Skip to content
/ SC-GS Public

[CVPR 2024] Code for SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes

License

Notifications You must be signed in to change notification settings

CVMI-Lab/SC-GS

Repository files navigation

SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes

This is the code for SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes.

Given (a) an image sequence from a monocular dynamic video, we propose to represent the motion with a set of sparse control points, which can be used to drive 3D Gaussians for high-fidelity rendering.Our approach enables both (b) dynamic view synthesis and (c) motion editing due to the motion representation based on sparse control points

Install

pip install -r requirements.txt

# a modified gaussian splatting (+ depth, alpha rendering)
git clone --recursive https://github.com/ashawkey/diff-gaussian-rasterization
pip install ./diff-gaussian-rasterization

# simple-knn
pip install ./submodules/simple-knn

Run

# Run with GUI (for the resolution of 400*400 with best PSNR)
CUDA_VISIBLE_DEVICES=0 python train_gui.py --source_path YOUR/PATH/TO/DATASET/jumpingjacks --model_path outputs/jumpingjacks --deform_type node --node_num 512 --is_blender --eval --gt_alpha_mask_as_scene_mask --local_frame --resolution 2 --W 800 --H 800 --gui

# Run with GUI (for the resolution of 800*800)
CUDA_VISIBLE_DEVICES=0 python train_gui.py --source_path YOUR/PATH/TO/DATASET/jumpingjacks --model_path outputs/jumpingjacks --deform_type node --node_num 512 --is_blender --eval --gt_alpha_mask_as_scene_mask --local_frame --W 800 --H 800 --random_bg_color --white_background --gui

Editing

sc-gs-editing-guidance.mp4

Acknowledgement

@article{yang2023deformable3dgs,
    title={Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction},
    author={Yang, Ziyi and Gao, Xinyu and Zhou, Wen and Jiao, Shaohui and Zhang, Yuqing and Jin, Xiaogang},
    journal={arXiv preprint arXiv:2309.13101},
    year={2023}
}
  • Credits to authors of 3D Gaussians for their excellent code.
@Article{kerbl3Dgaussians,
      author       = {Kerbl, Bernhard and Kopanas, Georgios and Leimk{\"u}hler, Thomas and Drettakis, George},
      title        = {3D Gaussian Splatting for Real-Time Radiance Field Rendering},
      journal      = {ACM Transactions on Graphics},
      number       = {4},
      volume       = {42},
      month        = {July},
      year         = {2023},
      url          = {https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/}
}

Citing

If you find our work useful, please consider citing:

@article{huang2023sc,
  title={SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes},
  author={Huang, Yi-Hua and Sun, Yang-Tian and Yang, Ziyi and Lyu, Xiaoyang and Cao, Yan-Pei and Qi, Xiaojuan},
  journal={arXiv preprint arXiv:2312.14937},
  year={2023}
}

About

[CVPR 2024] Code for SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published