graphdeco-inria/gaussian-splatting

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

24196 stars · 3471 forks observed · 2026-10-15

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

The official reference implementation of 3D Gaussian Splatting, a method for real-time radiance field rendering that reconstructs scenes from photos into 3D Gaussians. It supports training from calibrated image sets and real-time (≥30 fps) 1080p novel-view synthesis.

Use cases

• reconstruct 3D scenes from photos
• render novel views of a captured scene in real time
• train a radiance field model on my own image dataset
• reproduce results from the 3D Gaussian Splatting paper
• build 3D Gaussian splatting research on top of a reference codebase

When to choose

• you want the canonical, paper-faithful implementation of 3D Gaussian Splatting
• you need high-quality real-time novel-view synthesis at 1080p
• you have GPU resources and calibrated multi-view images
• you are doing graphics or vision research on radiance fields

When to avoid

• you need a permissively licensed library for commercial products (license is custom/non-standard)
• you have no CUDA-capable GPU
• you want a polished end-user photogrammetry application rather than research code
• you need simple object scanning with minimal setup

Facets

library · maturity active

graphics computer-vision machine-learning simulation computer-vision graphics machine-learning python windows gaussian-splatting novel-view-synthesis radiance-fields 3d-reconstruction neural-rendering reference-implementation linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="graphdeco-inria/gaussian-splatting")