Glossary

Drone mapping, in plain English

39 terms you will meet in a DroneFlow project — what each one means, and why it matters to the result you get.

🔍

What DroneFlow produces File formats Coordinates and elevation Capture and flight Accuracy Platform

What DroneFlow produces

Orthomosaic

One flat, distortion-corrected aerial image of the whole site that you can measure on.

Hundreds of overlapping photos stitched into a single image, with the lens and perspective distortion taken out so every pixel sits where it really is on the ground. That correction is what separates it from a photo collage: distances and areas measured on an orthomosaic are real measurements, not approximations.

DSM

An elevation model of the site including everything on it — buildings, stockpiles, trees.

A Digital Surface Model records the height of the top of whatever the drone could see. Roofs, equipment, spoil piles and tree canopy are all part of the surface. It is the right model for measuring a stockpile, and the wrong one for a ground contour under a tree line.

DEM

An elevation model of the bare ground, with surface objects removed.

A Digital Elevation Model (also called a DTM, or 'bare earth') estimates the ground surface with buildings and vegetation stripped out. Use it for grading, drainage and contours. DroneFlow compares surveyed checkpoints against the DEM when a project has one, because a checkpoint is a shot on the ground.

Contours

Lines joining points of equal elevation — the topo drawing your civil team expects.

Contours are generated from the elevation model at a chosen interval: a 1 ft interval draws a line every foot of height change. Tighter intervals show more detail and more noise. They are delivered as DXF for CAD and as a shapefile set for GIS.

Point cloud

Millions of individually measured 3D points — the rawest form of the reconstruction.

Every point carries a position and a colour sampled from the photos. It is the densest and most flexible output, and the largest: a site can run to hundreds of millions of points. DroneFlow streams it in the browser rather than expecting you to download it, and the raw download is LAZ.

3D textured model

A continuous surface with the photography wrapped onto it — what most people picture as 'the 3D model'.

Where a point cloud is discrete measurements, a mesh is a solid skin built from them with the photo texture applied. It looks more finished and is easier to present, but it is an interpretation of the points, so fine detail and sharp edges can be smoothed away. Downloads as OBJ with its textures.

QA report

The automated checks DroneFlow runs on a finished project before you see it.

Each check returns PASS, WARN or FAIL and says what it looked at. A WARN is information, not a rejection — it usually points at something about the capture rather than the processing. The report is deliberately descriptive: it does not certify a survey and it does not label anything 'survey-grade'.

File formats

GeoTIFF

A TIFF image carrying its own map position — the standard raster for GIS.

The orthomosaic, DSM and DEM download as GeoTIFF. Because the georeferencing is embedded, QGIS or ArcGIS opens the file directly in the right place with no extra step.

LAS

The standard interchange format for point clouds.

An open ASPRS format that essentially every survey and GIS package reads — QGIS, CloudCompare, Civil 3D and ReCap among them. LAZ is the compressed version of the same thing.

LAZ

A losslessly compressed LAS file — same data, a fraction of the size.

DroneFlow delivers the raw point cloud download as LAZ (or a zip of LAZ for a tiled cloud). Nothing is thrown away in the compression, and the tools that read LAS read LAZ. Very large raw datasets are capped for download because the browser viewer, not a local copy, is the intended way to work with a cloud of that size.

OBJ

The mesh format the 3D textured model downloads in.

An OBJ arrives with its texture images alongside it, so keep the whole folder together — an OBJ on its own renders untextured grey. Opens in Blender and most 3D packages.

3D Tiles

The streaming format that lets a multi-gigabyte model open in a browser tab.

Rather than sending the whole model, the site is cut into a pyramid of tiles and the viewer loads only the detail your current view needs. That is why a model far too large to download opens in seconds, and why detail sharpens for a moment after you stop moving.

KMZ

The Google Earth format — a zipped KML with its imagery inside.

DroneFlow produces a KMZ of the map so a stakeholder with no GIS software can double-click it and see the site in Google Earth. KML is the same thing uncompressed, and is one of the formats you can export map annotations to.

Shapefile

The ESRI vector format — actually a set of files that must travel together.

Contours ship as a shapefile set, and map annotations export as a zipped shapefile for ArcGIS or Civil 3D. The .prj file in the set is what tells the receiving software which coordinate system the geometry is in, so a .shp sent on its own often lands in the wrong place. One note when importing: the shapefile's text column is capped at 254 characters by the format itself, so long notes are better exported as GeoJSON or CSV.

DXF

The CAD interchange format — how contours reach AutoCAD and Civil 3D.

Contours are delivered as DXF, which AutoCAD, Civil 3D and any CAD package that reads DXF will open. A 3D DXF is also one of the two ways to upload a design surface for cut/fill; when you do, the elevation unit matters — declaring feet as metres puts every number out by a factor of about 3.28.

GeoJSON

A plain-text vector format, and the most forgiving way to move map markup around.

Annotations export to GeoJSON and import from it. Because it is text with no sidecar files and always in latitude/longitude, it avoids the coordinate-system confusion that trips up shapefile and DXF transfers.

Coordinates and elevation

CRS

The agreement about what the numbers in your file mean — where zero is, and what the units are.

A coordinate reference system (also called an SRS) pins numbers to real places. Your mapping software normally writes it into the export, so DroneFlow reads it and you leave the field blank at upload. You only type one in if processing reports it cannot determine one, or you know the export is missing it.

EPSG code

A short number that names a coordinate system exactly — EPSG:32617, for example.

EPSG codes remove the ambiguity from phrases like 'UTM zone 17'. If DroneFlow asks you for a source coordinate system, this is what it wants: the code your mapping software reports for the export. Leave it blank unless processing says it could not detect one.

