Repository files navigation

A free and open source implementation of 3Dgaussian splatting written in C++, focused on being portable, lean and fast.

OpenSplat takes camera poses + sparse points inCOLMAP,OpenSfM,ODM,OpenMVG ornerfstudio project format and computes ascene file (.ply or .splat) that can be later imported forviewing, editing and rendering in othersoftware.

Graphics card recommended, but not required! OpenSplat runs the fastest on NVIDIA, AMD and Apple (Metal) GPUs, but can also run entirely on the CPU (~100x slower).

Commercial use allowed and encouraged under the terms of theAGPLv3. ✅

We even have asong 🎵

If you're on Windows, you canbuy the pre-built program. This saves you time and helps support the project ❤️. Then jump directly to therun section. As an alternative, check thebuild section below.

If you're on macOS or Linux check thebuild section below.

You can build OpenSplat with or without GPU support.

Requirements for all builds:

• OpenCV:sudo apt install libopencv-dev should do it.
• libtorch: See instructions below.

For libtorch visithttps://pytorch.org/get-started/locally/ and select your OS, for package select "LibTorch". For compute platform you can select "CPU".

Then:

git clone https://github.com/pierotofy/OpenSplat OpenSplatcd OpenSplatmkdir build&&cd buildcmake -DCMAKE_PREFIX_PATH=/path/to/libtorch/ ..&& make -j$(nproc)

Additional requirement:

• CUDA: Make sure you have the CUDA compiler (nvcc) in your PATH and thatnvidia-smi is working.https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html

For libtorch visithttps://pytorch.org/get-started/locally/ and select your OS, for package select "LibTorch". Make sure to match your version of CUDA if you want to leverage GPU support in libtorch.

Then:

git clone https://github.com/pierotofy/OpenSplat OpenSplatcd OpenSplatmkdir build&&cd buildcmake -DCMAKE_PREFIX_PATH=/path/to/libtorch/ ..&& make -j$(nproc)

Additional requirement:

• ROCm: Make sure you have the ROCm installed at /opt/rocm.https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html

For libtorch visithttps://pytorch.org/get-started/locally/ and select your OS, for package select "LibTorch". Make sure to match your version of ROCm (5.7) if you want to leverage AMD GPU support in libtorch.

Then:

git clone https://github.com/pierotofy/OpenSplat OpenSplatcd OpenSplatmkdir build&&cd buildexport PYTORCH_ROCM_ARCH=gfx906cmake -DCMAKE_PREFIX_PATH=/path/to/libtorch/ -DGPU_RUNTIME="HIP" -DHIP_ROOT_DIR=/opt/rocm -DOPENSPLAT_BUILD_SIMPLE_TRAINER=ON ..make

In addition, you can leverage Jinja to build the project

cmake -GNinja -DCMAKE_PREFIX_PATH=/path/to/libtorch/ -DGPU_RUNTIME="HIP" -DHIP_ROOT_DIR=/opt/rocm -DOPENSPLAT_BUILD_SIMPLE_TRAINER=ON ..jinja

There's several ways to build on Windows, but this particular configuration has been confirmed to work:

• Visual Studio 2022 C++
• https://github.com/Kitware/CMake/releases/download/v3.28.3/cmake-3.28.3-windows-x86_64.msi
• https://developer.download.nvidia.com/compute/cuda/11.8.0/network_installers/cuda_11.8.0_windows_network.exe
• https://download.pytorch.org/libtorch/cu118/libtorch-win-shared-with-deps-2.1.2%2Bcu118.zip
• https://github.com/opencv/opencv/releases/download/4.9.0/opencv-4.9.0-windows.exe

Then run:

"C:/Program Files/Microsoft Visual Studio/2022/Community/VC/Auxiliary/Build/vcvars64.bat"git clone https://github.com/pierotofy/OpenSplat OpenSplatcd OpenSplatmd buildcd buildcmake -DCMAKE_PREFIX_PATH=C:/path_to/libtorch_2.1.2_cu11.8/libtorch -DOPENCV_DIR=C:/path_to/OpenCV_4.9.0/build -DCMAKE_BUILD_TYPE=Release ..cmake --build . --config Release

Optional: Edit cuda target (only if required) beforecmake --build .

C:/path_to/OpenSplat/build/gsplat.vcxprojfor example: arch=compute_75,code=sm_75

If you're usingHomebrew, you can install Cmake/OpenCV/Pytorch by running:

brew install cmakebrew install opencvbrew install pytorch

You will also need to install Xcode and the Xcode command line tools to compile with metal support (otherwise, OpenSplat will build with CPU acceleration only):

1. Install Xcode from the Apple App Store.
2. Install the command line tools withxcode-select --install. This might do nothing on your machine.
3. Ifxcode-select --print-path prints/Library/Developer/CommandLineTools,then runsudo xcode-select --switch /Applications/Xcode.app/Contents/Developer.

