This video, Part 2 of a gentle introduction to Rerun, demonstrates the fundamentals of logging and visualizing time-series data through animation. The core example features four points rotating in 2D space over 360 distinct frames. The tutorial walks through the code structure necessary to set frame timestamps and continuously log changing data components (position, color, and size) within a time loop. Crucially, the video explains how scalar values (X/Y coordinates and rotation angle) are logged to automatically generate interactive time series charts. Finally, the speaker highlights key viewer capabilities, including navigation of the entity path hierarchy and accessing the raw underlying logged data via the Data Frame view, emphasizing Rerun's importance for time-critical applications like robotics and autonomous systems.
Rerun: Working with Time and Animation (Part 2)
0:00 Introduction to Time: This installment focuses on how to work with time in Rerun, specifically animating data over multiple frames, which is critical for visualizing data from robotics and autonomous systems.
1:12 Demonstration of Animation: The primary example shows four points continuously rotating in 2D space, leveraging the timeline controls for looping and frame-by-frame inspection.
2:43 Code Setup: The script initializes the Rerun viewer and sets up logging.
2:51 Static Data Logging: An origin point is logged once and marked as "static." This provides a hint to Rerun that the data will not change over time, allowing for more efficient storage.
3:12 The Time Loop: The animation is driven by a loop that iterates over 360 frames. In each iteration, the points are rotated, and attributes like color (using RGB calculation) and size (radius) are recalculated and updated.
3:55 Logging Positional Data: Inside the loop, the rotated coordinates are logged to Rerun, creating the continuously moving positions seen in the 2D view.
4:32 Logging Scalar Data: To generate the time series chart view, scalar values are logged, specifically the calculated X/Y positions of the points and the monotonically increasing rotation angle (radians).
6:10 Entity Paths: The data is organized hierarchically along entity paths (e.g., position -> point zero -> x entity), allowing the user to precisely select and visualize specific data streams.
7:29 Data Frame View: The viewer offers a Data Frame view, which allows users to inspect the raw underlying logged data (position arrays, components, etc.) for the specific moment in time selected on the timeline.
8:30 Key Takeaway: Understanding how to visualize data changes over time is essential for complex engineering tasks such as feeding machine learning models and assessing robotic perception and decision-making. Part 3 will cover transforms.