# autonomousvision/gaussian-opacity-fields

[SIGGRAPH Asia'24 & TOG] Gaussian Opacity Fields: Efficient Adaptive Surface Reconstruction in Unbounded Scenes

Repository: https://github.com/autonomousvision/gaussian-opacity-fields
Canonical: https://ross.abutalabs.com/products/gaussian-opacity-fields
Homepage: https://niujinshuchong.github.io/gaussian-opacity-fields/
Language: Python
License: NOASSERTION
License Family: other
Topics: 3d-reconstruction, gaussian-splatting, novel-view-synthesis, surface-reconstruction, tetrahedral-meshing
Last push: 2024-11-15T11:57:45+00:00

## Health v2 (maintenance only)
Score: 25/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 62
- inputs: {"age_days": 869, "days_push": 656, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1017, forks 69 (observed 2026-08-28T04:03:14.697987+00:00)

## What it is
Gaussian Opacity Fields (GOF) is a Python/CUDA research implementation for efficient, adaptive surface reconstruction in unbounded scenes using 3D Gaussian Splatting. It extracts geometry directly from 3D Gaussians by identifying their level set and uses Marching Tetrahedra for compact, scene-adaptive mesh extraction.

## Use cases
- reconstruct 3D surfaces from multi-view images with gaussian splatting
- extract meshes from 3D gaussian splatting scenes
- novel view synthesis on unbounded scenes like Mip-NeRF 360
- reconstruct surfaces of Tanks and Temples or DTU scenes
- compare 3DGS-based surface reconstruction against neural implicit methods
- train gaussian splatting models with geometry regularization

## When to choose
- you need high-quality mesh extraction from 3D Gaussians without Poisson reconstruction or TSDF fusion
- you want fast training and adaptive, compact meshes in unbounded scenes
- you are doing research on gaussian splatting surface reconstruction and want a SIGGRAPH Asia 2024 baseline

## When to avoid
- you only need novel view rendering without geometry or meshes
- you lack a CUDA GPU or cannot build custom CUDA submodules
- you need a production-ready tool with a polished UI rather than research code

## Facets
- artifact type: library
- maturity: active
- function: computer-vision, image-processing, graphics, machine-learning
- domain: computer-vision, graphics, machine-learning
- platform: python
- tags: gaussian-splatting, 3d-reconstruction, surface-reconstruction, novel-view-synthesis, marching-tetrahedra, mesh-extraction, research-code, linux, gpu

## Member repositories
- autonomousvision/gaussian-opacity-fields (main) score 25

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:14.697987+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:11:10.869894+00:00, confidence not recorded.
  - readme: https://github.com/autonomousvision/gaussian-opacity-fields (fetched 2026-08-28T04:03:14.697987+00:00, sha ad2fd5c88504)
  - homepage: https://niujinshuchong.github.io/gaussian-opacity-fields/ (fetched 2026-08-29T13:10:00.803678+00:00, sha 0928967487ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.