1. The Two Ways a Drone Measures the World
A surveyor walks a construction site for three days with a total station, capturing 400 ground points. A drone flies the same site in 25 minutes and captures 15 million points. The difference is not just speed — it is data density. The surveyor's 400 points tell you where the ground is at 400 locations. The drone's 15 million points tell you what the ground looks like everywhere, including the subtle 15-centimeter depression that a 20-meter grid interval completely missed.
But not all drone surveying methods produce the same data. LiDAR and photogrammetry measure the world in fundamentally different ways, and choosing the wrong method for your terrain wastes flight time on data you cannot use — or worse, produces survey-grade outputs that fail accuracy validation because the method was not matched to the surface conditions.
2. Photogrammetry: Making 3D Models From 2D Photos
Photogrammetry works by overlapping aerial photos — typically 70-80% forward overlap and 60-70% side overlap — and using software to identify the same feature in multiple images. By knowing the camera's position and angle for each shot from onboard RTK GPS and IMU, the software triangulates the 3D position of every pixel that appears in at least three overlapping images.
2.1 The Hardware You Need
- Camera: A mechanical shutter is essential. Rolling shutters create geometric distortion during forward motion. The DJI Zenmuse P1 uses a full-frame 45MP sensor with a global mechanical shutter, capturing distortion-free nadir images at 0.7-second intervals.
- Positioning: RTK GPS with ground control points (GCPs) or an RTK base station. Without RTK correction, the camera's position is accurate to plus or minus 2-3 meters. With RTK, it is plus or minus 2 centimeters horizontally and 3 centimeters vertically.
- Platform: A DJI Matrice 350 RTK provides RTK positioning, 2.7 kg payload capacity for the P1 camera (approximately 800 grams), and 47 minutes of flight time per battery pair — covering roughly 1.5-2 square kilometers per flight at 5 cm/pixel GSD.
2.2 Where Photogrammetry Excels
- Bare earth, construction sites, open-pit mines — clean surfaces with visible texture produce the highest reconstruction accuracy at roughly 60-70% less cost per square kilometer than LiDAR for bare-earth terrain mapping.
- Orthomosaic generation — photogrammetry natively produces orthorectified imagery where every pixel is corrected for terrain distortion, creating a true-to-scale map that can be overlaid on CAD drawings or GIS layers directly.
2.3 Where Photogrammetry Fails
- Vegetation canopy — photos capture the top of the trees, not the ground underneath. In forested terrain, photogrammetry produces a digital surface model (DSM) of the canopy, not a digital terrain model (DTM) of the actual ground.
- Low-texture surfaces — sand dunes, snow fields, calm water, and freshly paved asphalt lack the visual features that software needs to match between images. The output has data holes where the software could not find enough tie points.
- Low-light conditions — photogrammetry requires ambient light. Pre-dawn, dusk, and nighttime operations produce underexposed imagery that degrades reconstruction accuracy.
3. LiDAR: Measuring Distance With Laser Pulses
LiDAR fires laser pulses at the ground — typically 240,000 to 480,000 pulses per second — and measures the time each pulse takes to reflect back to the sensor. Since the speed of light is constant and the aircraft's position is known from RTK GPS and IMU, each returned pulse yields a precise 3D coordinate.
3.1 The Hardware
- LiDAR sensor: The DJI Zenmuse L2 integrates a Livox Avia LiDAR with a 26MP RGB mapping camera on a 3-axis stabilized gimbal. The LiDAR emits 240,000 points per second with up to 5 returns per pulse — a single laser pulse can reflect off tree canopy, mid-story branches, and finally the ground, producing three separate elevation measurements from one beam.
- IMU: The L2's built-in high-precision IMU tracks the sensor's orientation at 200 Hz, correcting for aircraft vibration, wind gusts, and gimbal movement. Without this correction, a 1-degree pitch error at 100 meters altitude creates a 1.75-meter horizontal position error in the point cloud.
- Platform: Same DJI Matrice 350 RTK as photogrammetry. The L2 weighs approximately 930 grams and draws approximately 25W through the SkyPort V2 interface.
3.2 Where LiDAR Excels
- Vegetation penetration — this is the decisive LiDAR advantage. Multiple returns per pulse mean the laser passes through gaps in the canopy, reflects off branches, and reaches the ground. The resulting point cloud can be classified into ground points and vegetation points, producing a bare-earth DTM even under dense forest canopy — something photogrammetry cannot do.
- Corridor mapping — power line corridors, pipeline routes, and highway alignments are narrow, linear features where photogrammetry's wide-area coverage is wasted. LiDAR captures infrastructure details in a single pass along a 50-meter-wide corridor.
- Low-light and night operations — LiDAR provides its own illumination. It works identically in full daylight and complete darkness, enabling 24-hour operations.
