FORGE/GLOSSARY/IPAD LIDAR SCANNING
.Glossary — ipad lidar scanning

What is iPad LiDAR scanning, and what does it actually measure?

The depth sensor in iPad Pro and iPhone Pro is good enough to bid commercial roofs from. Here's what it does, what its limits are, and how Forge uses it.

ENTRY
iPad LiDAR scanning
LAST REVIEWED · APRIL 27, 2026

LiDAR — "Light Detection and Ranging" — is a measurement technique that fires invisible infrared pulses at a surface, measures the time each pulse takes to bounce back, and uses the timing to build a three-dimensional map of what it sees. It's been used in surveying, autonomous vehicles, and aerospace for decades. What changed in 2020 is that Apple put a small one in the iPad Pro.

The sensor in iPad Pro M-series and recent iPhone Pro models is a direct time-of-flight (dToF) LiDAR, paired with the device's IMU and ARKit's fused-depth pipeline. It captures depth at room scale and slightly beyond — useful range is roughly 0.5 to 5 meters of close-coupled high-resolution depth, with rougher depth out to about 30 meters when the scan is registered against IMU drift.

What it can measure honestly

Geometry: walls, floors, ceilings, surface area, opening sizes, parapet heights, penetrations through a roof membrane. Anything that has a hard, mostly-opaque surface inside the sensor's range registers cleanly.

Spatial relationships: how far the equipment pad sits from the parapet, where the rooftop unit's curb is relative to the drain, how the conduit run cuts across the rack room — all become measurable in the captured point cloud.

On commercial flat and low-slope roofs, the iPad LiDAR sensor delivers gross-area accuracy of approximately ±0.8% against EagleView and certified manual takeoffs in Forge's stated methodology. On small-residential pitched roofs that figure widens to ±2% — the depth-resolution gap on residential geometry is wider, and the published number stays honest about that.

What it can't

Glass and mirrored surfaces don't register reliably; the infrared pulse passes through or reflects in ways the time-of-flight calculation can't disambiguate.

Direct sun on a hot membrane can introduce noise — the sensor's ambient-IR rejection is good, not perfect. Scans at noon in July on a black TPO roof are noisier than scans at 7 a.m. on the same roof.

Sub-millimeter detail is out of reach. The LiDAR is built for room and structure scale; it won't measure a hairline crack or a bolt-pattern offset to surveyor precision.

Long-range geometry past about 30 meters degrades; for properties bigger than a single LiDAR session can capture, scans are stitched, and stitching introduces its own error budget.

How it differs from aerial imagery

Aerial-imagery providers — EagleView, Nearmap, GAF QuickMeasure, Aurora Solar's design pipeline — start from a satellite or drone photo, identify roof edges, and reconstruct a 3D model from photogrammetry plus a known camera position. The result is a measurement report typically delivered the next business day.

Aerial-imagery has two structural limits. First, the imagery is as old as the last flyover (typically 6–18 months). A roof that was reroofed last summer measures the old shingles. Second, aerial-imagery can't see what's behind a parapet or under an overhang or inside a courtyard; commercial geometry where the contractor cares most about measurement precision is the geometry hardest for aerial to capture.

iPad LiDAR captures the geometry the contractor walked through this morning. The roof under their feet is the roof being measured. For commercial re-roof and retrofit work, that distinction is the difference between bidding from the truck door and bidding from a stale aerial report.

How Forge uses it (Hyperion)

Hyperion is Forge's iPad and iPhone LiDAR scanning system. The operator walks the structure; the sensor captures geometry, materials, and penetrations to a fused point cloud registered in real time. On-device segmentation classifies facets, zones, and obstructions. The federation kernel pulls assemblies, waste factors, and material costs. A line-item estimate writes itself — typically in 60–90 seconds in field conditions.

The methodology, sample plan, and the limits of the published accuracy figure are documented at /proof/hyperion-accuracy. Independent third-party verification is not yet complete; the published figure stays hedged until it lands. The page tracks updates with stated dates so the reader can see what changed and when.

Hyperion is included in Forge Master tier and in the Charter Member program. It is not sold separately.

.Related on the site

Where this entry connects.

THE SPINE

The needs have been the same for 4,000 years.
What we name them changes. The work doesn't.

← Back to the glossaryApply for Charter →
ENTRY · IPAD LIDAR SCANNING · LAST REVIEWED APRIL 27, 2026