Coal package on PyPI for the currently published source distribution and platform wheels.

Precise BRep collision¤

Precise BRep collision operates on native Open CASCADE topology through the optional COMPAS OCC backend. COMPAS OCC is not distributed through PyPI and therefore is intentionally not declared as a Python package extra. Install it from conda-forge, then install COMPAS Model with pip in the same environment:

conda create -n model -c conda-forge python=3.12 compas_occ
conda activate model
python -m pip install compas_model

See the official COMPAS OCC installation instructions for supported Conda setups.

Complete optional setup¤

To use precise Mesh collision, precise BRep operations, and the viewer in one environment:

conda create -n model -c conda-forge python=3.12 coal compas_occ
conda activate model
python -m pip install "compas_model[viz]"

Development¤

git clone https://github.com/blockresearchgroup/compas_model.git
cd compas_model
conda env create -f environment.yml
conda activate model-dev

The development environment intentionally includes Coal, COMPAS OCC, and COMPAS Viewer so the complete optional test suite can be run locally. The standard CI matrix installs only the package and development tools from PyPI; separate jobs verify the pip-installed viewer and the Conda-installed Coal and COMPAS OCC backends.