The new depth APIs in ARCore 1.56 use floats in meters. Understand how this reduces conversions and improves shared pipelines between mobile AR and XR.

Direct answer

ARCore 1.56 adds AcquireDepthImageMeters and AcquireRawDepthImageMeters, APIs that expose depth as float values ​​in meters. Google relates the change to the unification of data between ARCore and OpenXR. For teams that share perception, occlusion or reconstruction between mobile and headset, a common physical unit reduces conversions, but does not eliminate differences in resolution, noise, range, confidence, origin and sensor orientation.

Common unit simplifies contracts

Meters like float make interfaces and tests clearer between modules. Also document axis convention, origin, orientation and capture time.

Same unit does not mean same sensor

Cell phones and headsets can produce depth with different range, density, latency and noise. Algorithms need to consume trust and explicit tolerances.

Occlusion depends on temporal alignment

Correct depth may fail visually if image, pose and map are not synchronized. Measure latency and reprojection during movement.

Physical data requires privacy treatment

Depth maps can reveal room geometry. Minimize collection, retention and transmission and inform the user of purpose.

Nexus Reading

Adopt the new representation as an interoperability layer, with normalization, per-device testing, and clear boundaries for each use case.

FAQ

Does depth now always come from a dedicated sensor?

No. Source and quality depend on the device and supported mode; the API standardizes the representation in meters.

Does this guarantee OpenXR compatibility?

It helps align data units, but formats, coordinates, capabilities and behavior still need adaptation.

Which test is essential?

Compare absolute error, stability, holes, latency, and occlusion across representative distances and surfaces.

Essential guides to delve deeper into the decision

Primary sources and references

This editorial analysis was produced by Nexus from the official sources below, consulted on October 8, 2026. The text is original and interprets practical implications for companies.

Date reported by main source: official documentation consulted on October 8, 2026.