[3D CV 연구] 3DGS SuGaR implementations missing point cloud

SuGaR는 먼저 학습하고 output mesh를 생성하고, 이후 mesh와 gaussian에 대한 joint-refinement를 수행해야하므로 command를 두번 나눠서 쳐줍니다.

run GS firstly. like this:

python -Bu gaussian_splatting/train.py -s ../data/image/resin/output/ --iterations 7000 -m ./output/_guass_splat/

then:

python -Bu train.py -s ../data/image/resin/output/ -c ./output/_guass_splat/ -r density

modify to your real paths: data path and checkpoint path.

https://github.com/Anttwo/SuGaR/issues/51