C-DRONE GUIDE · 15 JUNE 2026
Drone photogrammetry and BIM: the construction site measured to the centimetre
Turning three hundred aerial photos into a measurable digital twin of the site: that is the promise of drone photogrammetry, and it has been kept for years — provided you understand what drives accuracy, what the deliverables are worth, and how they connect to the BIM model. A practical guide for site managers, surveyors, project managers and clients.
The principle: photos that become measurements
Photogrammetry reconstructs the 3D geometry of a scene from overlapping photos: the software identifies the same points across dozens of images taken from different angles and derives their position in space, exactly as our binocular vision estimates distances. A typical mapping flight — automatic grid trajectory, 75 to 80% image overlap, constant height — produces 200 to 400 photos for a one-hectare site, then processed in a few hours of computation.
Resolution is driven by flight height: that is the GSD (ground sampling distance), the pixel size on the ground. At 60 m with a recent sensor you get a GSD of 1.5 to 2 cm — every orthophoto pixel represents less than 2 cm of terrain. Positional accuracy depends on georeferencing: an RTK drone such as the Matrice 4E corrected by a base station or a network (Centipede RTK, widely used in France, or commercial networks) attaches centimetre coordinates to every photo; ground control points (GCPs) measured with GNSS verify and constrain the computation. RTK plus control points is the professional standard: absolute accuracy of 1 to 3 cm horizontally, 2 to 5 cm vertically — enough for most site uses, but not a replacement for the licensed land surveyor on boundary matters, which remain their legal monopoly in France.
The deliverables and what they enable
Four deliverables cover most needs. The orthophoto: a geometrically corrected aerial image that overlays onto drawings — the site's base map, on which distances and areas are measured. The DSM/DTM (digital surface and terrain models): the site's elevation, the basis of cut-and-fill calculations. The point cloud: millions of colourised XYZ points, the equivalent of an aerial laser scan, usable in trade software. Finally the textured 3D mesh, more visual than metrological, valuable for project communication and client meetings.
| Deliverable | Common formats | Main use |
|---|---|---|
| Orthophoto | GeoTIFF, georeferenced JPG | plans, 2D measurements, visual tracking |
| DTM / DSM | GeoTIFF raster, DXF contours | earthworks, hydraulics |
| Point cloud | LAS/LAZ, E57 | as-built checks, sections, CAD |
| 3D mesh | OBJ, 3D Tiles | communication, virtual tours |
| Volume reports | PDF report + reference surfaces | stockpile and excavation quantities |
Volume computation deserves a mention: measuring aggregate stockpiles or excavated volumes by drone has become the norm in quarries and earthworks — accuracy around 1 to 3% on a stockpile, versus far larger errors from truck counting, and a full quarry inventory takes one morning without stopping the machines.
From point cloud to BIM model
The link with BIM runs both ways. "Scan-to-BIM" first: the point cloud serves as the geometric reference for modelling existing conditions in Revit or Archicad — indispensable in renovation, where original drawings always lie a little. The drone covers the roofs, façades and surroundings that terrestrial scanners reach poorly; aerial and terrestrial clouds are routinely merged for full coverage. The resulting model, generally at LOD 200 to 300 for the envelope, is exchanged as IFC.
Then the control direction, the biggest value creator during construction: overlaying the "as-built" point cloud on the "as-designed" model to detect deviations — a buried network laid out of tolerance, a misplaced wall, a non-compliant platform level. Caught before pouring or backfilling, a deviation costs hundreds of euros; discovered six months later, tens of thousands. Monitoring platforms (Pix4D, Propeller, DroneDeploy and equivalents) automate this comparison at every flight and publish the measurements to the whole team in a browser: the site manager measures a distance or a volume without installing anything. This shift from "the nice monthly picture" to "shared management data" is what genuinely brings the drone into the site's BIM process.
Organising site monitoring: frequency and method
Effective monitoring rests on repeatability: same flight paths, same parameters, same permanent control points sealed around the site perimeter, so successive deliverables are rigorously comparable. The typical frequency is monthly during structural works and earthworks, tightened to fortnightly or even weekly during phases with heavy earth movement. Each campaign follows the same cycle: flight in the morning (even light, quieter site), same-day processing, deliverables online within 24 to 48 h.
On the regulatory side, a construction site is often a favourable environment: flying over a controlled ground area — the fenced site, with informed personnel and controlled access — fits well within the specific category's scenarios, and in rural areas the open category sometimes suffices. The hard cases remain urban sites under a CTR or in city centres (prefecture notifications for each campaign, unless a framework agreement exists) and co-activity with tower cranes, which requires coordination with crane operators and a flight plan that keeps clear of jibs and lifting zones. Writing the drone into the site's health and safety plan, still uneven, is becoming standard practice among major French contractors.
2026 prices and photogrammetry quote pitfalls
Prices observed in France in 2026: €700 to €1,500 for a one-off mapping mission up to 5 ha (RTK flight, control points, orthophoto + DSM + point cloud), €1,500 to €2,500 beyond that or with reinforced accuracy requirements; €350 to €900 per visit for contracted recurring monitoring; €500 to €900 for a quarry stockpile inventory. Scan-to-BIM is priced case by case — allow €2,000 to €6,000 for the acquisition and LOD 200 modelling of an average building, modelling being the bulk of the cost. Monitoring platform subscriptions (€100 to €500/month by volume) come on top or are bundled in.
Three pitfalls recur in quotes. No quantified accuracy: require an absolute accuracy stated in centimetres, verified with independent check points and recorded in a processing report — "centimetre accuracy" without the report commits to nothing. The GSD/accuracy confusion: a 2 cm GSD in no way guarantees 2 cm accuracy if georeferencing is neglected. And data ownership: require delivery of the source files (geotagged photos, LAS cloud) and not just access to a proprietary platform — the day the subscription ends, your site data must remain yours. A serious provider accepts all three clauses without argument.