Processing performance — real-world benchmarks

DroneField runs entirely on your own machine — no cloud upload, no processing queue, no recurring storage fees. The total time from flight to application map depends on three things: hardware, flight configuration and number of images.

Below are real measurements from a typical production setup.

Benchmark: 20 hectares

Flight configuration:

• Altitude: 120 m AGL
• Overlap: 70–80% (front and side)
• Camera: standard multispectral drone (DJI Phantom 4 Multispectral or Mavic 3 Multispectral)
• Target GSD: 5 cm/px (DroneField default)

Hardware used:

• Mac mini M4 / 24 GB RAM

Total processing time: under 5 minutes — from the start of processing to a finished orthomosaic and NDVI analysis, ready for building the application map.

How the time scales

DroneField scales with:

• Available RAM — more images held in memory in parallel
• CPU cores — image alignment is parallelised
• GPU — orthomosaic blending is hardware-accelerated where supported

For larger areas, a finer GSD (3 cm/px) or older hardware, the time increases proportionally. For smaller areas and a coarser GSD (10–20 cm/px) it is significantly shorter.

The advantage of local processing

Because it runs locally, DroneField avoids:

• Uploading gigabytes of imagery (which can take hours on a slow rural connection)
• Waiting in a cloud processing queue
• Recurring storage costs charged per hectare or per flight
• Depending on the vendor’s server being available
• Privacy risk to your own field data

The result: from flight to application map in minutes, in the field, without an internet connection — which matters most in remote areas where connectivity is limited.

Minimum vs recommended

See System requirements for the minimum supported configuration. The benchmark above represents a recommended production setup for daily professional work.