# ziyc/drivestudio

A 3DGS framework for omni urban scene reconstruction and simulation.

Repository: https://github.com/ziyc/drivestudio
Canonical: https://ross.abutalabs.com/products/drivestudio
Homepage: https://ziyc.github.io/omnire/
Language: Python
License: MIT
License Family: permissive
Last push: 2025-08-27T08:42:40+00:00

## Health v2 (maintenance only)
Score: 40/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 39, release rhythm 35, longevity 52
- inputs: {"age_days": 734, "days_push": 371, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1254, forks 157 (observed 2026-08-28T04:04:08.798436+00:00)

## What it is
DriveStudio is a Python framework for 3D Gaussian Splatting (3DGS) based reconstruction and simulation of dynamic urban driving scenes. It jointly reconstructs backgrounds, vehicles, and non-rigid actors like pedestrians and cyclists from driving logs, and supports major autonomous driving datasets including Waymo, NuScenes, KITTI, Argoverse2, PandaSet, and NuPlan.

## Use cases
- reconstruct 3D urban driving scenes from lidar and camera logs
- simulate autonomous driving scenarios with dynamic actors
- train gaussian splatting models on Waymo or NuScenes datasets
- reconstruct pedestrians and cyclists in driving scenes
- generate novel view renders of street scenes for AV research
- benchmark neural scene reconstruction methods on driving datasets

## When to choose
- you need high-fidelity reconstruction of full dynamic urban scenes including non-vehicle actors
- you work in autonomous driving simulation or AV perception research
- you want a unified codebase supporting multiple driving datasets and Gaussian representations
- you want the official OmniRe implementation for research reproduction

## When to avoid
- you need real-time rendering on consumer hardware without a GPU
- your project involves indoor scenes or non-driving imagery
- you need a production-ready simulator with physics and sensor modeling rather than a research codebase
- you are not comfortable working with research-grade Python and large driving datasets

## Facets
- artifact type: framework
- maturity: active
- function: simulation, computer-vision, graphics, machine-learning, data-science
- domain: autonomous-vehicles, computer-vision, simulation, deep-learning, robotics
- platform: python, cross-platform
- tags: 3dgs, gaussian-splatting, urban-scene-reconstruction, autonomous-driving, neural-scene-graphs, waymo, nuscenes, kitti, argoverse, pandaset, nuplan, research-code, iclr-2025, omnire, linux, gpu

## Member repositories
- ziyc/drivestudio (main) score 40

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:08.798436+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 2026-08-30T05:07:26.805088+00:00, confidence not recorded.
  - readme: https://github.com/ziyc/drivestudio (fetched 2026-08-28T04:04:08.798436+00:00, sha 8bd0d1999ad3)
  - homepage: https://ziyc.github.io/omnire/ (fetched 2026-08-29T12:18:09.108706+00:00, sha a90f6e14b692)
- Data as of 2026-08-30T08:39:29.467469+00:00.