XRReferenceSpace

XRReferenceSpace is a coordinate system used in WebXR to interpret pose data during an immersive session. It defines the reference frame in which positions and orientations of the headset, controllers, and other input sources are reported, enabling consistent placement of virtual content relative to the real world or the user.

A reference space is created by calling a session method such as requestReferenceSpace with a specific type. The common types are viewer, local, local-floor, bounded-floor, and unbounded. Each type establishes how the origin and orientation are defined and how the environment is referenced.

The viewer space is centered on the user’s eyes and moves with head motion; poses in this space are useful for gaze-related calculations and overlays tied to the user’s view. The local space uses a fixed origin established at the start of the session and provides a stable, world-relative frame. The local-floor space is similar but aligns the origin to the real or estimated floor, enabling floor-relative placement of objects and gravity-aware interactions. The bounded-floor space extends local-floor by including a defined play area boundary, which helps constrain user movement in room-scale experiences. The unbounded space is intended for AR or open-world scenarios where content is positioned relative to an infinite, device-anchored coordinate system.

In each animation frame, developers obtain poses by requesting them relative to a chosen XRReferenceSpace, typically via frame.getPose or frame.getViewerPose. Availability and behavior of specific spaces may vary by device and platform.