A takeoff is the act of pulling quantities off a project so you can price it — square footage of membrane, linear feet of conduit, count of rooftop penetrations, number of devices, run lengths between panels. It is the first real work of every commercial bid, and historically it has been done one of three ways: by hand off a plan set, by walking the building with a tape and a wheel, or — more recently — off an aerial photo run through photogrammetry.
A LiDAR takeoff is the same job done off a three-dimensional point cloud. Instead of reading a dimension off a drawing or stretching a tape across a roof, the quantities are measured directly from a scan of the actual structure — a dense map of points, each one a real surface the sensor saw and ranged. LiDAR ("Light Detection and Ranging") fires invisible infrared pulses, times how long each one takes to bounce back, and reconstructs geometry from the timing. The takeoff is then computed from that geometry: areas from facets, lengths from edges, counts from segmented objects.
The distinction that matters to an operator is simple. A manual takeoff measures what the drawing says. An aerial takeoff measures what the building looked like at the last flyover. A LiDAR takeoff measures the building as it stands right now, with the equipment, penetrations, and field conditions that are actually there. For commercial installation work — re-roof, retrofit, low-voltage rough-in, rooftop equipment — that gap between the drawing and the building is exactly where bids bleed.
What a LiDAR takeoff actually measures
Surface quantities. Roof membrane area, wall area, floor area, parapet length, the footprint of an equipment pad — anything that resolves to a facet or an edge in the point cloud becomes a number. On a flat or low-slope commercial roof, that means gross and net area, perimeter, and the area subtracted by every curb, drain, and penetration the scan captured.
Spatial relationships. How far the rooftop unit's curb sits from the nearest drain, where a conduit run cuts across a rack room, how much clear height there is above a drop ceiling, the offset between two panel locations. These are the measurements a tape gives you one at a time and a plan set gives you only if someone drew them. The point cloud gives you all of them at once because the geometry is already captured.
Counts and penetrations. Roof penetrations, rooftop units, devices, openings, structural obstructions — the things you have to find before you can price them. Segmentation classifies them out of the cloud so they land on the takeoff as discrete line items instead of as something the estimator has to remember to count.
What it does not measure on its own is intent. A point cloud tells you a pipe is 2 inches and runs 40 feet; it does not tell you it's the chilled-water line versus the condensate drain. That is still the operator's call. A LiDAR takeoff narrows the measurement work so the human time goes to judgment, not to counting.
How accurate is a LiDAR takeoff?
Accuracy depends on the sensor, the geometry, and the registration — how well the scan holds together as the operator moves. The direct time-of-flight LiDAR in recent iPad Pro and iPhone Pro models is good enough to bid commercial work from, with honest limits. It captures high-resolution depth at close range and coarser depth out to the tens of meters, registered against the device's motion sensors.
On commercial flat and low-slope geometry, Forge publishes a gross-area figure of approximately ±0.8% against certified manual takeoffs and total-station survey, widening to about ±2% on tighter, more complex geometry where the depth-resolution gap is larger. That is currently published as a stated methodology and sample plan at /proof/hyperion-accuracy, not as an independently verified benchmark — independent third-party verification is not yet complete, and the figure stays hedged until it lands. We cite it the way the proof page does, with dated updates so a reader can see what changed and when.
The honest limits are worth stating plainly, because they decide whether a LiDAR takeoff is the right tool for a given job. Glass and mirrored surfaces don't register reliably — the infrared pulse passes through or scatters. Direct sun on a hot dark membrane adds noise; a 7 a.m. scan beats a noon scan on the same black TPO roof. Sub-millimeter detail is out of reach; LiDAR is a structure-scale instrument, not a metrology gauge for a bolt-pattern offset. And geometry at longer range degrades, so large properties get stitched from multiple sessions, and stitching carries its own error budget.
LiDAR takeoff vs manual takeoff
A manual takeoff — tape, wheel, and a plan set marked up by hand — is still the right call in a few situations, and it's worth being straight about them. If the geometry is genuinely sub-millimeter critical, if the surfaces are mostly glass or polished metal that LiDAR can't see, or if you're working a clean new-construction plan set where the drawings are trustworthy and complete, a careful manual takeoff is accurate and defensible. Plenty of excellent estimators have built whole careers on it.
