2D gaussian splatting的配置和可视化

长安游2024-12-19 19:19

继3D gaussian splatting，2D gaussian splatting除了渲染新视角，还能够生成mesh模型。

2D gaussian splatting的配置

两者的运行环境基本一致

GitHub - hbb1/2d-gaussian-splatting: $SIGGRAPH'24$ 2D Gaussian Splatting for Geometrically Accurate Radiance Fields

增加的内容有

git clone https://github.com/hbb1/2d-gaussian-splatting.git --recursive

pip install submodules/diff-surfel-rasterization

pip install submodules/simple-knn

pip install opencv-contrib-python

pip install plyfile

pip install matplotlib

pip install mediapy

pip install open3d

pip install trimesh

pip install tqdm

SIBR可视化

GitHub - RongLiu-Leo/Gaussian-Splatting-Monitor: Monitor Gaussian Splatting additional real-time viewable and differentiable outputs

git clone https://github.com/RongLiu-Leo/Gaussian-Splatting-Monitor.git

cd SIBR_viewers

cmake -Bbuild . -DCMAKE_BUILD_TYPE=Release

cmake --build build -j24 --target install

运行SIBR_remoteGaussian_app，再

train-s /home/zousiyuan/data/DTU/DTU/scan105 -m output/data/scan105 -r 2 --depth_ratio 1

render.py -s /home/zousiyuan/data/DTU/DTU/scan105 -m output/data/scan105 -r 2 --depth_ratio 1 --skip_test --skip_train