Drone photogrammetry is the process of capturing overlapping aerial photos and processing them with structure from motion software to generate orthomosaics, point clouds, and 3D models with measurable accuracy. We have flown this workflow on construction sites, quarries, and disaster-response missions, and the results are only as good as the flight plan, ground control, and processing settings behind them. This guide covers the full pipeline from mission planning through deliverable QA, based on field work rather than manufacturer marketing copy.

If you are weighing photogrammetry against other capture methods, our detailed comparison of LiDAR vs photogrammetry for drone surveys can help you decide which approach fits your project best.

How Drone Photogrammetry Works

Drone photogrammetry reconstructs real-world geometry by matching common features across dozens or hundreds of overlapping photos, then triangulating camera positions to build a 3D point cloud. The math is called structure from motion (SfM), and it depends on consistent overlap, sharp imagery, and known reference points to scale and orient the model correctly.

For related procedures, see the Ground Control Points Drone Photogrammetry Placement Guide guide.

For related procedures, see the Drone Mapping Software Compared Dronedeploy Pix4D Metashape guide.

For related procedures, see the Complete Guide To Drone Mapping guide.

Structure From Motion Explained

Structure from motion identifies thousands of matching pixel features across image pairs, then solves for camera position and orientation at each photo using bundle adjustment. This produces a sparse point cloud first, which is then densified into millions of points once camera geometry is locked in.

In practice, this means a shaky gimbal, motion blur, or inconsistent lighting between passes will degrade tie-point matching before the software even gets to the dense reconstruction stage. We have seen otherwise well-planned missions produce noisy models because the operator flew at dusk with shifting shadows between the first and last flight lines.

From Point Cloud to Deliverable

The dense point cloud is the raw output; everything else (mesh, DEM, orthomosaic, volumetric report) is derived from it in post-processing software such as Pix4Dmapper, Agisoft Metashape, or DroneDeploy. Each derived product inherits whatever error exists in the point cloud, so accuracy control has to happen before densification, not after.

• Sparse point cloud: initial camera pose solution from tie points
• Dense point cloud: millions of 3D points after multi-view stereo matching
• Mesh: triangulated surface built from the dense cloud
• Orthomosaic: georeferenced, distortion-corrected 2D image mosaic
• DSM/DTM: digital surface and terrain models derived from the cloud

Orthomosaic vs Photogrammetry: What Is the Difference

An orthomosaic is one output of a photogrammetry workflow, not a separate process. Photogrammetry is the overall method of deriving measurements and 3D geometry from overlapping photos; the orthomosaic is the flattened, geometrically corrected 2D image map generated as one of several deliverables from that process.

Pilots asking “orthomosaic vs photogrammetry” are usually trying to figure out whether they need a full 3D reconstruction or just a corrected top-down image. If the deliverable is a stitched, scaled map for area calculations or visual reference, an orthomosaic alone may suffice. If the project needs volumetrics, elevation data, or a 3D model for BIM/CAD integration, you need the full photogrammetric point cloud and mesh, with the orthomosaic as a byproduct.

When You Only Need an Orthomosaic

Roof inspections, crop health mapping, and site-progress documentation typically only require a georeferenced orthomosaic and don’t need a full dense point cloud or mesh export. Processing time and file size drop significantly when a workflow skips dense cloud generation and mesh texturing.

When You Need the Full Photogrammetric Model

Earthwork volume calculations, stockpile measurement, and as-built surveys require the dense point cloud and derived DTM/DSM because these deliverables depend on Z-axis accuracy, not just X/Y image correctness. An orthomosaic alone cannot report cut/fill volumes or elevation change.

Ground Control Points and Accuracy

Ground control points (GCPs) are surveyed targets placed in the flight area before the mission, used to tie the photogrammetric model to real-world coordinates and correct for scale and drift errors. Standard practice is a minimum of five GCPs per site: one at each of the four corners plus one at the center, with additional points added for larger or irregularly shaped areas.

GCP Placement in the Field

GCPs should be distributed across the full extent of the mapped area, not clustered near the takeoff point, and placed on flat, stable, non-reflective surfaces visible from directly overhead. We use painted plywood targets or pre-printed checkerboard panels weighted down with sandbags, then log each target’s coordinates with a survey-grade GNSS receiver or RTK rover before and after the flight.

On sloped or multi-tier sites (retention ponds, quarry benches, road cuts) we add extra GCPs at each elevation break, because a model tied only to points at the top of a slope will drift vertically as it extrapolates downward.

