Rerun is a versatile SDK, time-series database, and visualizer designed for logging and visually exploring computer vision and robotics data over time. It aids in debugging and understanding the internal state and data of systems with minimal code. Rerun is particularly useful in fields like robotics, spatial computing, 2D/3D simulation, and finance for verifying, debugging, and explaining data.
The tool supports multimodal data streams, allowing users to log data such as tensors, point clouds, and text, and correlate input, intermediate states, and outputs from multiple sources. It provides powerful interactive visualizations of live and recorded streams, enabling users to time travel through data dynamically aligned to any timeline. Rerun's composable data model allows users to express the semantics of their data, build custom layouts, and tailor visualizations either through code or interactively within the UI.
Rerun is built with a high-performance, cross-platform 2D/3D visualization engine using Rust and wgpu, capable of running natively or in the browser with Wasm. It handles serialization, transport, out-of-order and multi-process ingestion, and multi-timeline indexing. The tool also understands spatial relationships by default, supporting projections and transform hierarchies to connect spatial data.
The SDK is accessible in multiple languages, including C++, Python, and Rust, with quick start guides available for each. Rerun is open source and continually evolving, with regular updates and new features being added.
Pricing
Rerun offers a free and open-source SDK for visualizing multimodal data, licensed under MIT and Apache 2, targeting individual developers. A commercial product is under development, aimed at teams handling intelligent processing of multimodal data, providing robust sharing and collaboration features from prototype to production. For more information on the commercial offering, contact the team.