Echo-3 reconstructs real spaces from an iPhone capture

Capture a real place, then explore it

Echo-3 is SpAItial’s reconstruction model for turning a phone capture into a 3D environment. The product page describes recording a space, generating a Gaussian Splat scene and exploring or sharing the result in a browser.

The official launch announcement identifies the companion iPhone app as the starting point. It says Android support is coming later and describes an enterprise API as forthcoming. Those access paths should not be treated as interchangeable or assumed to be generally available today.

Reconstruction also completes missing regions

SpAItial says Echo-3 uses a more detailed capture than its earlier single-image or panorama approach. It aims to preserve visible parts of a place while filling regions the camera did not capture. The vendor also describes the reconstruction as having metric, real-world scale.

That combination is useful to understand before relying on the output. A completed-looking area may include inferred content rather than direct observation. A claim of metric geometry is not an independent certification that every displayed dimension is suitable for a professional measurement task.

Better coverage starts during capture

The same-publisher capture guide recommends steady movement, consistent lighting and views from multiple corners. It advises preparing doors before a multi-room scan and taking care around reflective surfaces. Floors, ceilings and areas behind furniture need deliberate coverage rather than being assumed complete.

For a first trial, choose a small room and compare the reconstruction with the actual space, including known dimensions and easily missed surfaces. Check subscription terms and the current device requirements in the app before a larger capture. Echo-3 offers a route from a phone recording to an explorable model; its usefulness for a particular workflow still depends on what was captured, what was inferred and what accuracy that workflow requires.