A $999 iPhone Scan Found 38 Framing Errors Before Drywall. Your $400 Inspector Found 4.

Last month, a production builder in Austin ran an experiment. Before drywall went up on a 2,400-square-foot spec house, he handed his project manager an iPhone 15 Pro and the Polycam app. Walk every room, scan every wall. Twenty-two minutes later, the PM had a point cloud of the entire frame.

They compared it against the architectural plans, and thirty-eight deviations exceeded 1/4 inch. Nine walls were out of plumb by 3/8 inch or more. Two bearing walls had studs at 17.5-inch centers instead of 16. One header sat 3/4 inch too low for the specified window unit.

His code inspector had visited that same morning, signed off with four notes on the report, and every single one was about fire blocking.

What Your Inspector Actually Checks

Pre-drywall inspections are mandatory in most jurisdictions, but inspectors verify life-safety items: fire blocking between floors, electrical rough-in, plumbing pressure tests, insulation installation, and structural connector placement. Important work, all of it, but none of it involves holding a tape measure against a stud wall and checking whether the framer hit his marks.

No section of the IRC or IBC mandates dimensional tolerances for light wood-frame construction. WoodWorks confirms this gap explicitly, and the advisory guidelines that do exist from NAHB (3/8 inch in 32 inches vertically) and the Gypsum Association (framing faces within 1/8 inch of the plane of adjacent members, per GA 216/ASTM C-840), but they are advisory, and nobody enforces them.

A wall that is half an inch out of plumb passes code. It also creates a visible gap behind your kitchen cabinets, a wavy tile line in your bathroom, and a trim carpenter who bills by the hour while shimming every piece of baseboard.

Fix Costs Multiply After Drywall

Framing corrections follow a brutal escalation curve:

A $150 shim job on Tuesday becomes a $6,000 tile demo in November, and that math gets ugly fast. Multiply it across nine walls in the Austin house and you are staring at $50,000 in avoidable rework, all of which could have been caught with $1,200 worth of iPhone time on a single morning.

What LiDAR Actually Catches (and What It Misses)

iPhone Pro LiDAR achieves 1 to 5 centimeter accuracy indoors, according to a 2023 MDPI study on smartphone LiDAR for building documentation. Good enough to flag a wall 3/8 inch out of plumb over 8 feet. Not accurate enough to verify the Gypsum Association's 1/8-inch standard.

For tighter tolerances, professional scanners close the gap. A Leica BLK360 ($20K) delivers 1 to 6 millimeter accuracy. A Matterport Pro3 ($6K) splits the difference between consumer and survey-grade. Scan-to-BIM comparison software from OpenSpace and Buildots can overlay the point cloud against a digital model and auto-flag every deviation beyond a set threshold.

Limitations matter, and they are real: LiDAR reads surfaces, not structure, so it cannot tell you if a stud is warped internally, if a nail plate is missing behind insulation, or if a connection uses the wrong fastener. Those are still your inspector's job, and always will be. Scanning catches geometric errors while inspection catches safety errors, which makes them complementary rather than competitive.

Why Nobody Does This on Houses

Commercial construction embraced reality capture years ago, and the gap between those sites and residential job sites gets wider every quarter. Buildots equips superintendents with hardhat-mounted 360-degree cameras that auto-compare to BIM models. OpenSpace auto-maps walk-throughs against plans and flags deviations before the super finishes his coffee. Even production home builders like Meritage and David Weekley have piloted camera-based QC programs on their tract developments.

Residential custom and semi-custom builders have not followed, and the reasons are frustratingly structural rather than technological.

First, most residential projects have no BIM model to compare against because framers work from 2D plans, and without a digital reference, a scan produces a pretty point cloud with nothing to measure it against except manual spot-checks and a good eye.

Second, commercial scan-to-BIM software prices out single-family work entirely: OpenSpace runs roughly $2,500 per month, Buildots prices per project at commercial scale, and no vendor in the space currently offers a $200-per-house residential tier that would make economic sense on a one-off custom build.

Third, the tolerance that matters depends on the finish, and framers do not know what finish goes where. A wall receiving paint can tolerate 1/4 inch of variation in 10 feet. A wall receiving large-format porcelain tile needs 1/8 inch in 8 feet. Whoever framed it used the same technique on both walls.

What a Builder Should Do Right Now

If you are a GC building custom or semi-custom homes in the $500K-and-up range, here is a protocol that costs under $1,500 and catches most of what matters:

1. Buy one afternoon of a scan technician's time. Local survey firms increasingly offer single-visit LiDAR scans for $500 to $1,500. Ask for a deliverable point cloud in .e57 or .las format.

2. If you cannot justify the cost, use an iPhone Pro. Polycam and SiteScape both export point clouds that you can measure against plans. Walk each room slowly, holding the phone at consistent height. Accuracy drops in large open rooms and near reflective surfaces, but closets, hallways, and bathrooms scan well.

3. Spot-check the scan against critical walls. Without BIM software the overlay is manual, so focus on walls receiving tile, stone, or cabinetry. Measure plumb at top and bottom of each wall segment in the point cloud. Flag anything beyond 1/4 inch in 8 feet.

4. Hand the deviation list to your framing sub before drywall. A good framer can fix 30 deviations in four hours for $400 to $800 in labor, which is less than a single post-tile callback will cost you.

5. For homeowners: if your builder does not scan, request it yourself. You are entitled to inspect your home before drywall. Bring a tape measure, a 4-foot level, and an iPhone. Or hire an independent scan for $500 to $1,000. For a $1M+ build, this is insurance that costs less than a single appliance upgrade.

What I Could Not Verify

That Austin builder's 38-deviation count comes from a single house, and I want to be upfront about that. I have no controlled study comparing LiDAR scan results against inspector findings across a statistically meaningful sample of residential projects. Commercial data from Buildots and OpenSpace shows 60-80% defect detection improvements over manual walk-throughs, but those numbers come from multi-story concrete and steel projects with BIM models, not stick-frame houses built off 2D plans.

iPhone LiDAR accuracy varies significantly with environmental conditions: direct sunlight degrades it, large open rooms with few reference surfaces reduce precision, and the 1 to 5 centimeter range from the MDPI study represents controlled indoor conditions rather than a job site with sawdust in the air and tarps over the windows.

Residential-specific AI comparison software does not exist yet at an accessible price point, and until someone builds a $200-per-house scan-to-plan tool that works with 2D drawings instead of BIM, the workflow I described above remains partially manual. It works, but it requires a person who knows what they are looking at.

Finally, rework cost estimates vary wildly by market, and my $50 to $200 pre-drywall range and $5,000+ post-tile range reflect conversations with Bay Area and Texas contractors rather than a nationally representative survey. Your numbers will differ depending on local labor rates, material costs, and how much your tile guy charges to come back. But the ratio holds everywhere I have looked: fixing framing after finishes costs 25 to 100 times more than fixing it before drywall, and that math does not care what zip code you are in.