Cowboy guide to drone mapping: pragmatic, no bullshit

1. Concepts you need to know without the theory shock

Drone mapping or photogrammetry is simply taking overlapping photos from the air and letting a computer stitch them into 2D or 3D models. Here are the key parameters:

• GSD (Ground Sampling Distance): How big one pixel is in reality, for example 1.5 cm/pixel. Lower is sharper, but you must fly lower and take many more photos.
• Front and side overlap: How much photos overlap in flight direction and sideways, typically 70 to 80 percent.
• Gimbal angle: The angle your camera points. Nadir (90 degrees) looks straight down, perfect for flat 2D maps. Oblique (for example 60 to 80 degrees) looks forward or down at an angle, essential for capturing sides of a chimney or building for 3D.
• Single vs. double lawn mower grid: A single grid is the classic back and forth. A double grid adds a second flight 90 degrees to the first. It takes longer but closes gaps in your data.
• Orthophoto (2D): A flat, measurable aerial photo corrected for lens and perspective distortion.
• Elevation model (DSM / DTM): A digital map of terrain and surface height.
• Point cloud and 3D mesh: A 3D model built from millions of coordinated points.

2. The myth of GCPs, RTK and the bowl effect

Every time you mention drone mapping, someone shouts: You need GCPs and RTK or your map is useless. Let us debunk that myth.

Relative vs. absolute accuracy

• Absolute accuracy: Is my map millimetre perfect on the global cadastre? Spoiler: This is the unicorn of photogrammetry. Even Google Earth satellite images are rarely millimetre precise.
• Relative accuracy: Is a 10 metre wall on my map also 10 metres in reality?

What is RTK and do I need it?

RTK (Real-Time Kinematic) is GPS on steroids. While normal drone GPS can be a few metres off, an RTK drone constantly talks to a base station or network to know its exact position in the air down to centimetres.

• What does it do better? It gives you absolute accuracy out of the box. You largely avoid physical GCPs because the drone already knows exactly where it took the photo.
• Why is it often overkill? It costs a fortune, both a pricier drone and often an expensive mobile network subscription. If you just need to measure a roof or a pile volume, standard GPS and relative accuracy is plenty.
• When can it be needed? If your map must align with existing survey data, municipal utility maps, or if you fly the same site every week and want maps from week 1, 2 and 3 to stack perfectly automatically.

Why use GCPs? (Ground Control Points)

3. Dataset hygiene: do not mix pears and bananas

A classic mistake is to fly a building thin with both 90 degree nadir and oblique photos, dump all 500 images in one folder and expect the computer to magically spit out both a perfect 2D map and a 3D model. It does not work. Here is the rule when the client wants both:

• Dataset A (for 2D orthophoto): Use only nadir (90 degrees). For a flat map the camera must look straight down. Fly a grid over the roof or area. If you feed the engine photos looking into a wall or toward the horizon, it chokes trying to force them into 2D. Separate your nadir images and use them only for the orthophoto.
• Dataset B (for 3D model): Use only oblique. Nadir images do not work for 3D, the roof becomes flat and the chimney has no sides. Fly a new pass with the gimbal at for example 60 to 75 degrees for depth, and orbit the facades. Collect these in a separate folder and use them only for the 3D model.

4. Risk, gear choice and the expensive refly

In drone business the math for all mapping is: Flight time plus number of images equals cost. Batteries, time on site and heavy server processing cost money. But before you cut everything to the bone to save money, know the real pitfalls:

1. Risk factor (why use a cannon for sparrows?)

Every flight is a risk. Birds, wind gusts and invisible cables exist. Why risk smashing an enterprise drone with a 150,000 kr. camera over a construction site if the job is just a 2D orthophoto at 2 cm/pixel? A small DJI Mini 5 Pro can deliver a result few mortals and clients can tell apart. Choose gear by risk.

2. The refly penalty

Mathematically double flight time and double images sounds like a budget killer. But the most expensive thing is not taking 50 photos too many. The absolute most expensive is sitting at the office, discovering overlap was too low, and being forced to pack the car and drive out to refly the damn thing. That is why we often fly a double grid even though we think a single grid is enough. It is cheap insurance.

3. The scale paradox and geography

What is smart depends entirely on job size and location:

• The small roof (200 m2): Here it takes 2 minutes extra to take twice as many photos. The drone still runs on the same battery. Overkill costs nothing but secures the data.
• The big field (80 ha): Here double flight time is a disaster. You go from 1 battery to suddenly needing 4 batteries and field charging. Here you cut to the bone and fly minimal overlap.
• Geographic distance: Is the job 2 hours away by car? Then always fly with a massive safety margin of extra photos. You do not want to drive 2 hours back tomorrow.

5. Real world examples (use cases)

Here are 3 everyday scenarios:

Frequently asked questions

Sources and further reading

If you want to see how we handle your images or what it costs, start here.

• GDPR-compliant drone processing in the EU

  Where your data lives, how long we keep it, and who can see it.

• Do you need a subscription?

  When pay per job makes sense and what 2,000 images cost.

• Pricing

  See current prices for orthophoto, DSM and full 3D on the front page.