Installation

This tutorial was developed and tested on Windows 11. Most steps also work on Linux, with some differences that will be addressed in a separate guide.

Recommended environment

For efficient training and reproducibility, use a system with the following configuration:

Hardware

NVIDIA GPU with CUDA support

Software

Windows 11
Git
Miniconda or Anaconda
Microsoft Build Tools for Visual Studio 2022
Python 3.10
CUDA Toolkit 12.6 or a compatible version (install separately)
COLMAP (with CUDA support)
gsplat
SuperSplat

Note: Minor differences in software versions or CUDA drivers may require small adjustments.

Note: The CUDA Toolkit must be installed separately from the GPU driver to enable GPU acceleration. Without it, tools like PyTorch will not detect CUDA support. For instructions and compatibility details, see the CUDA Toolkit documentation.

Tested setup

This tutorial was verified using the following configuration:

OS: Windows 11 (24H2)
GPU: NVIDIA GeForce RTX 4070 Laptop GPU
CPU: Intel Core i9-13900H (13th Gen), 2.6 GHz, 14 cores
RAM: 32 GB
CUDA Toolkit: Version 12.6
Python: 3.10 (installed via Miniconda)
COLMAP: v3.11.1 with CUDA https://github.com/colmap/colmap/releases
gsplat: v1.5.1 https://github.com/nerfstudio-project/gsplat/releases/tag/v1.5.1
SuperSplat: https://superspl.at/editor