Then run:

git clone https://github.com/pierotofy/OpenSplat OpenSplatcd OpenSplatmkdir build && cd buildcmake -DCMAKE_PREFIX_PATH=/path/to/libtorch/ -DGPU_RUNTIME=MPS .. && make -j$(sysctl -n hw.logicalcpu)./opensplat

If building CPU-only, remove-DGPU_RUNTIME=MPS.

⚠️ You will probably get alibc10.dylib can’t be opened because Apple cannot check it for malicious software error on first run. OpenSystem Settings and go toPrivacy & Security and find theAllow button. You might need to repeat this several times until all torch libraries are loaded.

⚠️ If you get aLibrary not loaded: @rpath/libomp.dylib error, try runningbrew link libomp --force before running OpenSplat.

Navigate to the root directory of OpenSplat repo that has Dockerfile and run the following command to build the Docker image:

docker build -t opensplat.

The-t flag and other--build-arg let you tag and further customize your image across different ubuntu versions, CUDA/libtorch stacks, and hardware accelerators.For example, to build an image with Ubuntu 22.04, CUDA 12.1.1, libtorch 2.2.1, and support for CUDA architectures 7.0 and 7.5, run the following command:

docker build \  -t opensplat:ubuntu-22.04-cuda-12.1.1-torch-2.2.1 \  --build-arg UBUNTU_VERSION=22.04 \  --build-arg CUDA_VERSION=12.1.1 \  --build-arg TORCH_VERSION=2.2.1 \  --build-arg CMAKE_CUDA_ARCHITECTURES="70;75;80" \  --build-arg CMAKE_BUILD_TYPE=Release.

Navigate to the root directory of OpenSplat repo that has Dockerfile and run the following command to build the Docker image:

docker build \  -t opensplat \  -f Dockerfile.rocm.

The-t flag and other--build-arg let you tag and further customize your image across different ubuntu versions, CUDA/libtorch stacks, and hardware accelerators.For example, to build an image with Ubuntu 22.04, CUDA 12.1.1, libtorch 2.2.1, ROCm 5.7.1, and support for ROCm architectures gfx906, run the following command:

docker build \  -t opensplat:ubuntu-22.04-cuda-12.1.1-libtorch-2.2.1-rocm-5.7.1-llvm-16 \  --build-arg UBUNTU_VERSION=22.04 \  --build-arg CUDA_VERSION=12.1.1 \  --build-arg TORCH_VERSION=2.2.1 \  --build-arg ROCM_VERSION=5.7.1 \  --build-arg PYTORCH_ROCM_ARCH="gfx906" \  --build-arg CMAKE_BUILD_TYPE=Release.

Note: If you want to use ROCm 6.x, you need to switch to AMD version of pytorch docker as a base layer to build:

docker build \  -t opensplat:ubuntu-22.04-libtorch-2.1.2-rocm-6.0.2 \  -f Dockerfile.rocm6.

To get started, download a dataset and extract it to a folder: [banana ] [train ] [truck ]

Then run from a command line prompt:

cd c:\path\to\opensplatopensplat.exe c:\path\to\banana -n 2000

cd build./opensplat /path/to/banana -n 2000

The program will generate an outputsplat.ply file which can then be dragged and dropped in one of the manyviewers such ashttps://playcanvas.com/viewer. You can also edit/cleanup the scene usinghttps://playcanvas.com/supersplat/editor. The program will also output acameras.json file in the same directory which can be used by some viewers.

To run on your own data, choose the path to an existingCOLMAP,OpenSfM,ODM ornerfstudio project. The project must have sparse points included (random initialization is not supported, see#7).

There's several parameters you can tune. To view the full list:

./opensplat --help

To generate compressed splats (.splat files), use the-o option:

./opensplat /path/to/banana -o banana.splat

You can resume training of a .PLY file by using the--resume option:

./opensplat /path/to/banana --resume ./splat.ply

To train a model with AMD GPU using docker container, you can use the following command as a reference:

1. Launch the docker container with the following command:

docker run -it -v~/data:/data --device=/dev/kfd --device=/dev/dri opensplat:ubuntu-22.04-libtorch-2.1.2-rocm-6.0.2 bash

2. Inside the docker container, run the following command to train the model:

export HIP_VISIBLE_DEVICES=0export HSA_OVERRIDE_GFX_VERSION=10.3.0# AMD RX 6700 XT workaroundcd /code/build./opensplat /data/banana -n 2000

We recently released OpenSplat, so there's lots of work to do.

• Support for running on AMD cards (more testing needed)
• Improve speed / reduce memory usage
• Distributed computation using multiple machines
• Real-time training viewer output
• Automatic filtering
• Your ideas?

https://github.com/pierotofy/OpenSplat/issues?q=is%3Aopen+is%3Aissue+label%3Aenhancement

We welcome contributions! Pull requests are welcome.

A single gaussian takes ~2000 bytes of memory, so currenly you need ~2GB of GPU memory for each million gaussians.

The methods used in OpenSplat are originally based onsplatfacto.

The code in this repository, unless otherwise noted, is licensed under the AGPLv3.

The code fromsplatfacto is originally licensed under the Apache 2.0 license and is © 2023 The Nerfstudio Team.