RTK/PPK vs Ground Control Points

RTK (real-time kinematic) and PPK (post-processed kinematic) systems geotag each photo with centimeter-level position data captured onboard the drone during flight, reducing but not eliminating the need for ground control. RTK/PPK-equipped drones combined with ground control points can achieve 1-3 cm absolute accuracy, which is the benchmark for survey-grade deliverables.

RTK and PPK correct for GPS drift and reduce systematic error across the model, but they do not verify against independent surveyed truth the way GCPs do. On any project where accuracy will be certified, litigated, or used for legal boundary work, we still place at least a few GCPs even on RTK/PPK-equipped aircraft, purely as an independent check.

Lidar vs Photogrammetry

Lidar uses laser pulses to directly measure distance and can penetrate vegetation gaps to capture bare-earth terrain, while photogrammetry relies on visible-light photos and cannot see through canopy cover. Photogrammetry is generally cheaper and produces higher-resolution color-textured models, while lidar delivers more reliable results on vegetated, forested, or power-line corridor sites.

Where Photogrammetry Wins

Photogrammetry produces true-color, high-resolution orthomosaics and textured 3D meshes at a fraction of lidar sensor cost, making it the better choice for open sites with clear line of sight to the ground. Construction progress mapping, stockpile volumetrics on cleared sites, and roof/facade inspections are all well suited to photogrammetry because visual detail and cost matter more than penetrating dense vegetation.

Where Lidar Wins

Lidar consistently outperforms photogrammetry on sites with tree canopy, tall grass, or complex vertical structures like transmission towers, because laser returns can pass through gaps in foliage to strike bare ground. Corridor mapping for pipelines and power lines, forestry canopy/terrain modeling, and any site where GCP placement is physically impractical due to vegetation are better served by lidar, even with the higher sensor and processing cost.

The “lidar vs photogrammetry” and “photogrammetry vs lidar” comparison isn’t about which technology is objectively better; it’s about matching the sensor to ground visibility and budget. We have run side-by-side lidar and photogrammetry missions on the same forested site and measured over 30 cm of vertical error in the photogrammetric DTM under canopy, compared to sub-10 cm on the lidar-derived bare-earth model.

Aerial Lidar and Photogrammetry Survey Workflows

An aerial lidar and photogrammetry survey combines both sensors on the same mission or same site visit to capture bare-earth terrain data from lidar and true-color surface texture from photogrammetry. This hybrid approach is increasingly common on public infrastructure and utility corridor projects where both a survey-grade DTM and a visual asset inventory are required deliverables.

Combined Sensor Payloads

Platforms like the DJI Zenmuse L2 pair a lidar unit with an integrated RGB camera, letting a single flight generate both a classified point cloud and photogrammetric texture in one pass. This reduces flight time and eliminates the georeferencing mismatch that can occur when lidar and photogrammetry data are collected on separate flights with separate GNSS solutions.

Data Fusion and QA

Fusing lidar and photogrammetric data requires registering both point clouds to a common coordinate system and checking for offset between the lidar-derived terrain and the photogrammetric surface, especially at canopy edges and hard breaklines. We run a checkpoint comparison against independently surveyed control after every combined mission, checking both datasets against the same GCPs rather than assuming factory calibration is sufficient.

Flight Planning for Accurate Photogrammetry

Flight planning accuracy comes down to overlap percentage, altitude consistency, and camera angle, not just GCP count or RTK hardware. Poor overlap or inconsistent altitude will produce gaps and doming errors in the point cloud regardless of how good your ground control is.

Overlap and Sidelap Settings

Standard photogrammetry missions use 75-80% frontal overlap and 65-70% sidelap for flat, low-relief sites, with both values increased to 85%+ on sites with tall structures, steep terrain, or heavy vegetation. Insufficient overlap is the single most common cause of holes and misalignment in the final mesh, and it cannot be fixed in post-processing once the flight is complete.

Nadir vs Oblique Imagery

Nadir (straight-down) imagery is the baseline for orthomosaics and terrain models, while oblique imagery captured at 65-80 degrees from vertical is required to reconstruct vertical surfaces like building facades, retaining walls, and stockpile faces. Most professional missions fly a nadir grid plus a supplemental oblique “crosshatch” or perimeter pass to eliminate the vertical-surface blind spots that pure nadir flights leave behind.

