Skip to content

Quickstart

Get from zero to exploring a RealSim-CP scene in three steps.

Vehicle multi-camera preview

1. Download one scene

Grab a single variant to start — the Daiba night_1 scene has the most complete sensor coverage.

:material-google-drive: Download RealSim-CP

Download daiba_station_scenario/night_1/ and note its local path, e.g.:

C:\data\RealSim-CP\daiba_station_scenario\night_1

2. Install the visualizer

The RealSim-CP Visualizer is an Open3D desktop viewer. It needs Python 3.12 (Open3D does not yet support 3.14 on Windows).

# from the repository root
cd tools/realsim-cp-visualizer
py -3.12 -m venv .venv
.\.venv\Scripts\activate
python -m pip install -e .

On macOS / Linux:

cd tools/realsim-cp-visualizer
python3.12 -m venv .venv
source .venv/bin/activate
python -m pip install -e .

3. Run it

Point the viewer at the variant folder you downloaded:

python -m realsim_viz "C:\data\RealSim-CP\daiba_station_scenario\night_1"

You'll get a unified 3D scene with the fused multi-agent point cloud, 3D bounding boxes, and agent markers — plus a per-agent dashboard for inspecting individual camera and LiDAR streams.

Key controls
  • Space — play / pause
  • ← / → — step one frame
  • View buttons — Top / Side / 3D camera presets
  • Sidebar — toggle layers, individual agents, all RSUs, all vehicles
  • Dashboard — pick an agent, add camera/LiDAR tiles
  • Distortion — toggle the k1…k6 camera distortion model
  • Overlay LiDAR — project LiDAR points onto a camera tile

Full reference: Visualizer.

Prefer code?

To load the data programmatically (no GUI), head to Loading the data for a minimal Python walkthrough of the OpenLABEL file, the transform chain, and cuboid decoding.