The DJI Zenmuse L2 and L1 are capable LiDAR payloads that produce LAZ point clouds and GeoTIFF orthomosaics from a standard drone platform. The capture workflow is well-documented by DJI. The delivery workflow — getting these files from your desktop to a client who can actually see them — is not, and that is where most operators leave value on the table.
Here is the complete picture: what the L2 and L1 produce, what the processing looks like in DJI Terra, and how to go from processed outputs to a professional client delivery.
What the Zenmuse L2 and L1 produce
The L2 and L1 are Livox LiDAR sensors integrated with a nadir-facing RGB camera, mounted on DJI Matrice drones (the M300 RTK and M350 RTK are the primary platforms). During a flight, the LiDAR fires laser pulses at a high scan rate and records the time-of-flight return for each — each return becomes a point in the final cloud. The integrated camera captures RGB photos simultaneously, which are used to colour the point cloud.
Raw data from the sensor:
.lvx2LiDAR data files (proprietary Livox format, processed by DJI Terra).jpg/.DNGRGB images from the integrated camera- GNSS and IMU trajectory data
After processing in DJI Terra:
| Output | Format | Notes |
|---|---|---|
| Point cloud (merged, classified) | LAZ | Ground (class 2), vegetation (3–5), buildings (6) |
| Orthomosaic | GeoTIFF | Georeferenced, stitched from RGB camera |
| Digital surface model (DSM) | GeoTIFF | Highest return elevation raster |
| Digital terrain model (DTM) | GeoTIFF | Ground return elevation raster |
| Quality report | Accuracy summary, GCP residuals, point density |
The LAZ and GeoTIFF outputs are industry-standard formats. The LAZ is a compressed LAS file — see what is a point cloud for a full explanation of the format. The GeoTIFF is a georeferenced raster image that sits correctly in the world coordinate system you specified during processing.
The delivery problem
You now have a folder on your desktop containing pointcloud.laz, orthomosaic.tif, dsm.tif, dtm.tif, and quality_report.pdf. The challenge: your client can open the PDF, cannot open the LAZ, and probably cannot open the GeoTIFFs either without GIS software.
If you send a Dropbox link, the client will:
- Download the PDF and read it
- Try to open the LAZ file, hit an error, and give up
- Try to open the
.tiffiles in Preview or Photos, and get a black square or nothing
The LAZ and GeoTIFF files are perfectly processed. They are simply inaccessible to anyone without CloudCompare and QGIS installed — which is most of your clients.
This is the delivery gap. It is not a data quality problem. It is a viewing problem, and it is entirely solvable.
The professional delivery workflow
Upload your DJI Terra outputs to a spatial data delivery platform. Swyvl classifies each file automatically and assigns the correct viewer. Your client gets a single portal link that opens everything in their browser.
Step-by-step:
- Complete processing in DJI Terra. Export the LAZ, GeoTIFFs, and report PDF.
- Log into Swyvl. Create a site (the physical location) and a capture session (the date of the flight).
- Upload all outputs. Swyvl detects the file types: LAZ gets the Potree viewer, GeoTIFFs get the MapLibre viewer, PDF gets the inline reader.
- Optionally, add a context note for the client — the coordinate system, the GCP accuracy, any areas of note.
- Send the share link to your client.
The client opens the link. They see the point cloud in a Potree viewer: they can orbit, zoom, measure point-to-point distances, and slice cross-sections. They see the orthomosaic as an interactive map with zoom and pan. The quality report is inline — they do not need to download anything to read it.
This is a complete delivery. The client engages with the actual data, not just a summary PDF.
How each format renders in the browser
LAZ — Potree viewer
Potree is the standard for browser-based point cloud rendering. It uses an octree structure to stream point cloud data progressively — the viewer loads a coarse version immediately, then adds detail as the user zooms in. This means even very large LAZ files (several GB) load quickly in the browser.
Your client can:
- Orbit, pan, and zoom freely
- Switch colouring modes: RGB from the camera, elevation gradient, intensity from the LiDAR return, or classification colour coding
- Toggle classification classes on and off (useful for showing ground vs vegetation vs buildings)
- Measure point-to-point distances and area
- Create elevation profiles / cross-sections
GeoTIFF — MapLibre viewer
Orthomosaics and elevation models render as interactive map layers. The GeoTIFF is tiled on the server and served as raster tiles — the client sees the image load progressively from low to high resolution as they zoom. They can overlay the orthomosaic on a satellite basemap for context, or switch to the DTM/DSM to examine the elevation model.
