How-To Guides
Step-by-step setup and execution instructions for each tool included in the experimental analysis. Each guide covers prerequisites, installation, dataset preparation, training/loading, and export.
Trainer guides
| Tool | Platform | Prerequisites | Difficulty |
|---|---|---|---|
| Inria gaussian-splatting | Windows / Linux | CUDA, Python, Conda, Git | Medium |
| gsplat | Windows / Linux | CUDA, Python, Conda | Medium |
| OpenSplat | Windows / Linux | CUDA, CMake, LibTorch, OpenCV | Advanced |
| Nerfstudio | Windows / Linux | CUDA, Python, Conda | Medium |
| LichtFeld Studio | Windows | NVIDIA GPU ≥ 8 GB VRAM | Easy (precompiled binary) |
All trainer guides assume a pre-computed COLMAP reconstruction or provide instructions for running COLMAP as part of the pipeline. The guides are written for and tested on Windows 11 with an NVIDIA RTX 4060.
Viewer guides
| Tool | Platform | Prerequisites | Difficulty |
|---|---|---|---|
| Unity-VR-Gaussian-Splatting (ninjamode) | Windows VR | Unity 2022.3.47f1, VR headset | Advanced |
| GaussianSplattingVRViewerUnity (clarte53) | Windows VR | VR-ready PC, RTX 2060+, VR headset | Medium (precompiled binary available) |
Both viewer guides assume you already have a cleaned .ply file produced by one of the trainer guides above.
Contributing a new guide
To add a guide for a new tool:
- Create a new
.mdfile indocs/how-to/trainers/ordocs/how-to/viewers/following the structure of an existing guide (prerequisites → installation → dataset preparation → execution → export) - Add a row to the table above in this file
- Add an entry to the
nav:section ofmkdocs.yml
See CONTRIBUTING.md at the repository root for full guidelines.