Advanced Installation

The minimal dependancies for trimesh are numpy, scipy and networkx.

All other dependancies are ‘soft’, or trimesh will only fail if a function is called that requires something not installed. If you do the most basic pip install, it will only require those three packages:

pip install trimesh

Conda Install

The easiest way to get going on the most platforms is through Conda.

# install Miniconda if you have no conda:
# if you are on Linux and feeling lazy, just run:
wget  --no-check-certificate -O
bash -b -p /opt/conda; rm
export PATH="/opt/conda/bin:$PATH"
conda update -q conda
conda create -q -n conda_env python=3.6
source activate conda_env

# install most trimesh requirements with built components from conda-forge
conda config --add channels conda-forge  # rtree, shapely, pyembree
conda install shapely rtree graph-tool pyembree numpy scipy
conda install -c conda-forge scikit-image

# install trimesh and all possible dependancies
# if this fails try: pip install trimesh[easy]
pip install trimesh[all]

Ubuntu Notes

Blender and openSCAD are soft dependancies used for boolean operations with subprocess, you can get them with apt:

sudo apt-get install openscad blender

Windows Notes

The easiest way to get going on Windows is to install the Anaconda Python distribution.

Most requirements are available as above, but to get meshpy the easiest way is from the Unofficial Windows Binaries from Christoph Gohlke