Getting Started With 3D Gaussian Splatting for Windows(submodules error, create env error)

Getting Started With 3D Gaussian Splatting for Windows

References

https://www.youtube.com/watch?v=UXtuigy_wYc

https://github.com/jonstephens85/gaussian-splatting-Windows

GitHub - jonstephens85/gaussian-splatting-Windows: 3D Gaussian Splatting for Real-Time Radiance Field Rendering - Detailed Windo

 

1. Install Dependencies/Requirements

• CUDA 11.8
• Visual Studio 2019
  • 3DGS has an issue with Visual Studio 2022 !
  • Install VS 2019
  • https://learn.microsoft.com/en-us/visualstudio/releases/2019/history#release-dates-and-build-numbers

• COLMAP
  
• FFMEG(It's for convert a video to images)
  https://www.youtube.com/watch?v=JR36oH35Fgg
• ImageMagik(not necessary. It's for convert a video to images)

COLMAP install on Windows
https://demuc.de/colmap/

COLMAP - Structure-from-Motion and Multi-View Stereo

 

FFMPEG install on Windows

https://www.youtube.com/watch?v=JR36oH35Fgg

2. Run anaconda3 prompt

3. Cloning the repository

# SSH
git clone git@github.com:graphdeco-inria/gaussian-splatting.git --recursive

# HTTPS
git clone <https://github.com/graphdeco-inria/gaussian-splatting> --recursive

4. Setting up the optimizer

1. go into the cloned repo

cd gaussian-splatting

2. type the followings

SET DISTUTILS_USE_SDK=1 # Windows only
conda env create --file environment.yml
conda activate gaussian_splatting

if failed to create conda env, follow the steps below "Conda env create error"

5. Preparing images for the optimizer

Convert a video to images(FFMEPG installed)

cd (root폴더)\data\tube\input
ffmpeg -i (root폴더)\data\tube\tube_input.mp4 -qscale:v 1 -qmin 1 -vf fps=10 %04d.jpg

1. Create data folder in the repo.
2. Create input folder and put images into it.(playroom/input, in my case)
3. Run the following command at gaussian-splatting

(gaussian_splatting) C:\\Users\\ghdth\\gaussian-splatting> python convert.py -s data/playroom

Results are like below

6. Run the optimizer

python train.py -s <path to COLMAP or NeRF Synthetic dataset>

For example,

 

python train.py -s data/playroom

 

Errors

Conda env create error

 

...
× Encountered error while trying to install package.
╰─> simple_knn
...

 

Follow the steps below

# init conda environment
conda create --name gaussian_splatting -y python=3.8
conda activate gaussian_splatting
pip install --upgrade pip
# or it would be like this C:\Users\ghdth\miniconda3\envs\gaussian_splatting\python.exe -m pip install --upgrade pip

# install PyTorch and cuda-toolkit
pip install torch==2.1.2+cu118 torchvision==0.16.2+cu118 --extra-index-url https://download.pytorch.org/whl/cu118
conda install -c "nvidia/label/cuda-11.8.0" cuda-toolkit
conda install -c conda-forge vs2019_win-64

# clone the source
git clone <https://github.com/graphdeco-inria/gaussian-splatting> --recursive
cd gaussian-splatting

# install dependencies
pip install plyfile tqdm ninja

# install submodules
pip install submodules/diff-gaussian-rasterization
pip install submodules/simple-knn

No module named ‘cv2’

pip install opencv-python