UTM

A worldwide projected grid, split into numbered zones, measured in metres.

A common export coordinate system for drone processing software. It is projected — coordinates are metres on a grid rather than degrees — which is why an export in UTM cannot be treated as latitude and longitude without conversion.

State Plane

The US state-by-state projected grid, often in US survey feet.

What most American civil and survey deliverables are drawn in. The unit is the thing to watch: US survey feet and international feet are very slightly different, and both differ from metres — a mismatch shows up as a consistent scale error rather than an obvious failure.

WGS84

The global latitude/longitude system GPS reports in.

DroneFlow writes exported annotations in WGS84 longitude/latitude. Site plan overlays do not have to be — a Shapefile or GeoPackage carries its own coordinate system and is reprojected for you, and a DXF carries none, so you tell us the zone off the title block. Heights measured against the WGS84 ellipsoid are ellipsoidal heights, which are not the elevations civil work expects.

Vertical datum

What 'elevation zero' means — and the single most common cause of elevations that look wrong.

You choose the vertical datum when you upload. Get it wrong and the geometry is still perfect but every height is out by roughly the same amount, which is the signature to look for. DroneFlow only offers datums whose coverage actually reaches your site.

Ellipsoidal height

Height above a mathematical model of the earth — what raw GPS and RTK report.

It is a clean measurement and almost never the number a civil drawing wants. In Florida an ellipsoidal height reads roughly 90 ft below the orthometric height for the same point; the offset differs everywhere else.

Orthometric height

Height related to mean sea level — the elevation site and civil work actually use.

Converting from ellipsoidal to orthometric height needs a geoid model for your part of the world. DroneFlow applies one at upload when you pick a datum, which is why choosing the datum matters more than any other field on the form.

Capture and flight

Photogrammetry

Deriving 3D measurements from overlapping 2D photographs.

If a point on the ground appears in enough photos taken from enough different positions, its position in space can be solved for. Everything DroneFlow produces comes out of that one idea, which is also why overlap and image sharpness matter so much.

GSD

How much ground one pixel covers — the resolution of your map.

A 2 cm GSD means each pixel is about 2 cm across on the ground. It is set mostly by flying height and camera: lower flights give finer GSD and take more photos to cover the same area. Detail smaller than roughly two pixels will not be reliably measurable.

Overlap

How much each photo repeats its neighbours — the main lever on reconstruction quality.

Front overlap is along the flight line, sidelap is between adjacent lines. Too little and the reconstruction has holes or fails outright; more overlap costs flight time and photos but is the cheapest insurance available. Vegetation, water and uniform surfaces need more of it than hard urban detail.

Nadir

Camera pointed straight down — the standard geometry for a 2D map.

Nadir photography is what an orthomosaic is built from. It sees roofs and ground well and vertical surfaces badly, which is why a nadir-only capture produces a good map and a 3D model with smeared building sides.

Oblique

Camera tilted off vertical — what gives a 3D model real walls and façades.

Oblique passes see the sides of things a nadir pass cannot. Adding them to a nadir capture is the difference between a model that looks melted at the edges of buildings and one that stands up to inspection. They add flight time and photos.

Accuracy

GCP

A marked point on the ground whose position you surveyed, used to anchor the reconstruction.

Control points are fed into processing so the model is built onto known coordinates. They improve the result — and precisely because they were used to build it, they cannot also measure how accurate it turned out. That is what a checkpoint is for.

Checkpoint

A surveyed point deliberately held back from processing, used to measure the result.

Because the reconstruction never saw it, the difference between where the model puts a checkpoint and where you surveyed it is an honest measure of error. Upload checkpoints as a CSV with an ID, latitude, longitude and elevation, in latitude/longitude — a State Plane or UTM file is refused rather than quietly measured in the wrong place.

RTK

Corrected satellite positioning that fixes the drone's position to centimetres in flight.

An RTK drone records far better photo positions than a standard GPS one, which usually means fewer or no ground control points. The correction source matters: if the base station position was wrong, every photo inherits that error and no amount of later processing recovers it.

PPK

The same correction as RTK, applied after the flight instead of during it.

PPK logs raw satellite data and corrects it later against a base station record. It removes the dependence on a live radio or network link during flight, at the cost of a processing step afterwards.

RMSE

A single number summarising how far the model sits from your surveyed checkpoints.

DroneFlow reports RMSE in metres and feet alongside the mean error, the scatter, the worst point and every individual point. It deliberately does not tell you whether that number is good enough: the standard that matters is the one on your contract or spec, and only you can apply it. One reading worth knowing — if every point is off by roughly the same amount, that is a vertical datum mismatch, not bad data.

Platform

Cut and fill

How much material has to come off, and how much has to go on, to reach a design grade.

DroneFlow compares your surveyed surface against a design surface you upload — a GeoTIFF or a 3D DXF — and reports the volumes separately. Two things decide whether the answer is meaningful: that both surfaces are on the same vertical datum, and that the DXF's elevation unit was declared correctly.

Base surface

The 'floor' a volume is measured down to — and the choice that decides the number.

A stockpile volume is only defined relative to something. Measuring to a flat plane, to the surrounding ground, or to an earlier flight's surface gives three different and individually defensible answers. Say which you used when you report a volume.

Photo credit

The unit DroneFlow bills processing in — one credit, one photo.

A capture's cost is its photo count, so an estimate is available before you commit. Reprocessing an existing project and using the training sample do not consume credits.

Annotation

A pin, line, area or note you draw on the map or the 3D model.

Annotations belong to the project, so a pin dropped on the 3D model is the same object as one on the 2D map and travels with a share link. They import and export as GeoJSON, KML, zipped shapefile or CSV.

No term matches that search. Try a different word, or browse the Help Center.