C‑DRONE
Top-down orthophoto-style view of farmland parcels captured by drone

C-DRONE GUIDE · 12 AUGUST 2026

Drone photogrammetry deliverables: orthophoto, DSM/DTM, point cloud, 3D mesh

Ordering "a drone photogrammetry survey" is not enough to know what you will actually receive. From a single flight, photogrammetric processing can produce several radically different files — a flat image for mapping, a surface model for computing a volume, a point cloud or a mesh to feed a 3D model — each with its own format, use and level of accuracy. A design office, a surveyor, a local authority or an industrial site that does not specify the expected deliverable in its brief risks receiving a file that does not match its real need, or paying for a precision the project has no use for. Here is what each deliverable represents, what it is for, and what actually determines its accuracy.

Published on 12 August 2026, reviewed on 18 August 2026 — regulations in force as of August 2026.

One flight, several possible deliverables

A company that orders a drone photogrammetry mission rarely receives a single file. From the same flight and the same set of photos, processing can produce several distinct products, each answering a specific need: a rectified flat image for mapping, a surface or ground model for computing a volume, a point cloud or a mesh to feed a 3D model. A design office that asks for "a drone survey" without specifying which of these deliverables it expects risks receiving a file that is useless for its real need — an orthophoto says nothing about elevation, and a good-looking mesh is not enough for a volume calculation.

Understanding what each deliverable represents, before writing a brief or requesting a quote, avoids the back-and-forth that costs a project weeks. Our guide on drone photogrammetry and 3D modelling covers the general principle of the technique; this one focuses on the files it produces and what can actually be done with them.

The orthophoto: a rectified, georeferenced photo map

The orthophoto (or orthophotomap) is a mosaic of the photos taken in flight, recomposed to correct perspective and relief distortion: every point in the image is brought back to its true position as seen from directly overhead, as on a map. The delivered file is a georeferenced image — typically a GeoTIFF, which opens in a GIS (QGIS, ArcGIS) just as in CAD software — on which a distance or a ground area can be measured to the stated accuracy, but not an elevation: the orthophoto is a flat, two-dimensional document.

It is the most universal and cheapest deliverable to produce: a municipality updating its local land-use plan, a farm operator documenting the state of a plot, a site archiving its progress month after month are all well served by it — our guide on the municipal orthophoto covers this use for local authorities. The specification to request in a brief is the ground resolution (GSD, ground sample distance, expressed in centimetres per pixel): a site that needs to spot a fine crack requires a far lower GSD than one that is enough for institutional communication.

DTM and DSM: modelling the ground and the surface

The digital surface model (DSM) and the digital terrain model (DTM) both represent an elevation at every point of an area, as a grid of values — but not the same elevation. The DSM captures everything the drone saw: the top of vegetation, roofs, vehicles, on top of the terrain itself. The DTM, by contrast, represents bare ground only: a classification algorithm strips out points belonging to buildings or vegetation to keep only those that actually touch the terrain.

This distinction is not a technicality: an earthworks volume calculation, a flood-zone assessment or a water-flow study that used a DSM instead of a DTM would wrongly fold the height of trees and buildings into the ground volume or elevation. Conversely, standard photogrammetry only sees what the camera photographs: under dense vegetation cover, it cannot reconstruct a reliable DTM for lack of visible ground points — one of the cases where an airborne or ground-based LiDAR remains necessary, as detailed in our guide on choosing between photogrammetry and laser scanning.

Point cloud and 3D mesh: usable geometry

The point cloud is the intermediate product of photogrammetric processing: every point carries an X, Y, Z position and, most often, a colour taken from the photo. Delivered as LAS or LAZ — the geomatics industry standard, compatible with most surveying, CAD and BIM software — or sometimes E57, it is the most usable deliverable for precise measurement: extracting a cross-section, positioning an object, checking a dimension. A brief that asks for "a point cloud" should specify that it means a dense cloud — produced after a thorough image-correlation computation — rather than the far sparser cloud that comes out of aerial triangulation alone.

The 3D mesh turns this cloud into a continuous surface of triangles, wrapped in the photo texture — usually as OBJ, FBX or glTF. It serves visualisation: a client presentation, integration into a BIM model for rendering, a virtual walkthrough of a site — our guide on the building digital twin covers this use. For a binding measurement, on the other hand — a contractual dimension, a compliance check — the raw point cloud remains the reference: the mesh smooths and simplifies the original geometry.

What actually determines the stated accuracy

The accuracy stated on a spec sheet is never down to camera quality alone. A study by Agüera-Vega, Carvajal-Ramírez and Martínez-Carricondo, published in 2017 in the American Society of Civil Engineers' Journal of Surveying Engineering, compared sixty photogrammetric projects combining different flight altitudes, terrain shapes and numbers of ground control points (GCPs): the orthophoto's planimetric accuracy turned out to be barely affected by flight altitude, while the surface model's vertical accuracy depended directly on the number of ground control points placed and surveyed before the flight (see the study on Google Scholar).

An alternative to ground control points is fitting the drone with an onboard GNSS RTK receiver, which records the exact position of every photo at the moment it is taken. A study by Štroner, Urban, Reindl, Seidl and Brouček, published in 2020 in the journal Sensors, measured the georeferencing accuracy obtained this way on a photogrammetric model: centimetre-level accuracy, comparable to what ground control points achieve, without the need to place and survey them (see the study on Google Scholar). Our guide on RTK/PPK and centimetre accuracy details when this requirement is genuinely justified for a given project, without systematically paying its extra cost.

What to specify in a brief or a quote request

Before requesting a quote, four specifications avoid nearly every misunderstanding about a photogrammetric deliverable. The expected deliverable first: an orthophoto for mapping or communication, a dense point cloud for measurement or to feed specialist software, a DTM for a volume or flow calculation, a mesh for a presentation or a BIM model — a single flight can produce several of these deliverables at once, but each carries its own processing cost. The file format next, matching the software that will use it: GeoTIFF for a GIS, LAS/LAZ for surveying software, OBJ or IFC for a BIM model. The expected coordinate system — Lambert-93/RGF93 for the vast majority of French projects, a local projection for some international sites. And finally the required accuracy, expressed in centimetres rather than as a vague approximation: it is what determines whether a standard flight is enough or whether ground control points, or even an RTK drone, are needed.

Request a quote describing the final use of your data rather than just the service — earthworks volume, cadastral update, BIM model, communication — and we will work out with you the deliverable, format and level of accuracy your project genuinely needs, without charging for a requirement it does not have.

Frequently asked questions

What is the difference between a DTM and a DSM?

The DSM (digital surface model) includes everything the drone saw — buildings, vegetation, vehicles — on top of the terrain. The DTM (digital terrain model) represents bare ground only, obtained by filtering out those elements. An earthworks volume calculation must use a DTM, never a DSM.

Does a photogrammetric point cloud replace laser scanning for a BIM model?

For a building's outer envelope or an open-air site, yes in most cases. For interiors, areas hidden by vegetation, or millimetre-level accuracy, a terrestrial laser scanner remains necessary.

Which file format should I request for a point cloud usable in my software?

LAS or LAZ work with nearly all surveying, CAD and BIM software; E57 for some specialised tools. Always specify the expected coordinate system (Lambert-93/RGF93 in France) in your request.

Request a free quote

Put it into practice

Also worth reading