PlotJuggler 4 beta cuts memory 5x, loads MCAPs 4x faster
New beta handles 6 pointclouds and 3 videos on just 50% of a single core.
PlotJuggler 4 (beta) is here, and it’s a dramatic leap for robotics data analysis. Facontidavide has brutally optimized the tool: playing a compressed MCAP with lazy file loading, rendering 3 videos and two 3D scenes (6 pointclouds total) uses only 50% of a single CPU core. On the core side, a refactored data engine reduces memory usage by up to 5x when loading large datasets, and a new parallel MCAP loader is about 4x faster for compressed files. Support now includes 2D images, compressed video (H264, AV1), depth, markers, and real-time WebRTC streaming. For 3D, it handles meshes, occupancy grids, TF2, 3D markers, and pointclouds—including compressed formats like Draco and Cloudini—plus multi-camera controls similar to RViz.
Beyond raw performance, PlotJuggler 4 introduces a marketplace for Extensions (think VSCode) to share plugins, and integration with Mosaico for direct cloud data access. While still in beta with rough edges, the update delivers massive speed and efficiency gains for ROS developers working with multimodal data. The combination of lower memory, faster loading, and rich 2D/3D visualization—all from a single lightweight application—makes it a compelling upgrade for anyone debugging or analyzing complex robotic systems.
- Uses 5x less memory on large datasets; loads compressed MCAPs 4x faster with parallel loader.
- Renders 6 pointclouds, 3 videos, and 2 3D scenes using only 50% of a single CPU core.
- Adds real-time WebRTC streaming, image rectification, compressed pointclouds (Draco, Cloudini), and cloud integration via Mosaico.
Why It Matters
Robotics engineers can now analyze massive multimodal datasets in real time without bogging down their machines.