The flight is the glamorous part of a LiDAR project. The deliverables are the part you actually live with. And if you have ever received a hard drive full of unfamiliar file extensions — or worse, paid for products your team never opened — you know the format list deserves more attention than it usually gets at quoting time.
Here is what each standard LiDAR deliverable actually is, who consumes it, and how to order only what your workflow needs. This is the menu behind our point cloud processing service, explained without the jargon.
The classified point cloud: LAS / LAZ
Everything starts here. The raw output of a drone LiDAR flight is a point cloud — millions to billions of 3D points, each one a laser measurement with coordinates, elevation, intensity, and return information.
The industry-standard container is LAS, an open format every serious geospatial package reads. LAZ is the same data losslessly compressed to a fraction of the size — prefer it for delivery and archival.
The word that matters most in a scope, though, is classified. Raw points don’t know what they are; classification assigns each point a standard code — ground, low/medium/high vegetation, building, water, noise — following the ASPRS scheme. Ground classification is the make-or-break step: every surface, contour, and volume downstream is built from the points labeled ground, so sloppy classification quietly poisons everything built on top of it. When you compare quotes, “classified LAS/LAZ” versus “raw point cloud” is a real difference, not a nuance.
Who uses it: engineers and surveyors who want to build their own surfaces, GIS analysts, anyone doing QA. Even if your daily tools are CAD-based, keep the classified cloud on file — it is the master dataset every other product derives from, and reprocessing it later is far cheaper than reflying.
Elevation models: DEM vs. DSM vs. DTM
Three acronyms, one family, endless confusion:
| Model | What it includes | What it’s for |
|---|---|---|
| DSM — digital surface model | Everything: canopy, rooftops, stockpiles | Viewsheds, line-of-sight, obstruction analysis, volumes against a surface |
| DTM / bare-earth DEM — digital terrain model | Ground only, vegetation and structures removed | Grading design, drainage, floodplain modeling, earthwork |
| DEM — digital elevation model | Umbrella term; in US practice usually means bare earth | Depends entirely on which of the above is meant |
The practical guidance: almost all design work wants bare earth, and because “DEM” is used loosely across the industry, say “bare-earth DEM” explicitly in any RFP or scope. The vegetation-removal step is precisely where LiDAR earns its cost premium, since it is the method that can actually see the ground under canopy.
Who uses it: civil engineers (grading, drainage), hydrologists (flood modeling), landscape architects, earthwork estimators.
GeoTIFF: the raster that carries its own map
Elevation models are usually delivered as GeoTIFF — a TIFF image whose pixels hold elevation values instead of colors, with the georeferencing (coordinate system, extent, resolution) embedded in the file itself. Drop it into QGIS, ArcGIS, or Civil 3D and it lands exactly where it belongs on earth, no manual alignment.
Resolution is specified as ground distance per pixel; drone LiDAR supports very fine rasters, but finer is not automatically better — match the resolution to the analysis, because file sizes grow with the square of it. Derived rasters like hillshades (for visualization) and slope maps (for erosion and constructability review) come from the same source and are cheap to add.
Who uses it: GIS teams, hydraulic modelers, anyone doing raster analysis or map production.
CAD contours: DWG / DXF
For most engineering consumers, contours are where LiDAR data finally looks like a survey. Processing generates contour lines — typically 1-foot intervals for design work, with index contours at 5-foot — delivered as DWG or DXF, layered and elevation-attributed so they drop straight into Civil 3D or AutoCAD.
Two things worth knowing. First, good contours are smoothed and cartographically cleaned from the surface model; raw unsmoothed contours from a dense LiDAR surface are jagged and nearly unusable, so contour generation is craft, not a button. Second, when contours must carry a professional land surveyor’s certification for platting or permitting, our licensed surveyor partners review and certify them — Remote Optix Inc. and Pro LiDAR are not professional land surveyors and do not represent themselves as such. The topographic packages on our topographic mapping page cover both the engineering-grade and partner-certified paths.
Who uses it: civil designers, site engineers, architects, permitting reviewers.
Shapefiles: SHP / DBF / SHX
The shapefile is GIS’s stubborn old workhorse for vector features — contours as GIS lines, site boundaries, planimetric features (edges of pavement, buildings, utilities), breaklines. One “shapefile” is actually a bundle of files traveling together: SHP holds the geometry, DBF holds the attribute table, SHX is the index linking them, and a PRJ carries the coordinate system. Deliver or receive them zipped, always — a shapefile missing its siblings is broken.
It is a dated format with real limits, and alternatives like GeoPackage are technically better; but SHP remains the interchange standard that everything reads, which is why it stays on the standard menu.
Who uses it: GIS departments, utility and municipal clients, environmental consultants.
The supporting cast: breaklines, planimetrics, and intensity
Beyond the core five formats, three add-ons appear on scopes often enough to know by name.
Breaklines are 3D lines digitized along abrupt terrain changes — curb lines, ditch flowlines, tops and toes of slopes. Added to a surface model, they force the triangulation to honor those edges instead of smearing across them, which is the difference between a swale that drains in the model and one that mysteriously ponds. Design-grade surfaces for civil work usually deserve them.
Planimetrics are the mapped 2D/3D features themselves: edge of pavement, building footprints, fence lines, visible utilities, signage. Extracted from the point cloud (often with the help of paired imagery), they arrive as CAD layers or shapefiles and save a design team days of manual digitizing.
Intensity rasters come free with the physics: every LiDAR return records how strongly it reflected, and mapping that intensity produces a grayscale image resembling a black-and-white aerial photo. Pavement markings, for instance, reflect differently than asphalt — useful for feature extraction even without a camera on board.
None of these are automatic; each adds processing hours and belongs on the scope only when someone downstream will actually consume it.
Ordering the right package
A useful default for a development site: classified LAZ + bare-earth DEM GeoTIFF + 1-ft CAD contours, with shapefiles added if a GIS team is involved. Corridor and campus projects layer on planimetrics and profiles.
Two final buying tips. Specify the coordinate system and vertical datum (for most US work, your State Plane zone and NAVD88) in the scope — it is a one-line request that prevents a world of rework. And remember that formats are only half the specification; the other half is accuracy, which has its own vocabulary of control points and checkpoints. Our guide to drone LiDAR accuracy covers how to specify that side, and our cost guide shows how deliverable choices move the price.