trimesh

trimesh is a pure Python 3.10+ library for loading and using triangular meshes, with an emphasis on watertight surfaces.

Key capabilities include:

• Import/Export: Supports STL, OBJ, PLY, GLTF/GLB, 3MF, XAML, 3DXML, and more.
• Analysis: Calculate volume, center of mass, moment of inertia, watertightness, and convexity.
• Geometry Operations: Convex hulls, slicing, boolean operations (via Manifold3D or Blender), voxelization, and smoothing.
• Queries: Ray-mesh queries and nearest point calculations.
• Primitives: Built-in support for Box, Cylinder, Sphere, and Extrusion objects.
• Visualization: OpenGL preview via pyglet or in-line in Jupyter/marimo notebooks via three.js.

To create a lightweight Docker image, use an official first-party base image (such as python:3.11-slim-bullseye) and install trimesh with the [easy] extra via pip.

FROM python:3.11-slim-bullseye
RUN pip install trimesh[easy]

To install the minimal version of trimesh, which only requires numpy and enables loading basic formats like STL, PLY, OBJ, GLB, and GLTF into numpy arrays, use:

pip install trimesh

To install trimesh along with a set of "easy" soft dependencies that generally install cleanly on Linux (x86_64), MacOS (ARM), and Windows (x86_64), use the [easy] extra:

pip install trimesh[easy]

When choosing between the two Non-Rigid ICP methods, consider the following:

Users are advised to experiment with both algorithms using different steps parameter sets to find the best fit for their specific geometry.

When freezing dependencies (e.g., in pyproject.toml, requirements.txt, or uv.lock), it is recommended not to hard-code extras like trimesh[easy].

Instead, depend on specific packages (e.g., trimesh scipy). This helps avoid functionality mismatches with the latest versions used in Trimesh's test matrix and makes it easier to diagnose faults with specific specialized dependencies (like vhacdx).

To build the rendering image and run the provided example (which renders a sphere to a PNG), follow these steps in the directory containing the Dockerfile:

1. Build the image: Tag the image as renderworker.
2. Run the container: Mount your current working directory to the /output folder inside the container so that the rendered files are saved to your host machine.

Upon successful completion, the logs will indicate the number of rendered bytes, and an output.png file will appear in your current directory.

# build the image in the current directory
docker build -t renderworker .

# run the example
# `-v`: mount the current directory as a volume
# in the `/output` folder inside the image
docker run -v `pwd`:/output renderworker

You can install trimesh with just its hard dependency (numpy) or with a set of common soft dependencies for expanded functionality (like convex hulls, graph operations, and preview windows).

To install the minimal version:

pip install trimesh

To install with the recommended [easy] extra (includes common soft dependencies like scipy, networkx, pyglet, etc.):

pip install trimesh[easy]

If you are using a conda environment, you can install trimesh from the conda-forge channel:

conda install -c conda-forge trimesh

When selecting a mesh format for use with trimesh, consider the following recommendations:

• Recommended: GLB/GLTF. It is a scene format with flat arrays that allows for extremely fast loading via numpy.frombuffer. It has a well-defined specification and supports textures, point clouds, and more.
• For simple ASCII importers: OFF is a minimal text format containing only vertices and faces in ASCII.
• For wide tool compatibility: OBJ (Wavefront) is widely supported but can be slow to parse and may require re-indexing to match trimesh's internal data structure.
• For 3D printing: 3MF is an XML-based format focused on 3D printing.
• For CAD/Solidworks workflows: 3DXML is the easiest way to transition from Solidworks to a 3D scene in trimesh.