One drone flight produces one dataset — but that dataset gets processed into several different deliverables, each built for a different job. Knowing what each one actually is makes it a lot easier to ask for the right one.

The point cloud — the master dataset

Everything else on this list is built from the point cloud, so it's worth understanding first. It's a dense collection of individual points in three-dimensional space — millions of them on a typical site — each one describing a precise X, Y, and Z location, usually with color information attached. Photogrammetric point clouds, built from overlapping photos, typically run somewhere in the range of 50 to 500 points per square meter; LiDAR point clouds are often denser still, and unlike photogrammetry, LiDAR pulses can penetrate vegetation to capture ground points hidden under canopy. Point clouds are delivered as LAS or LAZ files and import directly into CAD, GIS, and BIM software for measurement and modeling — this is the deliverable serious design and engineering work is actually built on.

The orthomosaic — the map you can measure

An orthomosaic is a single, seamless aerial image built by stitching together hundreds or thousands of overlapping drone photos and correcting each one geometrically so every pixel represents a true, consistent ground distance. That correction is what separates it from a regular aerial photo — you can measure a distance or an area directly off an orthomosaic and trust the number. Delivered as a GeoTIFF, it's the most accessible deliverable on this list: easy to view, easy to share with someone who doesn't have survey software, and still precise enough to serve as a base layer for GIS mapping, progress documentation, or a design overlay comparison.

DSM vs. DTM — surface versus bare earth

These two get confused constantly, and the difference matters. A Digital Surface Model (DSM) captures the elevation of everything visible from above — the ground, but also buildings, stockpiles, equipment, and vegetation. A Digital Terrain Model (DTM) strips all of that away and represents only the bare-earth ground surface underneath. A DSM is the right tool when you need to know what's physically present — visualizing how a structure interacts with its surroundings, for instance. A DTM is what grading, drainage, and cut/fill design actually need, because those calculations depend on the true ground surface, not what's sitting on top of it. Both are typically delivered as GeoTIFF rasters or gridded elevation files.

How accurate is each deliverable

Accuracy isn't a single number that applies to every deliverable equally — it depends on how the underlying point cloud was built. Standard photogrammetry without ground control typically holds absolute accuracy to somewhere between one and three times the ground sample distance, often in the 3–8 cm range depending on flight altitude. Add ground control points or RTK/PPK positioning, and that tightens into the 1–5 cm range for most of these deliverables. LiDAR-based point clouds often land in a similar or better range and hold up better in vegetation, where photogrammetry alone can't see the ground at all. Whatever deliverable you're working from, it's worth asking whether ground control was used — the answer changes how much weight the numbers can carry.

The 3D mesh — what it looks like

A 3D mesh takes the point cloud and builds a continuous, textured surface over it — the photorealistic model people can rotate, zoom into, and walk through in a viewer. It's built from the same underlying data as everything else, but it's optimized for visualization rather than precise measurement: a mesh is what you hand a client who wants to see the finished site, not what an engineer uses to calculate a volume. Delivered as OBJ or GLB, it's the deliverable most useful for stakeholder presentations, marketing, and general site walkthroughs.

Getting it into your CAD or GIS workflow

Contour lines, generated from the DTM or DSM at whatever interval a project needs, export as DXF or DWG for direct use in AutoCAD or Civil 3D. Point clouds and surfaces import into Civil 3D, Trimble Business Center, or Carlson for surface modeling, breakline extraction, and design comparison — though it's worth knowing that surface exports between platforms don't always translate perfectly. A LandXML export alongside the native format is common practice specifically because it carries breaklines and boundary data across software more reliably than a straight surface file does. For most projects, the practical move is agreeing on export format before the flight, not after the data's already processed.

How often you'll actually need each one

Not every project needs every deliverable on every flight. Progress monitoring and stockpile tracking typically run on orthomosaics and DSM/DTM pairs, generated on a routine cycle. Point clouds get pulled and archived on that same cycle even when they're not immediately needed, simply because they're the source data everything else is built from, and reprocessing later is more work than exporting once up front. Contour lines and CAD-ready surfaces tend to get generated on demand, tied to a specific design or grading milestone rather than every single flight. Meshes are usually the exception — generated for a specific presentation, marketing need, or project milestone rather than as a routine output.

Matching the deliverable to the question you're actually asking

If you need to...	Start with...
Measure a stockpile	DSM and DTM together, or the point cloud directly
Build a base map for GIS or documentation	The orthomosaic
Design a grading or drainage surface	The DTM
Show a client the finished site	The 3D mesh
Build your own CAD workflow around raw data	The point cloud