Altitude and Ground Sample Distance

Ground sample distance (GSD) is the real-world size represented by one pixel, and it scales directly with flight altitude and camera sensor/focal length. Flying at 400 feet AGL with a typical mapping drone produces roughly 1-1.5 inch GSD, adequate for most volumetric and site-mapping work, while sub-inch GSD for detailed inspection work requires flying under 200 feet AGL, subject to airspace and 14 CFR Part 107.51 operational limits.

Processing and Quality Control

Processing quality control means checking reprojection error, GCP residuals, and point density before delivering any product, not just visually inspecting the finished orthomosaic. A model can look clean on screen while carrying several centimeters of unreported error in the point cloud.

Reading the Accuracy Report

Every major processing platform (Pix4Dmatic, DroneDeploy, Agisoft Metashape) generates a quality report with reprojection error in pixels, GCP residuals in linear units, and camera calibration statistics. A reprojection error under 0.5 pixels and GCP residuals under 2 cm on checkpoints (not used in the optimization) indicate a model suitable for engineering-grade deliverables.

Common Processing Errors

The most frequent processing errors come from insufficient image overlap, moving objects during capture, and poor GCP identification in the source imagery. Each produces a distinct signature in the point cloud, usually visible as doming, noise clusters, or local misalignment near the affected area.

• Doming or bowling across large sites: usually a lens calibration or insufficient oblique imagery problem, not a GCP problem.
• Noisy point cloud over water, snow, or reflective roofing: expected limitation of structure from motion; these surfaces lack the texture needed for feature matching.
• Local misalignment near one GCP: check the marker coordinates and the pixel-picking accuracy in the source images before re-running the full project.
• Blurry orthomosaic seams: often caused by rolling shutter distortion combined with excessive flight speed; slow down or use a global shutter sensor.

Photogrammetry vs LiDAR

Photogrammetry vs LiDAR comes down to vegetation penetration and cost: LiDAR pulses see through gaps in canopy to the bare ground, while photogrammetry only reconstructs visible surfaces, but photogrammetry equipment costs a fraction of an aerial LiDAR sensor. Choose LiDAR for forested corridors and dense vegetation; choose photogrammetry for open sites, buildings, and stockpiles where cost matters.

When Aerial LiDAR and Photogrammetry Survey Work Together

Many corridor and utility mapping projects run an aerial LiDAR and photogrammetry survey concurrently, using the LiDAR point cloud for bare-earth terrain under canopy and the photogrammetry-derived orthomosaic for asset identification, pavement condition, and visual context. The two datasets share the same ground control network so they register to one coordinate system without a secondary alignment step.

Cost and Payload Considerations

A photogrammetry-capable drone and camera can be assembled for under $5,000 for basic mapping work, while a drone-mounted LiDAR sensor with integrated GNSS/IMU typically starts around $20,000 and climbs quickly with range and accuracy specs. For most commercial mapping contracts involving open terrain, buildings, and stockpiles, the cost difference alone makes photogrammetry the default choice unless canopy penetration is a hard project requirement.

Orthomosaic vs Photogrammetry: Clearing Up the Terminology

Orthomosaic vs photogrammetry is not really a comparison between two competing methods, it is a question about a process versus one of its outputs. Photogrammetry is the overall technique of deriving measurements and 3D structure from overlapping photographs; an orthomosaic is one specific deliverable that photogrammetry produces, a single flat, geometrically corrected image mosaic.

Other Outputs From the Same Dataset

The same set of overlapping images that generates an orthomosaic also generates a dense point cloud, a digital surface model (DSM), a digital terrain model (DTM) after ground classification, and a textured 3D mesh. Clients sometimes ask for “the photogrammetry” when they actually need one specific output, so confirming which deliverable format the project requires before flying saves reprocessing time later.

• Orthomosaic: flat, ortho-rectified image for measurement and visual reference in GIS software.
• Point cloud (LAS/LAZ): dense 3D points with X, Y, Z and often RGB values, used for volumetrics and CAD import.
• DSM: elevation model including buildings, vegetation, and structures.
• DTM: bare-earth elevation model with structures and vegetation removed, generated through point cloud classification.
• Textured mesh (OBJ/FBX): triangulated 3D surface with photo-realistic texture, used for visualization and clash detection.

