Loggable Data Types

Rerun comes with built-in support for a number of different types that can be logged via the Python and Rust Logging APIs and then visualized in the Viewer.

The top-level types are called archetypes to differentiate them from the lower-level data types that make up the individual components. For more information on the relationship between archetypes and components, check out the concept page on Entities and Components.

In Python every archetype is typically backed by one or more function calls. In contrast, the Rust API works by building up entities of a given archetype explicitly by assembling the required components.

Spatial Archetypes

The spatial archetypes represent 2d and 3d spatial data. These types have some notion of a coordinate system and generally support spatial transformations. All of these types can be visualized by the Spatial space view.

Image & Tensor Archetypes

Image and tensor archetypes all build on top of a common tensor component. The tensor component is a multi-dimensional generic container for arrays of data. Images are restricted to tensors of rank 2 or rank 3; these can be viewed in the Spatial space view. Generic tensors of greater rank can only be viewed in the specialized Tensor space view.

Other Archetypes

  • Scalar: a single scalar / metric value. Can be viewed in the TimeSeries space view.
  • TextEntry: captures text data. Can be viewed in the Text space view.
  • AnnotationContext: not viewed directly, but provides classes, labels, and connectivity information for other entities.