DroneDeploy's three-hour benchmark for a 1,500-image map depends on what you feed it
The July release adds faster Aerial Pro processing, denser 3D output, revised volumetrics, and new dock-health tools, while several headline AI and robotics features remain early access.
By Joel Kelly
Decision brief
Faster processing matters when it shortens a real delivery window, but a buyer should test the same aircraft, overlap, control, coordinate system, network, site size, and output on representative jobs before pricing around the company's best-case benchmark.
- 300 images
- < 1 hour
- 1,500 images
- < 3 hours
- Dock uploads
- 99.9%
Company-reported Aerial Pro benchmark
Company-reported megaproject benchmark
Vendor-reported success after infrastructure changes
Benchmark
Processing can be fast without the deliverable being finished.
DroneDeploy says its new Aerial Pro processing mode can finish a 300-image map in under one hour and a 1,500-image project in under three hours. The company describes large-dataset processing as up to twice as fast as standard mode, with twice as many 3D points on its 1,500-image comparison and as much as ten times the density on the largest sites. Those are company benchmarks, and the release does not publish the source imagery, hardware, control, accuracy checkpoints, network conditions, processing queue, or comparison methodology.
The practical promise is same-day delivery. A crew could fly in the morning and place a controlled map in a superintendent's hands before the shift ends. That only holds when the complete chain cooperates: capture plan, exposure, overlap, RTK or control, upload, processing, quality review, coordinate system, annotation, and export. A fast model that needs a reflight or cannot support the intended measurement is not a faster deliverable.
Measurement
A cleaner volume report still inherits every decision made before export.
The release also changes volumetric work. Users can apply compaction or swell factors to a volume annotation and carry the adjusted values into CSV exports. DroneDeploy says measurements now compute more precisely in the project's coordinate reference system. The feature can make a report easier to reconcile with estimating assumptions, but the factor itself still needs an accountable source and the underlying surface still needs appropriate control and review.
For docked operations, DroneDeploy reports 99.9 percent upload success from new infrastructure and adds a connection-health panel for the dock, aircraft, and network. The company's operating guides show why that last layer matters: media upload depends on stable connectivity, correct provisioning, compatible firmware and hardware, and a successful automated flight and transfer. The published 99.9 percent figure is a vendor-wide claim, not a guarantee for one site.
Proof run
Test the workflow on one job you already understand.
A useful benchmark holds the capture and output requirements steady, then measures both machine time and human correction.
- 01
Freeze the input
Use the same image set, coordinate system, control, and output targets for each processing mode.
- 02
Check the surface
Compare checkpoints, noise, seams, stockpile boundaries, and areas with weak geometry.
- 03
Time the people
Record upload, setup, processing, QA, annotation, export, and any rework separately.
- 04
Test recovery
Interrupt a transfer or reject a map and document how the team retrieves raw data and restores the job.
Availability
The release mixes shipping tools with early-access systems.
The release is broader than the generally available aerial updates. A new Ground app adds field notes, two-way Procore and Autodesk sync, mobile DXF overlays, and other field-inspection functions. Ground Pro RTK scanning, Progress AI schedule matching, and a robotic Capture Agent are marked early access. They should be evaluated as emerging workflows rather than features every customer can deploy immediately.
Buying test
Price the hours the software removes, not the features in the announcement.
For a small mapping business, the buying question is not whether the release sounds fast. It is whether the software removes enough labor, delay, rework, and uncertainty from the exact deliverable to cover its price. Run a representative project through the standard and premium modes, keep the raw imagery, compare checkpoints and volumes, time every human step after processing, and document failure recovery before changing a client promise.
Practical read
Where faster processing can change the margin
Strong fit
- Recurring construction, earthwork, quarry, and large-site capture with same-day decisions
- Teams that already use control, coordinate systems, checkpoints, and repeatable QA
- Dock programs where network and upload reliability are measured as part of the operation
Account for
- Comparing vendor benchmarks with a different aircraft, image count, overlap, control, or output
- Treating point density as a substitute for validated accuracy or fit for purpose
- Selling early-access AI or robotics functions as generally available client deliverables
Source file
Verify the reporting
Continue on Drone Authority
Put the report to work
Continue this beat