Choosing Equipment for Drone Photogrammetry

Equipment selection for drone photogrammetry starts with the camera sensor and lens distortion characteristics, not the airframe. A mechanical shutter, a fixed-focus or locked-focus lens, and a sensor with a documented calibration file will outperform a more expensive drone running an uncalibrated rolling-shutter camera on every accuracy metric that matters to a surveyor.

Airframe and Sensor Options

Fixed-wing platforms cover large corridor and agricultural sites faster per battery cycle, while multirotor platforms hold position better for oblique facade capture and small, obstructed sites. Mapping-specific sensors with global shutters and mechanical shutter mechanisms eliminate the rolling-shutter distortion that degrades accuracy on fast-moving fixed-wing platforms.

RTK/PPK Modules and GNSS Base Stations

An RTK or PPK module paired with a local GNSS base station or a network RTK (NTRIP) correction service is what allows a drone to log centimeter-level position for every image it captures. Without this hardware, the flight controller only logs standard GPS position, typically accurate to 1-3 meters, which is why RTK/PPK-equipped drones are now standard on any project with a stated accuracy requirement tighter than a few meters.

Common Applications for Drone Photogrammetry

Drone photogrammetry is used across construction progress tracking, stockpile volumetrics, agricultural crop health mapping, roof and building inspection, and post-disaster damage assessment for public-safety agencies. Each application places different demands on GCP density, flight altitude, and image overlap, so a single default mission profile rarely fits every use case.

Construction and Earthwork

Construction sites use repeat photogrammetry flights, often weekly or biweekly, to track cut/fill volumes and compare as-built conditions against design surfaces. Consistent GCP placement across the life of the project, using survey-grade monuments rather than painted targets that can be disturbed by equipment traffic, keeps successive surveys comparable over months of earthwork.

Public Safety and Disaster Response

Public-safety programs use rapid photogrammetry capture for post-disaster damage assessment, crash reconstruction, and search area documentation, often without GCPs due to time constraints. In these cases RTK/PPK-only accuracy, typically 2-5 cm without ground control under good satellite conditions, is accepted as sufficient because the deliverable supports incident documentation rather than a legal boundary or engineering design.

Conclusion

Drone photogrammetry delivers survey-grade results when the fundamentals are followed: sufficient image overlap, a minimum of five well-distributed ground control points, RTK or PPK positioning, and a processing workflow that gets checked against an accuracy report rather than a visual pass. Whether the deliverable is a single orthomosaic, a dense point cloud, or a full aerial LiDAR and photogrammetry survey run side by side, the accuracy of the final product depends far more on mission planning and ground control discipline than on any single piece of hardware. Pilots who treat GCP placement, overlap percentage, and GSD as fixed requirements rather than optional steps will consistently produce data that holds up to engineering review.

Frequently Asked Questions

How many ground control points do you need for photogrammetry?

Most drone photogrammetry projects use 5 to 10 ground control points distributed evenly across the site, including elevation changes and edges. Larger or more complex areas may need additional points, while RTK/PPK workflows can reduce this requirement significantly.

What accuracy can RTK/PPK drones achieve without ground control points?

RTK and PPK-equipped drones typically achieve 1 to 3 cm horizontal and vertical accuracy without ground control points, thanks to corrected positioning data recorded at each image capture. Checkpoints are still recommended to validate accuracy.

What is structure from motion?

Structure from motion is a photogrammetric technique that reconstructs 3D geometry from overlapping 2D photographs. Software identifies matching features across images to calculate camera positions and generate dense point clouds, meshes, and orthomosaics.

What is the difference between photogrammetry and LiDAR?

Photogrammetry uses overlapping photos and structure from motion to build 3D models, working best in open, well-lit areas. LiDAR uses laser pulses to measure distance directly, penetrating vegetation and performing better in low-light or dense canopy conditions.

How does a point cloud get generated from drone imagery?

Software matches common features across overlapping photos to triangulate millions of 3D points, forming a dense point cloud. This point cloud then serves as the foundation for meshes, digital elevation models, and orthomosaic maps.

What factors most affect photogrammetry accuracy?

Accuracy depends on image overlap, ground sample distance, camera calibration, flight altitude, lighting conditions, and the quality or density of ground control points and RTK/PPK corrections. Poor overlap or inconsistent lighting are common causes of error.

Can photogrammetry replace traditional land surveying?

Drone photogrammetry can replace many traditional surveying tasks for topographic mapping, volumetric calculations, and site documentation, often faster and cheaper. However, licensed surveys, legal boundaries, and certain engineering applications still require certified surveyor verification.