This README covers, first, dependencies and instructions for building Linux, then those for Windows.
- Install Docker.
- Post-installation steps for linux.
Make sure that
dockercan be executed without root privileges.
To verify that Docker is working, run:
# You should be able to run this without sudo.
docker run --rm hello-worldYou don't need to install Nvidia Docker to build CUDA container. You will need to install Nvidia Docker to run the CUDA container.
- Install Nvidia Docker. This is required for testing CUDA builds.
To verify that the Nvidia Docker is working, run:
docker run --rm --gpus all nvidia/cuda:12.1-base nvidia-smiYou can build and run ARM64 docker. This works on an ARM64 host including Apple Silicon. However, if your host is x86-64, you will need to install QEMU:
sudo apt-get --yes install qemu binfmt-support qemu-user-static
# Run the registering scripts
docker run --rm --privileged multiarch/qemu-user-static --reset -p yesTo verify that the ARM64 environment is working, run:
# This shall print "aarch64".
# The following warning message is expected: "WARNING: The requested image's
# platform (linux/arm64/v8) does not match the detected host platform
# (linux/amd64) and no specific platform was requested aarch64."
docker run --rm arm64v8/ubuntu:24.04 uname -pFor example:
cd docker
# Build Docker.
./docker_build.sh openblas-amd64-py38-dev
# Test Docker.
./docker_test.sh openblas-amd64-py38-devSee ./docker_build.sh and ./docker_test.sh for all available options.
You can build and test Open3D for Linux in a Docker container under Windows using the provided scripts thanks to Docker Desktop for Windows and WSL.
This guide walks you through installing Docker Desktop, setting up Windows Subsystem for Linux (WSL), configuring Docker integration with WSL, and building Open3D for Linux including its documentation and the Python wheel, and testing it, using the provided scripts (respectively docker_build.sh and docker_test.sh).
- Download and Install: Download Docker Desktop and follow the on-screen prompts to install.
- Launch Docker Desktop: After installation, open Docker Desktop to ensure it is running.
-
Enable WSL: Open PowerShell as Administrator and install WSL:
wsl --install -
Install a Linux Distribution (e.g., Ubuntu-24.04):
wsl --install -d Ubuntu-24.04
-
Restart your system if prompted.
- Open Docker Desktop.
- Go to Settings > Resources > WSL Integration.
- Enable Integration for your Linux distribution (e.g., Ubuntu-24.04).
- If necessary, restart Docker Desktop to apply the changes.
-
Open a terminal within WSL.
-
Clone and check out Open3D repository into the folder of your choice:
git clone https://github.com/isl-org/Open3D /path/to/Open3D
-
Open your WSL terminal.
-
Navigate to the Docker folder in the Open3D repository:
cd /path/to/Open3D/docker -
Disable PyTorch and TensorFlow ops if not needed:
export BUILD_PYTORCH_OPS=OFF export BUILD_TENSORFLOW_OPS=OFF
-
Run the Docker build script:
E.g.:
./docker_build.sh openblas-amd64-py312
Check the log of the build. After the build completes, you will have an Open3D Docker image ready to use, and the artifacts (binaries, documentation and Python package) will have been copied back to the host.
-
Run tests within the built Docker image:
E.g.:
./docker_test.sh openblas-amd64-py312