An iPhone LiDAR scan begins with depth information, but the file you receive depends on the app’s capture mode and processing. A point cloud contains spatial samples; a mesh connects surfaces; a structured room model represents recognised elements. These outputs serve different tasks and should not be treated as interchangeable 3D scans.
How does depth become geometry?
On supported devices, ARKit exposes scene depth and scene geometry to apps. Apple’s ARKit depth and scene-geometry explanation distinguishes depth maps from a mesh representation of the surrounding environment. The app then decides which data to keep, process, display and export. A LiDAR-capable device alone does not establish what an app’s export contains.
A point cloud can preserve samples without connecting them into faces. A mesh provides a connected surface representation, which is useful for visual review and many 3D tools. Neither representation guarantees that every part of the scene was observed or that a measurement has a known tolerance.
Which Voxelio modes use depth or LiDAR?
Mesh Mode captures room and environment geometry with OBJ + MTL export. Point Cloud Mode captures coloured AR depth samples and exports PLY. Each requires the corresponding device capability.
Room Plan uses Apple RoomPlan to identify structured room elements and export USDZ + JSON. It answers a different question from a freeform surface scan: which room components were recognised? Its output should not be confused with a complete measurement survey.
Why is a camera capture mode different?
Voxelio also provides 3D Objects for supported on-device object photogrammetry, Gaussian Splatting for supported small-scene capture, training and export, and Dual Camera for video and camera data. Dual Camera footage is not automatically a reconstructed mesh or trained splat.
Choose by the deliverable rather than assuming that every camera mode needs LiDAR or produces the same format. The six-mode workflow guide compares the current options and their requirements.
What makes a scan incomplete?
Surfaces hidden behind furniture or outside the capture path cannot be inspected as if they were observed. Movement in the scene, poor tracking, difficult surfaces and incomplete coverage can leave gaps or distorted areas. A visually plausible preview is not evidence that the geometry is complete or metrically reliable.
Inspect the file for the task you actually intend to perform. For a textured room reference, check the geometry as well as the appearance. For point-based analysis, inspect coverage, distribution and coordinate scale. For structured room work, check the recognised elements against the room.
Is iPhone LiDAR a guaranteed measurement tool?
No universal accuracy figure applies to every device, app mode, scene and capture. A scan may be useful as a spatial reference while remaining unsuitable for a particular tolerance or safety-critical measurement. Compare relevant dimensions with an independent reference before relying on them, and use appropriate specialist equipment where the task requires it.
What should you read next?
Follow the room-mesh capture checklist when OBJ surfaces are the goal. Follow the point-cloud capture and export checklist when you need PLY samples. All Voxelio modes and export are free; capture remains subject to the app’s weekly scan allowance, which resets each week.