Ground sample distance calculator
One pixel, measured on the ground. Work it forwards from altitude, or backwards from the resolution the job demands.
Researched from published specifications and verified owner reviews · updated 2026
The short answer
Ground sample distance in centimetres per pixel equals (sensor width in mm x altitude in m x 100) / (focal length in mm x image width in px). A 1-inch 20 MP camera with a 24 mm equivalent lens flying at 100 m gives roughly 2.7 cm per pixel, and GSD scales linearly with altitude.
Ground sample distance is the real-world size of one image pixel on the ground. It is the only resolution figure that means anything in aerial work, because it combines the camera and the altitude into a single number that says exactly how much detail your imagery contains. Megapixels alone tell you nothing: a 48 MP camera at 120 m resolves less than a 12 MP camera at 25 m.
Use it in both directions. Forwards, to find out what resolution a planned altitude will give. Backwards, to find the altitude required to hit the resolution a client asked for, which is the version that actually changes your flight plan.
Resolution
GSD and target altitude
Focal length must be the actual focal length in millimetres, not the 35 mm equivalent. If you only know the equivalent, the presets below convert it for you.
Estimates only, from published specifications. Wind, temperature, altitude, payload and pack age all move the real number, usually in the direction that costs you flight time. Always keep a reserve and follow the manufacturer's own limits.
The formula, and why each term is there
GSD in centimetres per pixel equals sensor width in millimetres, times altitude in metres, times one hundred, divided by focal length in millimetres times image width in pixels. The hundred is just the metre-to-centimetre conversion. Everything else is a similar-triangles argument: the sensor and the focal length define an angle, the altitude projects that angle onto the ground, and the pixel count divides the result into pixels.
Two practical consequences fall out of the algebra. First, GSD is linear in altitude, so flying at half the height gives twice the resolution, exactly. Second, GSD is linear in focal length too, in the opposite direction: a 70 mm equivalent telephoto at 100 m gives about the same pixel size as a 24 mm wide lens at 34 m. That is the single most useful fact for inspection work, because it lets you get roof-level detail from a standoff distance that is both safer and less intrusive.
Resolution versus accuracy
These are different claims and clients conflate them constantly. Resolution is how small a thing you can see. Accuracy is how close its measured position is to reality. Flying lower improves the first and does nothing at all for the second.
A photogrammetry model built from aircraft GPS alone typically lands within a few metres horizontally and worse vertically, no matter how sharp the imagery is. Absolute accuracy comes from surveyed ground control points or from onboard RTK as on the Autel EVO II Pro RTK V3 ($2,999). If a client says "I need centimetre accuracy", they are asking about control, not about altitude, and answering with a lower flight is the wrong answer.
GSD by altitude for common cameras
| Altitude | 1/1.3-inch, 12 MP | 1-inch, 20 MP | 4/3, 20 MP | 1-inch, 48 MP mode |
|---|---|---|---|---|
| 20 m (66 ft) | 0.71 | 0.54 | 0.70 | 0.35 |
| 40 m (131 ft) | 1.42 | 1.07 | 1.40 | 0.70 |
| 60 m (197 ft) | 2.13 | 1.61 | 2.10 | 1.05 |
| 80 m (262 ft) | 2.84 | 2.15 | 2.80 | 1.41 |
| 100 m (328 ft) | 3.55 | 2.68 | 3.50 | 1.76 |
| 121 m (397 ft) | 4.30 | 3.25 | 4.24 | 2.13 |
Figures are centimetres per pixel, assuming a 24 mm equivalent lens. The full grid, including telephoto options and imperial altitudes, is on the GSD and altitude chart.
The three-to-five pixel rule
A feature needs roughly three to five pixels across it before a person or an algorithm can reliably identify what it is, and considerably more before anyone can measure it. That converts a resolution requirement into a GSD requirement quickly: to identify a 10 cm object you want a GSD around 2 to 3 cm/px, and to identify a 5 mm crack you want 0.1 to 0.2 cm/px, which is why crack detection is flown at low altitude or with a telephoto.
Be sceptical of the theoretical number. Motion blur, lens softness away from the frame centre, atmospheric haze on a long lens, and the resampling inside photogrammetry software all remove detail that the geometry says should be there. If a job is close to the edge of what the GSD allows, fly a test line and look at the imagery before you commit to the deliverable.
Motion blur is the resolution killer nobody budgets for
If your aircraft moves more than about half a pixel of ground during the exposure, the image is already softer than its GSD. At 8 m/s ground speed and a 1/500 s shutter, the aircraft covers 1.6 cm during the exposure, which is comparable to a 2 cm GSD. That is a visible loss.
The fixes are all boring and all effective: fly slower, use a faster shutter, and use an ND filter only when you actually need to keep the shutter down. Mapping is the one context where the 180 degree shutter rule does not apply, because you are capturing stills for measurement rather than video for motion. Shoot as fast a shutter as the light allows. For video work, where the rule does apply, use the ND filter and shutter calculator.
Where this fits in a mission
GSD is the first decision in a mapping flight and it constrains every other one. Set the target resolution, get the altitude from this page, then take that altitude to the mapping mission planner to get line spacing, image count, airborne time and pack count. Check the pack count against the battery count calculator and the image count against the memory card capacity calculator.
On the gear side, sensor size is what buys you resolution at a sensible altitude. The options and what genuinely separates them are in best drones for mapping and surveying, and the complete kit at a stated total is the commercial mapping build.
Questions people ask
+ What is ground sample distance in plain terms?
Ground sample distance is the real-world size of a single image pixel measured on the ground. A GSD of 2 cm per pixel means one pixel covers a 2 cm square of the world. It is the honest measure of aerial resolution because it accounts for the camera and the altitude together, which megapixel count alone never does.
+ What GSD do I need to see a crack in a roof or a road?
A feature needs roughly three to five pixels across it before a human or an algorithm can reliably identify it. A 5 mm crack therefore wants a GSD near 0.1 to 0.2 cm per pixel, which means flying very low or using a longer lens. In practice, roof condition work is flown at 12 to 25 m and pavement crack detection at 20 to 35 m with a 1-inch or larger sensor.
+ Does a higher megapixel camera improve GSD?
Yes, if the sensor size stays the same, because more pixels across the same sensor width means each pixel covers less ground. Going from 20 MP to 48 MP on the same 1-inch sensor improves GSD by roughly 55 percent at the same altitude. The caveat is that the extra pixels are physically smaller, so image quality in low light gets worse and the improvement is smaller than the number suggests.
+ Is GSD the same as accuracy?
No, and confusing the two is the most common mistake in drone mapping. GSD is resolution, which is how small a thing you can see. Accuracy is how close the position of that thing is to its true coordinates. A model can have a 1 cm GSD and be two metres out of position. Accuracy comes from ground control points or RTK, never from flying lower.
+ How does GSD change with a zoom or telephoto camera?
GSD improves in direct proportion to focal length, so a 70 mm telephoto at 100 m gives roughly the same GSD as a 24 mm wide lens at 34 m. That is genuinely useful for inspecting a structure you cannot fly close to, and much less useful for mapping, because the narrow footprint means far more images and far more flight lines to cover the same area.
+ Why does my delivered orthomosaic look softer than the GSD suggests?
Because GSD is a geometric limit, not an optical one. Motion blur from flying too fast for the shutter speed, lens softness at the frame edges, atmospheric haze, and the resampling that happens when photogrammetry software builds an orthomosaic all subtract from the theoretical figure. Assume the delivered product resolves noticeably less detail than the raw GSD number implies.
