How to measure a roof from a drone photo
A drone photo measures a roof the same way satellite imagery does, once it has a scale. The shot to fly, how to set the scale from one known length, and where it goes wrong.
You measure a roof from a drone photo by flying one straight-down shot that holds the whole roof, giving the image a scale from a single length you know, then tracing the outline and applying the pitch factor to each plane. The photo replaces satellite imagery as the base layer; the routine after that is the same one in how to measure a roof from satellite imagery, and every step below assumes you know it.
Here is a worked job: a 1,900 sq ft rambler with an attached garage, trees over the back half, and the newest satellite image a couple of years old. The drone is in the air for about four minutes. The measurement takes longer than the flight.
Why a drone photo needs a scale
Satellite imagery arrives georeferenced: each pixel maps to a known distance on the ground, so a tracing tool reports feet by itself. A drone photo is just a JPEG. The camera knows its altitude roughly, not precisely enough to turn pixels into inches. So the photo measures nothing until you tell the software how long one thing in it is. That one length scales every line you draw afterwards; get it wrong by 5% and every area and bundle count is off by 5% with it.
The shot to fly
- Straight down. Camera at 90 degrees, not tilted. An oblique shot is useful for pitch and for photographing damage, but its outline is distorted by perspective and cannot be traced to scale.
- High enough for margin. Keep the roof in the middle 60% of the frame, away from the lens edge where distortion is worst. If the roof fills the frame, go higher.
- Centred over the ridge. A shot from over the driveway shows one slope more than the other. Hold position over the middle of the house.
- Something of known length in view. A garage door, a parked pickup, a tape laid flat on the driveway.
- Flat light. Overcast beats noon sun. Hard shadows hide the shaded eave and tempt you to trace the shadow line.
Fly it twice so you have a spare, and check sharpness before you land. What any usable image needs is in roof measurement from aerial photos: what a usable image needs.
Set the scale from one known length
Upload the photo and find your reference. On the rambler it is the double garage door, taped at the jamb before the flight: 15 ft 10 in. Draw a line across the opening on the photo, enter the length, and the image is scaled. Three habits make this reliable:
- Measure the reference yourself. A "16 ft" door is often 15 ft 8 in or 16 ft 2 in at the opening. A pickup runs 18 to 22 ft depending on cab and bed. Tape it or skip it.
- Pick something long. A 2 ft error on a 16 ft reference is 12%. The same 2 ft on a 40 ft driveway edge is 5%.
- Prefer a reference at roof height on a tall house. The driveway is further from the lens than the roof, so it reads slightly smaller per foot. On a two-storey, scale from a ridge vent or a dormer width you taped before, or expect the roof to read a few percent large.
In RoofStruct the upload flow asks for exactly this: draw a line over a length you know, type the length, and every measurement follows it. There is also an AI scale finder that looks for objects of known size in the photo, such as vehicles, garage doors or a printed scale bar; it runs on AI credits, and a free account starts with 10 of them for uploads. A roof traced on satellite imagery is scaled automatically and never needs this.
Then check the scale against a second known length. The front porch slab is 8 ft deep. Draw it on the photo: 8 ft 1 in means the scale is good; 8 ft 9 in means the reference is wrong, and you fix it before tracing a single roof line.
Trace the outline and label the edges
From here the routine is the satellite one. Trace each facet to the drip edge, not the wall. Label every edge as eave, rake, ridge, hip or valley so the line totals come out for starter, drip edge, ridge cap and valley metal.
| Facet | Flat area (sq ft) | Pitch | Roof area (sq ft) |
|---|---|---|---|
| Main front | 760 | 5/12 | 823 |
| Main back | 760 | 5/12 | 823 |
| Garage front | 310 | 4/12 | 327 |
| Garage back | 310 | 4/12 | 327 |
| Total | 2,140 | 2,300 |
The flat area comes straight from the traced polygons. The roof area is the flat area times the pitch factor for each plane, 1.083 at 5/12 and 1.054 at 4/12; the table and the maths are in how to calculate roof pitch. That is 23 squares before waste, about 25 squares of shingles at a 10% waste factor.
Note the trees. The satellite image hid the back half of the main roof under canopy; the drone flew under the tree line, and the back facet measures 760 sq ft rather than the 690 sq ft a guess from the front would have given.
Get the pitch on the same flight
A straight-down photo tells you nothing about pitch; the roof looks flat from above. Drop to eave height, back off until a gable end fills the frame, and photograph it square on. Count rise over 12 inches of run against the gable trim, or measure the rake and the half-span and work the triangle. A hip roof has no gable, so shoot the end hip square instead, or take the pitch from the ground with the methods in how to measure a roof without climbing it.
Where you trip
- Tracing an oblique photo. Perspective distorts the outline and the scale changes across the frame. Only a straight-down shot can be traced to scale.
- Scaling from a length you never measured. Garage doors, pickups and fence panels all come in more sizes than you think.
- Trusting the drone's altitude readout. Barometric altitude drifts by several feet. It is not a scale.
- Facets at the edge of the frame. Lens distortion stretches the outer part of the image, so the far garage reads bigger than the near one. Keep the roof in the middle.
- Shadow lines as eave lines. A hard shadow on the north side looks like an edge. Trace the drip edge.
- Flying without clearance. Airspace rules, altitude limits and the rules around people and property vary by country and change. Know them before the job.
FAQ
Is a drone photo more accurate than satellite imagery for roof measurement?
It can be: it is sharper, current to the day, and flies under tree cover. It is only as accurate as its scale. A taped reference lands in the same range as a good satellite tracing; a guessed one does not. Which source wins on which job is in satellite vs aerial vs drone roof measurements.
Do I need a survey-grade drone or ground control points?
Not for a shingle order. Ground control points and photogrammetry are for mapping and volumes. One sharp straight-down photo and a taped reference gets a residential roof within ordering tolerance.
What if nothing of known length is in the photo?
Land, lay a tape or a measured board flat on the driveway in view of the camera, and fly the shot again. Thirty seconds on the ground saves a bad scale.