rerun-sdk 0.34.0rc4

The Rerun Logging SDK

The Rerun Python SDK

Use the Rerun SDK to record data like images, tensors, point clouds, and text. Data is streamed to the Rerun Viewer for live visualization or to file for later use.

Install

pip3 install rerun-sdk

ℹ️ Note: The Python module is called rerun, while the package published on PyPI is rerun-sdk.

For other SDK languages see Installing Rerun.

We also provide a Jupyter widget for interactive data visualization in Jupyter notebooks:

pip3 install rerun-sdk[notebook]

Example

import rerun as rr
import numpy as np

rr.init("rerun_example_app", spawn=True)

positions = np.vstack([xyz.ravel() for xyz in np.mgrid[3 * [slice(-5, 5, 10j)]]]).T
colors = np.vstack([rgb.ravel() for rgb in np.mgrid[3 * [slice(0, 255, 10j)]]]).astype(np.uint8).T

rr.log("points3d", rr.Points3D(positions, colors=colors))

Resources

• Examples
• Python API docs
• Quick start
• Tutorial
• Troubleshooting
• Discord Server

Logging and viewing in different processes

You can run the Viewer and logger in different processes.

In one terminal, start up a Viewer with a server that the SDK can connect to:

python3 -m rerun

In a second terminal, run the example with the --connect option:

python3 examples/python/plots/plots.py --connect

Note that SDK and Viewer can run on different machines!

Building Rerun from source

We use pixi for managing dev-tool versioning, download and task running. See here for installation instructions.

pixi run py-build

This builds the SDK for Python (use pixi run py-build --release for a release build).

You can then run examples via uv:

pixi run uv run examples/python/minimal/minimal.py

To build a wheel instead for manual install use:

pixi run py-build-wheel

Refer to BUILD.md for details on the various different build options of the Rerun Viewer and SDKs for all target languages.

Installing a pre-release

Prebuilt dev wheels from head of main are available at https://github.com/rerun-io/rerun/releases/tag/prerelease.

While we try to keep the main branch usable at all times, it may be unstable occasionally. Use at your own risk.

Running Python unit tests

pixi run py-test

If you run into a problem, run rm -rf .pixi .venv and try again.

Running specific Python unit tests

pixi run py-build && pixi run uvpy -m pytest rerun_py/tests/unit/test_tensor.py

Profiling the Python SDK

Set RERUN_PUFFIN=1 to spawn a puffin_viewer attached to the SDK on startup. The Rust side of the SDK then streams scopes (anything wrapped in re_tracing::profile_function! / profile_scope!) to the viewer for the lifetime of the process.

RERUN_PUFFIN=1 pixi run uvpy your_script.py

Save a recording from the viewer for offline analysis (use the investigate-puffin skill in .claude/skills/).