esc
Start typing to search the docs
Navigate Open
Get started
Menu
Get started

See what you can build with Rerun

A tour of Rerun in action: live recordings, code, and walkthroughs across robotics, computer vision, and machine learning.

Spatial computing and XR

Fuse depth, meshes, calibrated cameras, and detections into a single scene you can scrub through frame by frame, from ARKit captures to multi-view reconstructions.

Live

ARKit scenes

Visualize the ARKitScenes dataset: color images, depth, the reconstructed mesh, and labeled 3D boxes around indoor furniture.

2D 3D Depth Mesh

3D line mapping revisited

Visual walkthrough of "3D Line Mapping Revisited": detection, matching, triangulation, and joint optimization.

2D 3D Structure from motion Time series

Objectron

Visualize Google Research's Objectron: per-frame images, camera poses, sparse point clouds, planes, and 3D boxes.

2D 3D Object detection Pinhole camera

SimpleRecon: 3D reconstruction without 3D convolutions

Visual walkthrough of "SimpleRecon: 3D reconstruction without 3D convolutions", showing cost volumes, depth, and normals per frame.

3D Depth Time series Pinhole camera

Decoupling human and camera motion from videos in the wild

Visual walkthrough of SLAHMR, which chains ViTPose, PHALP, DROID-SLAM, HuMoR, and SMPL across three optimization stages.

3D SLAM Keypoint detection Mesh

VRS viewer

A C++ example that loads VRS (Facebook's per-device sensor record format) into Rerun: images, audio, IMU, and more.

2D 3D VRS Viewer

Learning to render novel views from wide-baseline stereo pairs

Visual walkthrough of "Learning to Render Novel Views from Wide-Baseline Stereo Pairs" via cross-attention along rays.

2D 3D View synthesis Time series

ARFlow: a framework for simplifying AR experimentation workflow

External AR research framework that uses Rerun to visualize multi-modal sensor data from a mobile client in real time.

3D Augmented reality Spatial computing Integration

Robotics

Log multi-rate sensor streams, joint states, controllers, and policies into one timeline. Replay a run, compare two, or watch one live as it happens.

Live

Any scalar

Log structured data once and plot any nested field with `jq`-style selectors. Shown on PID telemetry and live stock data.

Any scalar Plotting DynamicArchetype
Live

IMU signals

Log multi-dimensional IMU signals from the TUM VI Benchmark using the columnar `send_columns()` API.

Plots
Live

RGBD

Visualize an indoor recording from the NYU Depth V2 dataset (Kinect): synchronized RGB and depth aligned in 3D.

2D 3D Depth NYUD
Live

RRT*

Visualize the RRT* sampling-based path planner: watch the tree expand and re-wire toward an optimal route.

2D
Live

nuScenes

Visualize the nuScenes self-driving dataset: 6 cameras, lidar, radar, GPS, IMU, and 3D boxes on one timeline.

Lidar 3D 2D Object detection
Live

URDF

Load a URDF file and animate its joints over time with the `UrdfTree` helper, which resolves frame names and origins per joint.

3D Mesh URDF Animation

DROID

Visualize the DROID robot manipulation dataset: joint states, wrist and side-view stereo cameras, and teleoperator actions.

2D 3D Depth Pinhole camera

DROID semantic frame search

Find moments in robot demonstrations by describing them in plain language. Embeddings flow from a Rerun dataset into a vector store and deep-link back into the viewer.

Robotics Semantic search Embeddings SigLIP

Robot data preprocessing example

Assemble a robot recording from multiple file sources with Rerun's chunk processing API and preprocessing steps.

API example

LeRobot dataset from RRD

Convert RRD robot recordings into LeRobot v3 training datasets using the OSS Rerun server, with video remuxing.

Robotics MCAP LeRobot Dataset

ROS node

A minimal ROS 2 Python node that forwards pose, images, scans, point clouds, and velocities into Rerun.

2D 3D Pinhole camera ROS

Robby fischer

Robby Fischer is an autonomous chess-playing robot arm that detects human moves with per-square camera masks.

3D URDF Blueprint

ROS 2 bridge

A C++ ROS 2 bridge that subscribes to all supported topics and logs the messages into Rerun, with topic remapping.

2D 3D Pinhole camera ROS

ROS bridge

A C++ ROS 1 bridge that subscribes to all supported topics and logs the messages into Rerun (Linux only).

2D 3D Mesh Pinhole camera

KISS-ICP

Visualize the KISS-ICP LiDAR odometry pipeline on the NCLT dataset: deskewing, local map matching, ICP, step by step.