PDF — inline reader
The quality report opens in an inline PDF reader. The client does not need to download it. They can scroll through the accuracy summary, GCP residuals, and point density statistics without leaving the portal.
DJI Terra vs post-processing in Metashape
DJI Terra is a capable processing tool and the natural starting point for L2 and L1 data. It is tightly integrated with DJI’s proprietary data formats, and for routine deliverables — construction site surveys, topographic surveys, vegetation mapping — it produces solid results quickly.
For engineering-grade deliverables, consider post-processing in Agisoft Metashape, LiDAR360, or LAStools:
| Consideration | DJI Terra | Metashape / LiDAR360 |
|---|---|---|
| Processing speed | Fast | Slower |
| Ground classification | Automated, limited control | More parameters, better results on complex terrain |
| Accuracy with GCPs | Good | Better — finer control over GCP weighting |
| Point cloud colouring | RGB or intensity | Full control, multi-return handling |
| Output formats | LAZ, GeoTIFF, report | LAS, LAZ, PLY, E57, many others |
| Cost | Included with L2/L1 | Separate licence (Metashape Standard ~$179) |
For most client deliveries, DJI Terra is the right choice. If you are delivering to a client who will import the point cloud into a structural analysis tool, a BIM workflow, or a geotechnical model, Metashape gives you better control over the output quality.
Point cloud colouring options
The Zenmuse L2 and L1 offer two inherently different colouring options that are worth understanding before delivery, because they produce very different client experiences.
RGB from the integrated camera
The point cloud is coloured using the RGB values from the photos taken during the flight. This gives you a photorealistic point cloud that looks like a 3D aerial photograph — every surface is the colour it actually is in real life. RGB colouring is the most intuitive option for clients who are not spatially technical. They see a 3D representation of the site that looks recognisably real.
The limitation: RGB colouring depends on good lighting conditions during the flight. Shadows, overcast skies, and direct sunlight at low angles can produce patchy or washed-out colouring.
Intensity from the LiDAR return
Intensity values represent how strongly each surface reflects the laser pulse. Asphalt is dark. Fresh paint lines are bright. Grass has a characteristic signature. Concrete has another. Intensity-coloured point clouds look grey-scale and technical — they do not look like photos.
Intensity colouring is useful for surveyors and engineers who want to distinguish surfaces by material rather than by colour. It is less useful for non-technical clients who are trying to orient themselves in the data.
For most client deliveries, RGB colouring is the right default. If you are delivering to a technical client who will be doing classification or analysis work, include both options.
Connecting the delivery to your broader workflow
The L2 and L1 sit within a broader drone survey workflow. If you are delivering repeat surveys of the same site — construction progress monitoring, mine volumetrics, environmental monitoring — the delivery platform needs to support organised time-series delivery, not just one-off uploads.
Swyvl organises deliverables by site and capture session. A client monitoring a construction site sees the January, February, and March surveys as separate sessions, each with the orthomosaic, point cloud, and report from that flight. They can navigate back through the history and compare periods visually.
For operators doing one-off surveys — a topographic survey, a planning application support, a stockpile volume calculation — the workflow is the same, just without the repeat session structure.
See how to deliver drone survey data to clients for the broader workflow context, and for drone operators for platform-specific guidance on setting up your delivery workflow in Swyvl.
What the client actually needs
The most important question is not which software you use to process the L2 data. It is whether your client can engage with the data you deliver.
A project manager overseeing a subdivision does not need CloudCompare. They need to be able to look at the orthomosaic, orient to the site, and understand what the survey captured. A browser-based delivery gives them that.
A civil engineer reviewing the point cloud for volume calculations needs the raw LAZ — and they probably have the software to open it. But they should also be able to view it in the browser first, to verify it looks correct, before importing it into their CAD workflow.
A client who can see the data you delivered is a client who understands what they paid for. That understanding translates into repeat work, referrals, and a professional relationship built on demonstrated value rather than a folder of files nobody opened.