Rerun 0.25 – transparency, table filtering and initial MCAP support (github.com)

🤖 AI Summary
Rerun 0.25 is a feature-focused release that brings three visible improvements: prettier syntax highlighting and table UIs, initial support for transparent rendering of 3D geometry, and experimental MCAP import for sensor/robot data. The release also advances the table widget with filtering (boolean, integer, floating point and string columns) for Arrow dataframes used by ViewerClient.send_table() and the Rerun server browser, plus UI niceties like full-text string filtering, an Arrow data tree, and other viewer UX refinements. These make large interactive recordings and telemetry tables far easier to explore and query. On the 3D side, the viewer now supports alpha/transparency for Mesh3D and a range of primitive archetypes (Boxes3D, Ellipsoids3D, Capsules3D, Cylinders3D) — with alpha sourced from shape color components, Mesh3D albedo factor, and Asset3D albedo where supported. MCAP support is experimental and partial: a basic loader and low‑level Arrow conversion of protobuf-encoded MCAP messages have been integrated (subset of ROS2/CDR and protobuf messages; scalar sensor parsers added), enabling import of robot/sensor recordings while development continues. Notable breaking changes: removal of some CLI flags (--serve, --drop-at-latency, -o), deprecation of Python 3.9, and a split of AnyValues into AnyValues and DynamicArchetype—see the migration guide before upgrading.
Loading comments...
loading comments...