How It Works
From upload to deliverable
in four steps
No software to install. Upload your processed photogrammetry output and receive professional deliverables — typically within 24 hours, though large or complex captures can take longer.

The app’s upload step · the raw photos straight off the card, or a processed export
- The raw flight photos (JPG/PNG — no zipping)
- or a DJI Terra export (zip)
- Pix4D / Metashape / WebODM export
- Orthomosaic, DSM/DEM (GeoTIFF)
- Point cloud (LAS/LAZ) or 3D mesh (OBJ/PLY)
Upload Your Photos — or a Finished Export
New Project opens on “Photos from my drone” — drag the photos straight off the SD card, or the folder that holds them. No zipping: your browser bundles them for you. We run the photogrammetry — orthomosaic, surface and terrain models, contours, point cloud and 3D model — then carry on through the same pipeline. It takes longer than uploading a finished export, and reconstruction quality depends on the flight: overlap and coverage decide whether the map is usable, and no software recovers from a capture that never saw the ground.
Already processed it? Choose “A processed export” and drag and drop your photogrammetry export into a new project — from DJI Terra, Pix4D, Metashape, WebODM, or whatever you already run (included with Pilot and Fleet, and during the free trial). DroneFlow.Ai validates file types, coordinate reference data, and your plan's upload limits before packaging starts. See every format you can upload
Positions come from the coordinates your images already carry, which anchor the result to the real world; where a DJI flight held a genuine RTK fix, its per-photo precision weights the solve, and RTK that was on but never fixed is positioned as standard GNSS, weighted accordingly, and the project page says so. Photos carrying only standard GPS still process — expect lower absolute accuracy. Actual accuracy depends on capture design, correction quality, and ground control; outputs are not a licensed land survey unless certified by a licensed professional.
Ortho · surface model · 1 ft contours · point cloud (9.2 M points, LAZ) · textured mesh
- Auto CRS detection + reprojection
- KMZ + web-tile packaging
- Branded PDF report
- Cross-layer validation
Automated Packaging
Your upload moves through the same QGIS/GDAL deliverables pipeline AeroLakeland has run for months — georeferenced, re-tiled, and packaged for the web with no manual GIS work on your end. On this path DroneFlow.Ai doesn't need to reconstruct anything: the photogrammetry is already done, so it picks up your finished output and carries it the rest of the way. Bring raw photos instead and it runs the reconstruction for you first — same pipeline from there on.

5,873,939 surface cells checked · 73.0 ft spread, 1st–99th pct · measured, not graded
- Tileset health checks
- Elevation cross-validation
- Coverage verification
- Second review pass
Automated QA Review
Before anything reaches your portal, it's checked against a real, repeatable QA pass — the same one used to catch real issues on real projects.

Case Study #1 as delivered · 32.6 acres at 1.7 cm per pixel
- Direct downloads
- Revocable client links
- View-count / date expiry
- No client account needed
Download or Share
Download your files directly, or hand a client a link that expires by date or view count — no download link that lives forever in someone's inbox.