3.3 Where LiDAR Falls Short
- Cost — the Zenmuse L2 costs approximately 2-3 times more than the P1. For open-terrain projects where photogrammetry produces equivalent accuracy, LiDAR's cost premium is difficult to justify.
- No native color imagery — LiDAR produces geometry without texture. The L2's integrated 26MP camera provides colorization of the point cloud, but the RGB data is lower resolution than a dedicated photogrammetry camera.
- Data volume — a 30-minute LiDAR flight generates 15-25 GB of raw point cloud data, requiring a workstation with 32GB+ RAM and a dedicated GPU for processing.
4. Direct Comparison: LiDAR vs Photogrammetry
Ground Sampling Distance: Photogrammetry — 1-5 cm/pixel (P1 at 100m). LiDAR — 2-5 cm point spacing at 100m altitude. Photogrammetry wins on raw resolution for bare-earth surfaces.
Vertical accuracy (RMSE): Photogrammetry — 3-10 cm with RTK and GCPs on bare earth. LiDAR — 2-5 cm with RTK, independent of surface type. LiDAR wins on consistency across all terrains.
Vegetation penetration: Photogrammetry — none. LiDAR — 3-5 returns per pulse, bare-earth DTM under dense canopy. LiDAR is the only viable option for forested terrain.
Operational lighting: Photogrammetry — daylight required. LiDAR — works in complete darkness.
Output data types: Photogrammetry — orthomosaic, DSM, 3D mesh, textured model. LiDAR — classified point cloud, DTM, DSM, contour lines, volumetric calculations.
Data processing time: Photogrammetry — 2-4 hours per flight. LiDAR — 1-3 hours with automated classification.
Cost per square kilometer: Photogrammetry lower on bare earth. LiDAR higher — but the only method producing valid results for vegetated terrain.
5. Mission Workflow: From Flight Planning to Deliverable
5.1 Pre-Flight
- Define project requirements: accuracy grade, deliverable types, coordinate reference system.
- Set ground control points: 5-8 GCPs across the site, surveyed with RTK GPS, using 60cm by 60cm checkerboard targets.
- Plan the flight in DJI Pilot 2 or DJI Terra: define survey boundary, set GSD, set overlap percentages, verify obstacle clearance.
- Check weather: wind under 10 m/s, no precipitation, cloud ceiling above planned altitude.
5.2 In-Flight
The aircraft flies parallel transects at constant altitude and speed. The operator monitors battery levels, RTK fix status, and data storage remaining on DJI Pilot 2. At 50% battery, the aircraft returns to the home point, the battery is swapped in 90 seconds, and the mission resumes from the interruption point automatically.
5.3 Post-Processing
- Transfer raw images or point cloud data from the sensor to the processing workstation.
- DJI Terra processing: for photogrammetry — aerial triangulation, dense point cloud, mesh reconstruction, orthomosaic output. For LiDAR — point cloud registration, strip adjustment, ground classification, DTM/DSM generation.
- Quality check: verify accuracy against GCPs. RMSE at each GCP should be within the project's specified tolerance.
- Export: GeoTIFF for orthomosaics, LAS/LAZ for point clouds, DXF/DWG for contour lines, OBJ/FBX for 3D models.
6. How to Choose: A Decision Framework
- Is the terrain vegetated? If yes, LiDAR is the only method that produces usable ground elevation data. If no, photogrammetry is the cost-effective choice.
- What deliverables does the client require? If they need an orthomosaic with true-to-scale imagery, photogrammetry is mandatory. If they only need a DTM and contour map, LiDAR alone suffices.
- What is the project timeline? If the deadline requires night operations or work during winter with limited daylight, LiDAR's light-independence justifies the higher equipment cost.
For most commercial surveying companies, the pragmatic answer is: own both sensors. Use the P1 for 70% of projects covering construction sites, open-pit mines, and bare-earth topographic surveys. Use the L2 for the 30% that involve vegetation, corridors, or tight deadlines. Both sensors mount on the same DJI Matrice 350 RTK platform, share the same batteries and controller, and process through the same DJI Terra software environment.
7. Summary: The Right Tool for the Surface
LiDAR and photogrammetry are not competing technologies — they are complementary tools that solve different measurement problems. Photogrammetry produces photo-realistic orthomosaics at lower cost for open terrain. LiDAR sees through vegetation and works in darkness at higher cost. The professional surveying operator keeps both in the toolkit and selects based on the project's surface conditions, not the equipment catalog's feature list.
Configuring a drone mapping payload for your surveying operations? We supply integrated solutions — DJI Matrice 350 RTK platforms with Zenmuse L2 LiDAR or P1 photogrammetry payloads, RTK base stations, ground control point kits, and DJI Terra processing software. Contact our technical team with your typical project terrain, accuracy requirements, and monthly survey volume for a customized payload configuration and pricing proposal.