3D Point cloud Lidar

Live depth sensor

Stream live RGB + depth from an Intel RealSense sensor into a Rerun 3D view with a pinhole camera model.

2D 3D Live Depth

Lidar

Visualize LiDAR data from the nuScenes self-driving dataset, focused on the point cloud side of the recording.

Lidar 3D

NV12

Display an NV12-encoded video stream from a webcam, a reference for non-RGB pixel formats from camera hardware.

2D Image encoding YUV

MCAP

Load, convert, and post-process MCAP files in Rerun, including conversion of older ROS 1/2 bags into MCAP and RRD.

MCAP RRD ROS ROS 2

Eye control

Programmatically configure and animate the 3D view camera with the Rerun Blueprint API. Useful for cinematic walkthroughs.

Eye control 3D Pinhole camera

ROS TF

Map ROS 2 tf2 frames to Rerun's named transforms, including how to spot misconfigured transform trees.

ROS TF Transform Coordinate Frame

Diffusion models, LLMs, and machine learning

Visualize tensors, embeddings, diffusion intermediates, and end-to-end pipelines side-by-side. See what your model sees, at every step.

ControlNet

Generate an image from text conditioned on detected edges with Hugging Face's ControlNet. Every diffusion timestep is logged.

ControlNet Canny Hugging Face Stable diffusion

Vista driving world model

Vista is a generative driving world model built on Stable Video Diffusion. Visualizes latent diffusion steps and outputs.

2D Diffusion Hugging Face Video

Single image 3D reconstruction using MCC, SAM, and ZoeDepth

Chain SAM, ZoeDepth, and Multiview Compressive Coding to reconstruct a 3D object from a single image.

2D 3D Segmentation Point cloud

Training a model on the LeRobot dataset

Train a LeRobot policy on Hugging Face's `pusht` dataset and watch the training progress live in Rerun.

2D Hugging Face Imitation learning

Depth compare

Side-by-side comparison of monocular depth estimation networks, with 3D back-projection into world space for richer inspection.

2D 3D Hugging Face Depth

Mini NVS solver

Minimal NVS Solver: a video diffusion model used as a zero-shot novel-view synthesizer along a chosen camera path.

2D 3D Hugging Face Depth

Point-E and Shap-E

Visual comparison of two OpenAI text-to-3D models: Point-E (point clouds) and Shap-E (implicit NeRF weights).

3D Diffusion Point Mesh

LLM embedding-based named entity recognition

BERT-based named-entity recognition with UMAP-projected embeddings: text, tokens, entities, and 3D space in one view.

LLM Embeddings Classification Hugging Face

TFRecord loader

An external importer plugin that opens TFRecord (Tensorboard) event files in the Rerun viewer, extending the viewer without modifying it.

2D Tensor Loader Time series

Image and video understanding

Detections, segmentations, keypoints, and depth, overlaid on the source video, synchronized in time, and ready to scrub. Built for the messy reality of perception work.

Live

Segment anything model

Run Meta AI's Segment Anything on an image and log every mask as a transparent overlay via Rerun's `Tensor` archetype.

2D SAM Segmentation
Live

Detect and track objects

Object detection from Hugging Face Transformers plus OpenCV optical-flow tracking, aligned to the source video.

2D Hugging Face Object detection Object tracking

Face tracking

Detect and track faces, landmarks, and blendshapes from images, video, or a live camera using MediaPipe.

2D 3D Camera Face tracking

Interactive 3D annotation app with Rerun and Gradio

Interactive 3D annotation app built with Gradio and SAM 2. Track objects in monocular video, or fuse masks across multi-view RGB-D.

2D 3D Pinhole camera Time series

SAM 3D body: robust Full-Body human mesh recovery

SAM 3D Body recovers 3D human meshes from in-the-wild images, with the recovered body shown alongside the input.

3D Human mesh Body tracking Single-view

Depth compare

Prompt Depth Anything fuses iPhone LiDAR depth with a high-res image to produce metric depth up to 4K.

2D 3D Lidar Depth

Sam2 + DepthAnything2

Combine SAM 2 video tracking with Depth Anything 2 to track an object in 3D from a single monocular video.

2D 3D Hugging Face Depth

PaddleOCR

Run PaddleOCR layout analysis and text detection on documents: regions, detected text, and recovered output in one view.

Text OCR 2D Blueprint

Live camera edge detection

Run OpenCV's Canny edge detector on a live webcam stream, logging RGB, grayscale, and edge images side by side.

2D Canny Live OpenCV

TAPIR: tracking any point with per-frame initialization and temporal refinement

Visual walkthrough of TAPIR's per-point video tracking: cost volumes, predicted tracks, and occlusion maps per frame.