Where the manual method breaks down is time and field truth. Walking a building with a tape is slow, it's error-prone the moment a number gets transcribed wrong, and it only captures the measurements someone thought to take. Come back to the office with a question the tape didn't answer and you're driving back to the site. A LiDAR takeoff captures the whole structure once; every measurement you didn't think to take in the field is still in the cloud when you need it.
The other manual-takeoff cost is the handoff. A marked-up plan has to be re-keyed into the estimate by hand — another place for a digit to flip. A LiDAR takeoff in a system like Forge skips the re-keying entirely: the geometry flows straight into the line items. The measurement and the estimate are the same act, not two acts with a transcription error between them.
LiDAR takeoff vs aerial-imagery takeoff
Aerial-imagery takeoff providers — Nearmap, GAF QuickMeasure, EagleView, and the aerial design pipeline inside tools like Aurora Solar — start from a satellite or drone photo, identify edges, and reconstruct geometry from photogrammetry plus a known camera position. For the right job they are genuinely excellent: if you need a roof measured on a building you can't safely or easily access, or you want a report for a whole portfolio of addresses without sending anyone to site, aerial is the faster path and often the only practical one. Credit where it's due.
Aerial-imagery has two structural limits, and they're not knocks on any one vendor — they're inherent to measuring from a photo taken from above at some point in the past. First, the imagery is only as current as the last flyover, which can be many months old. A building that was reroofed last summer measures the old roof. Second, an overhead photo can't see what's behind a parapet, under an overhang, inside a courtyard, or anywhere on a vertical face — and on commercial geometry, the places the contractor most cares about measuring are frequently the places aerial can't reach.
A LiDAR takeoff is ground-truth: it measures the building the operator walked through this morning, including the geometry an aerial photo never sees. The trade-off is honest — LiDAR requires someone on site with the device, and aerial does not. For commercial re-roof and retrofit work where the building has changed and the hard-to-see geometry is the whole point, that on-site requirement is a feature, not a tax. For a fast portfolio sweep of accessible roofs, aerial may still be the better tool. Forge's position is in-field LiDAR ground-truth over stale aerial imagery for the work that lives or dies on field accuracy — not a claim that LiDAR wins every job.
How Forge produces a LiDAR takeoff (Hyperion)
Hyperion is Forge's iPad and iPhone LiDAR scanning system, and a LiDAR takeoff is what it produces. The operator walks the structure; the sensor captures geometry, materials, and penetrations into a fused point cloud registered as they move. On-device segmentation classifies facets, zones, and obstructions out of the cloud. The federation kernel — the shared architecture every Forge module is built on, not a stack of integrations bolted together — pulls trade assemblies, waste factors, and material costs. The result is a line-item estimate generated from the scan rather than re-entered by hand.
That last part is the point of doing the takeoff inside the platform rather than as a standalone report. In Forge, the takeoff doesn't get exported and re-keyed somewhere else — it lands directly in the estimate, which lives in the same chest as the CRM record, the schedule, the documents, and Treasury's payroll and change-order ledger. The measurement becomes a bid, the bid becomes a project, and the project becomes labor cost and certified payroll without anyone retyping a number along the way. The whole point of one unified system is that the takeoff is the first link in a chain that never breaks for a copy-paste.
Hyperion was built first for commercial roofing — flat and low-slope membrane is the geometry LiDAR reads most cleanly — and it ships today across Forge's active verticals: roofing, security and fire, AV and low-voltage, solar, HVAC, and electrical. The spatial-intelligence approach behind it, including photogrammetric LiDAR augmentation and solar PV design from 3D capture, is the subject of provisional USPTO patent applications assigned to Dominus IP Holdings LLC.
Hyperion is a $399/mo module on Forge Core ($499/mo flat) — included in the Working Stack and Full Platform packages, and in the Charter founder program. It is not sold separately — the takeoff instrument and the operating system it feeds are one product, because a takeoff that has to be exported to be useful is the problem Forge was built to end.