[ICLR 2025] 3DGS-Drag: Dragging Gaussians for Intuitive Point-Based 3D Editing

Install

conda create -n draggs python=3.10
conda activate draggs

pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

# set CUDA_HOME
export CUDA_HOME=<path of cuda>


# Required for colmap data processing
conda install -c conda-forge colmap
conda install -c conda-forge imagemagick

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

Example Usage

Data process (Follow Gaussian Splatting)

python convert.py -s <location> [--resize] #If not resizing, ImageMagick is not needed

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

Drag Editing (Currently needs the drag configuration file)

python train_drag.py -s <dataset path> --ply <checkpoint path> --drag_path <config path>

Data examples: We provide one processed data from IN2N with config filese in cfgs. the example data can be downloaded from drive

✅ To-Do List

• GUI Interface

Citation

@inproceedings{3dgs-drag2025,
      author = {Dong, Jiahua and Wang, Yu-Xiong},
      title = {3DGS-Drag: Dragging Gaussians for Intuitive Point-Based 3D Editing},
      booktitle = {ICLR},
      year = {2025},
     } 

Acknowledgments

This work is greatly inspired aned based on 3D Gaussian Splatting (https://github.com/graphdeco-inria/gaussian-splatting)