3D Capture Pipelines for Inspection & Documentation

Laan Labs builds complete capture pipelines for organizations that need a reliable record of what something looked like and what condition it was in: insurers documenting property for policies and claims, utilities and infrastructure owners surveying sites and assets in the field, operators inspecting aircraft parts, and owners tracking buildings and equipment over time. A pipeline takes a capture made on an ordinary phone or tablet by someone who is not a 3D specialist, processes it automatically, and produces a measured, annotated record that an adjuster, inspector or engineer can open in a browser, with every finding tied to the place it was seen.

Condition documentation today is mostly photographs in a folder and notes in a form. Photos lack scale and context, notes are subjective, and neither can answer a question nobody thought to ask on site. A 3D capture changes that: the whole space or asset is recorded once, measurements can be taken afterwards, and a reviewer who was never there can look at the same evidence as the person who was. For an insurer, that means better estimates before a policy is written and faster, less contested claims afterwards. For an inspector, it means fewer site visits and a record that holds up later.

For sites and field assets, where something is matters as much as what it looks like. Laan Labs pipelines tie each scan to a surveyed position from a GNSS receiver, so a LiDAR mesh, point cloud or photo scan lands in real-world coordinates instead of floating in its own local frame. The same workflow runs at every accuracy tier: the phone's own location at 3 to 5 meters for reconnaissance and inventories, an external sub-meter receiver for GIS and utility mapping, or an RTK rover with NTRIP corrections at 1 to 3 centimeters for control and stakeout. Because the capture carries coordinates, it drops straight into GIS and CAD alongside the rest of the project, and it can be found again: augmented reality turns stored coordinates into pins in the camera view, with live direction and distance, so a crew can walk back to a valve, manhole, marker or inspection point months later. Laan Labs has hands-on experience with the hardware as well as the software, having integrated GNSS receivers and survey equipment from Emlid, Leica, Trimble and Bad Elf, among others. Its open-source OpenSurvey3D app shows this scan, anchor and find workflow end to end.

Laan Labs has built these pipelines in production. It delivered a custom scanning solution used by an insurance company for policy estimation and claims, and built RadianceView, which combines mobile scans, LiDAR, Gaussian splats and high-resolution photographs into one inspection record and has been used to document end-of-life airliners for parts reuse. The same components are assembled and adapted for each client's assets, staff and systems.

What we do

Guided capture for field staff

iPhone and iPad apps that walk an adjuster, inspector or technician through a capture, check coverage and quality before they leave the site, work without a network connection, and attach the job, policy or asset reference at the source.

Automated processing

Uploads are reconstructed without manual steps into the right representation for the job: LiDAR meshes and floor plans for rooms and buildings, photogrammetry for measurable detail, and Gaussian splats for a photorealistic view of the whole scene, aligned at true scale.

GNSS and RTK georeferencing

Every capture locked to a surveyed observation before the camera starts, using the device's location, a Bluetooth GNSS receiver or an RTK rover with NTRIP corrections. Fix state, estimated accuracy, satellites, correction age and antenna offsets are recorded with each observation, so a centimeter-grade fix is never confused with a meter-grade one.

Survey hardware integration

Direct integration with the receivers and instruments field teams already own, including equipment from Emlid, Leica, Trimble and Bad Elf: Bluetooth and Wi-Fi connections, NMEA streams, NTRIP corrections, antenna height and offset handling, and the practical differences between vendors that only show up on site.

GIS and CAD integration

Points, lines, areas and scans collected with the client's own feature codes, attributes and layers, and exported as GeoJSON, CSV, DXF and LandXML in the project's coordinate frame, or imported from existing GIS and design data in CSV, GeoJSON, GPX and KML, so field work and the office system of record stay in step.

AR asset finding and revisits

Stored coordinates shown as pins in the camera view, with bearing and distance to a chosen target, to guide staff back to a surveyed asset or inspection point for a repeat capture. AR placement inherits GNSS and compass uncertainty, so it is used to reach an asset, with numeric stakeout and check shots where a position has to be set or accepted.

Measurement and extraction

Dimensions, areas and volumes taken directly on the capture; floor plans and room geometry extracted automatically; damaged or changed regions marked up with their size and location for estimating.

Scan to CAD with AI

Raw scans converted into structured, editable geometry rather than left as a mesh: walls, openings and rooms as floor plans and DXF, and parts and equipment as exact solid models on a CAD kernel. Vision and language models interpret what is in the scan, recognize components and standard features, and propose the CAD operations that rebuild them, with an engineer reviewing and correcting the result instead of redrawing from scratch.

AI-assisted findings

Computer vision and language models that pre-screen captures for damage, defects and missing items, draft the description of each finding, and flag what a human reviewer should look at first, with the reviewer making the call.

Condition records over time

Repeat captures of the same property, part or structure aligned to each other so change is visible: before and after a loss, at intake and at return, or across scheduled inspections.

Review, reporting and sharing

A browser-based record that an adjuster, engineer, contractor or customer opens with nothing to install, where they can look around, measure, read the findings and export a report. Access is controlled per case. Field work can carry a QA report of check shots against known coordinates, so reviewers see observed accuracy rather than a receiver's own estimate.

Integration with existing systems

Captures, measurements and findings delivered into the claims, policy, asset management or maintenance systems an organization already runs, in standard formats, with single sign-on and audit trails where required.

How we work

Projects begin with one real workflow, such as a property claim, a pre-policy survey or a parts inspection, followed end to end with the people who do it. Laan Labs then builds a pilot pipeline for that workflow in weeks and runs it on live cases alongside the existing process, so its effect on time, cost and accuracy is measured before a wider rollout.

Pipelines can be delivered as a solution built on RadianceView, as a white-label app and portal under the client's brand, or as licensed components integrated into existing software. Sensitive captures can be processed on the client's own infrastructure. See 3D capture technology licensing for the licensing options.

Products built with this work

• RadianceView — Inspection-grade visual records in a browser
• OpenSurvey3D — Open-source 3D scanning, GNSS surveying and AR asset finding for iPhone and iPad
• OpenPlan3D — Open-source floor plans from iPhone LiDAR scans, with DXF export
• OpenShape3D — Open-source solid-modeling CAD on the OpenCASCADE kernel

Common questions