Home

Installation

Binaries are not provided yet. Please follow the instructions to build LichtFeld Studio from source.

Requirements

Software

• OS: Linux (Ubuntu 22.04+) or Windows
• CMake: 3.30 or higher
• Compiler: C++23 compatible (GCC 14+ or Clang 17+)
• CUDA: 12.8 or higher (required)
• LibTorch: 2.7.0 (setup instructions below)
• vcpkg: For dependency management

Hardware

• GPU: NVIDIA GPU with compute capability 7.5+
• VRAM: Minimum 8GB recommended
• Tested GPUs: RTX 4090, RTX A5000, RTX 3090Ti, A100, RTX 2060 SUPER

Build instructions

• Linux
• Windows
• Docker

Usage

The preferred way to use LichtFeld Studio is to import your data (undistorted images + pointcloud + camera locations) in COLMAP format.
Have a look at these 2 introduction videos on how to get your images ready for use in LichtFeld Studio:

Example datasets can be found here

Once your dataset is ready, you can use LFS to train your images to create a Gaussian Splat, either using the GUI or the command line.

• GUI: start LightFeld Studio and use "Import dataset" to load your dataset
• Command line: Basic training:

./build/LichtFeld-Studio -d data/garden -o output/garden

Training with evaluation and visualization:

./build/LichtFeld-Studio \
    -d data/garden \
    -o output/garden \
    --eval \
    --save-eval-images \
    --render-mode RGB_D \
    -i 30000

MCMC strategy with limited Gaussians:

./build/LichtFeld-Studio \
    -d data/garden \
    -o output/garden \
    --strategy mcmc \
    --max-cap 500000

More command line options: command line options