2D Point tracking Time series Tensor

DepthPro

Visual walkthrough of Apple's "Depth Pro": fast zero-shot monocular metric depth without needing camera intrinsics.

2D 3D Hugging Face Pinhole camera

EgoExo forge

A unified loader that visualizes multiple egocentric and exocentric human-activity datasets through one consistent layout.

3D Hugging Face Egocentric Exocentric

3D reconstruction and modelling

SLAM, structure-from-motion, photogrammetry, Gaussian splatting. Watch the pipeline build the scene step by step, with every intermediate inspectable.

Live

Raw mesh

Log a 3D scene as raw `Mesh3D` data or directly as a prepacked `Asset3D`.

Mesh
Live

Open photogrammetry format

Load and display a photogrammetric reconstruction in the Open Photogrammetry Format (OPF) using `pyopf`.

2D 3D Camera Photogrammetry
Live

Structure from motion

Visualize a sparse 3D reconstruction from COLMAP: camera frames, estimated poses, and an accumulating point cloud.

2D 3D COLMAP Pinhole camera

KISS-ICP

Visualize the KISS-ICP LiDAR odometry pipeline on the NCLT dataset: deskewing, local map matching, ICP, step by step.

3D Point cloud Lidar

Differentiable blocks world: qualitative 3D decomposition by rendering primitives

Visual walkthrough of "Differentiable Block Worlds", which reconstructs a textured 3D scene by differentiable rendering of superquadrics.

3D Mesh Pinhole camera Paper walkthrough

Mast3r slam - real-time dense slam with 3D reconstruction priors

Mast3r-SLAM is a real-time monocular SLAM system built on the Mast3r 3D reconstruction prior, robust in the wild.

2D 3D Pinhole camera Time series

Stereo vision SLAM

Stereo-vision SLAM on the KITTI self-driving dataset: vehicle trajectory plus the surrounding point cloud.

3D Point cloud C++

GLOMAP

GLOMAP is a global structure-from-motion pipeline. The video shows its joint global triangulation and camera-position step.

3D Point cloud GLOMAP

Hierarchical-Localization and GLOMAP

Combine the hloc toolbox with GLOMAP for fast structure-from-motion using deep features and matchers, runnable via Pixi.

2D 3D COLMAP Pinhole camera

InstantSplat

InstantSplat is a sparse-view, SfM-free framework that reconstructs large-scale Gaussian splatting scenes in seconds.

2D 3D Hugging Face Pinhole camera

VistaDream: sampling multiview consistent images for single-view scene reconstruction

VistaDream reconstructs a 3D scene from a single image with a two-stage coarse-scaffold + multi-view consistency pipeline.

3D Reconstruction Pinhole camera Diffusion

Integrations

ROS bridges, MCAP and rosbag readers, notebook embedding, custom dataloaders. Rerun meets your stack where it lives.

Live

Dicom MRI

Visualize a DICOM MRI scan with tensor slicing and 3D volume raymarching.

Tensor MRI DICOM

ROS node

A minimal ROS 2 Python node that forwards pose, images, scans, point clouds, and velocities into Rerun.

2D 3D Pinhole camera ROS

ROS 2 bridge

A C++ ROS 2 bridge that subscribes to all supported topics and logs the messages into Rerun, with topic remapping.

2D 3D Pinhole camera ROS

ROS bridge

A C++ ROS 1 bridge that subscribes to all supported topics and logs the messages into Rerun (Linux only).

2D 3D Mesh Pinhole camera

Notebook: minimal example

Embed the Rerun viewer directly inside Jupyter, Colab, or VSCode. Logged data shows up inline in the notebook.

Notebook API example 3D

Notebook: 2D neural fields

Train a small neural field to fit a 2D image inside Jupyter, with the live training progress logged into Rerun.

Notebook Neural Field 2D

LeRobot loader

Rerun's built-in importer for LeRobot datasets: point the viewer at a directory and it loads like any other recording.

2D Video Loader Hugging Face

VRS viewer

A C++ example that loads VRS (Facebook's per-device sensor record format) into Rerun: images, audio, IMU, and more.

2D 3D VRS Viewer

Revy - Rerun integration for Bevy

Revy is a time-travel debugger for Bevy: each frame's database is snapshot into Rerun for scrubbable engine state.

2D 3D Gamedev Bevy

TFRecord loader

An external importer plugin that opens TFRecord (Tensorboard) event files in the Rerun viewer, extending the viewer without modifying it.

