🎨 ReCoGS: Real-time ReColoring for Gaussian Splatting scenes

University of Modena and Reggio Emilia

Lorenzo Rutayisire, Nicola Capodieci, Fabio Pellacini

Overview

ReCoGS is a pipeline and a tool for interactively ReColoring a pre-trained gaussian splatting scene.

Additional scenes

How to run

Linux

Requirements:

• A NVIDIA GPU with Compute Capability 7.5 or higher
• vcpkg installation: make sure VCPKG_ROOT is defined.
• CUDA 12.6
• TensorRT 10.8.0.43

Clone and build the repository:

git clone https://github.com/loryruta/recogs update --init --recursive
cd recogs
# Download large assets
bash ./download_assets.sh
# Build
mkdir build
cd build
cmake .. \
  -DCMAKE_TOOLCHAIN_FILE=$VCPKG_ROOT/scripts/buildsystems/vcpkg.cmake \
  -DCMAKE_BUILD_TYPE=Release
cmake --build . --target recogs
cd ..

Download sample scenes from the MipNeRF360 dataset (any colmap dataset can be used as well):

bash ./download_data.sh

Run recogs:

./build/recogs ./data/mipnerf360/kitchen/point_cloud/iteration_30000/point_cloud.ply

⚠️ NOTE: on the first launch, recogs will take a long time to bootstrap (more than 10 minutes). This is caused by TensorRT having to compile the Stereo Matching model to an optimized version tailored to your specific device.

License

This code is distributed under the MIT license.

Citation

If you find our work helpful, you can cite us with this snippet:

@inproceedings{10.2312:stag.20251329,
booktitle = {Smart Tools and Applications in Graphics - Eurographics Italian Chapter Conference},
editor = {Comino Trinidad, Marc and Mancinelli, Claudio and Maggioli, Filippo and Romanengo, Chiara and Cabiddu, Daniela and Giorgi, Daniela},
title = {{ReCoGS: Real-time ReColoring for Gaussian Splatting scenes}},
author = {Rutayisire, Lorenzo and Capodieci, Nicola and Pellacini, Fabio},
year = {2025},
publisher = {The Eurographics Association},
ISSN = {2617-4855},
ISBN = {978-3-03868-296-7},
DOI = {10.2312/stag.20251329}
}