2D Tensor Loader Time series

Eigen and OpenCV C++ integration

Minimal CMake project combining Rerun with Eigen for point and pose math and OpenCV for image reading and conversion.

2D 3D C++ Eigen

Stock charts

Fetch the last five days of stock data and present it through several different Rerun blueprints: same data, many shapes.

Time series Blueprint

ARFlow: a framework for simplifying AR experimentation workflow

External AR research framework that uses Rerun to visualize multi-modal sensor data from a mobile client in real time.

3D Augmented reality Spatial computing Integration

Air traffic data

Display live air traffic data from INVOLI on a geospatial map view, with pyproj reprojection and GeoPandas-backed loading.

2D 3D Map CRS

MCAP

Load, convert, and post-process MCAP files in Rerun, including conversion of older ROS 1/2 bags into MCAP and RRD.

MCAP RRD ROS ROS 2

ROS TF

Map ROS 2 tf2 frames to Rerun's named transforms, including how to spot misconfigured transform trees.

ROS TF Transform Coordinate Frame

Feature showcase

Compact, focused snippets that show how individual Rerun features work, handy when you want to learn one thing at a time.

Live

Any scalar

Log structured data once and plot any nested field with `jq`-style selectors. Shown on PID telemetry and live stock data.

Any scalar Plotting DynamicArchetype
Live

IMU signals

Log multi-dimensional IMU signals from the TUM VI Benchmark using the columnar `send_columns()` API.

Plots
Live

Raw mesh

Log a 3D scene as raw `Mesh3D` data or directly as a prepacked `Asset3D`.

Mesh
Live

State timeline

Simulates a robot work cell and tours every feature of the state timeline view: transitions, custom styling, resets, and columnar logging.

States API example
Live

Plots

A tour of Rerun's plotting primitives: bar charts, line plots, time-varying scalars, styled series, and measurements with uncertainty bands, each built from a few lines of code.

2D Plots API example
Live

Graphs

Different graph types and layouts visualized with Rerun's built-in force-directed engine Fjädra: trees, lattices, and random graphs.

Graph Layout Node-link diagrams Bubble charts
Live

Helix

Tiny example that logs points and lines to draw a 3D helix. A friendly starting point for entity paths and primitive archetypes.

3D API example

Notebook: minimal example

Embed the Rerun viewer directly inside Jupyter, Colab, or VSCode. Logged data shows up inline in the notebook.

Notebook API example 3D

Notebook: viewer

Embed a pre-recorded RRD inside a Jupyter notebook with the Rerun viewer widget. Great for demos and docs.

Notebook Widget 3D

Notebook: viewer callbacks

React to timeline, time, and selection events from an embedded Rerun viewer widget in Jupyter, surfaced as live Jupyter widgets.

Notebook Interactive Callbacks 3D

Clock

An analog clock built from `Boxes3D`, `Points3D`, and `Arrows3D` primitives, animated along a Rerun timeline.

3D API example

Compressed camera video stream

Capture webcam frames with pyAV, encode to H.264, and stream live into the viewer via the `VideoStream` archetype.

2D Image encoding Video Streaming

Log file

How to log any file from the SDK using the Importer machinery, a one-call path for assets the viewer understands.

API example Loader

Lenses example

How to use lenses to transform log messages before they reach the underlying sink: filtering, renaming, or shaping data.

OpenStreetMap data

Query OpenStreetMap data via the Overpass API and render it on Rerun's geospatial map view. A starting point for ad-hoc world-map context.

Map Blueprint

Minimal example

The quick-start example: generate a 3D colored cube and log it as a point cloud. The shortest path to a working visualization.

3D API example

Multiprocess logging

Log to the same Rerun viewer from multiple Python processes with `multiprocessing` and `@rr.shutdown_at_exit`.

API example

Multithreading

Log to a Rerun recording from multiple Python threads, a reference for safe concurrent use of the SDK.

API example

Live scrolling plot

A live scrolling plot built from a blueprint that updates each `TimeSeriesView`'s `time_ranges` as new data arrives.

Plots Live

Air traffic data

Display live air traffic data from INVOLI on a geospatial map view, with pyproj reprojection and GeoPandas-backed loading.

2D 3D Map CRS

Eye control

Programmatically configure and animate the 3D view camera with the Rerun Blueprint API. Useful for cinematic walkthroughs.

Eye control 3D Pinhole camera

Webpage

Embed the Rerun viewer in a static webpage so visitors can scrub the 3D scene and timeline themselves.

Webpage Javascript Web-viewer Tutorial

Ready to log your own data?

Install the SDK in one line, log a few things, and open the viewer.

Get started Try in browser