Orthophoto from Drone Photos: A Photogrammetry Guide

Produce an orthophoto, DSM/DTM, a classified point cloud and class layers from drone photos as one package — in the CartaX cloud or on your own computer with CartaX Studio; photo preparation, GCPs and RTK, delivery coordinate system and time expectations.

2026-09-14 · 6 min read

A flight's photos are not a product on their own; to become measurable they have to be tied together and placed on the map through photogrammetry. CartaX does this in a single flow: you point it at the photo folder, the standard product package is produced together, and it is uploaded straight into your CartaX project. You run production either in the CartaX cloud (from the CartaX Studio page in the web app, or with Studio's Produce in the cloud button) or, on a capable computer, on your own machine with CartaX Studio. This guide explains what to watch for to get the best result. For installing Studio see What is CartaX Studio?.

What is produced?

Every job produces the same package at the same high quality; there is no output or quality selection:

Product	Format	What it is for
Orthophoto	GeoTIFF (COG)	A scaled, georeferenced aerial image; the base of the map.
Elevation data (DSM + DTM)	GeoTIFF (COG)	Surface and terrain elevation models; elevation readout, slope and volume on the map are calculated from these.
Classified point cloud	LAZ → Octree	A dense 3D point cloud; measurement, sections; with classes such as ground, vegetation and buildings.
Segmentation layers	Vector	Objects recognised on the orthophoto (buildings, roads, trees, water, vehicles…); open in the Classes window.

A textured 3D model (mesh) is not part of the production package today.

Preparing the photos

Result quality comes more from the photos than from the engine:

Overlap: at least 75% forward, at least 65% side. Low overlap = holes in the model and drifting edges.
Photo count: hundreds of photos is normal; both Studio and the CartaX Studio page on the web scan the root folder including its subfolders (JPG/PNG/TIF) and find the RTK .MRK logs by themselves.
Position data: photos should carry EXIF GPS. On RTK drones (e.g. DJI .MRK) centimetre-level positions are used automatically: when an .MRK is present the RTK position is used, otherwise the photos' EXIF position.
Avoid: moving water surfaces, uniform snow/sand, very dark shadows, mixing different lighting conditions in a single run.

Position accuracy: RTK and GCPs

A ground control point (GCP) file is optional. If you provide one, the product is fitted to those points; without it, positions come from the RTK logs or from the photos' own position data.

File format: .txt or .csv; one point per line: name X Y Z. Values may be separated by spaces, tabs, commas or semicolons; the decimal separator is a dot.
No file? In Studio you can type the coordinates in with Enter and mark points.
Lines named sbt, baz, base, ref, cors and similar are treated as base stations; they are listed but do not enter the solution.
Marking: the photos are scanned, targets are found automatically and the frames where each point appears are brought to you; you check the position of each mark and correct it if needed. Marking each point in at least 3 photos, with at least 3 points in total, is recommended. Once a file is given, production does not start until the points are marked; if you prefer, Produce without GCPs drops the file and production continues on the RTK positions.
For a job sent from the web, the photos are uploaded first and marking happens on the same page once the upload has finished.

Every marked point enters the model. The residuals in the accuracy report are the in-solution differences of these points, not an independent accuracy measurement. The accuracy of the result depends on how well the points were surveyed and marked.

Delivery coordinate system

The photogrammetry engine's raw output is in UTM. The coordinate system is chosen under Advanced and the products are delivered in that system. Left on Automatic, the system is detected from the GCP file; without GCPs the products are delivered in UTM.

Where should it run: this computer or the cloud?

On this computer (Studio) — no tokens are spent and the raw photos never leave your machine. It needs Windows, an NVIDIA graphics card (at least 8 GB of memory), at least 16 GB of RAM and engines downloaded once; a single job handles at most 800 photos and 20,000 megapixels. Studio makes the call: if the computer is not up to it, it shows why and routes the job to the cloud. Today the dense point cloud step uses GPU acceleration best on the RTX 40 series (on other cards this step runs on the CPU; beta).
In the cloud (Produce in the cloud in Studio, or CartaX Studio page → New production on the web) — the photos are uploaded, production runs on CartaX's cloud machines and the result lands in your project. Billed in tokens: the cost follows the total megapixels processed; as soon as the photo count is known you see the estimated tokens and your balance, and you are warned before the job starts if tokens are short. Up to 2,000 photos in a single job. Once the photos are uploaded you can shut your computer down; once production has started it cannot be stopped.

Time expectations: in the cloud usually 2–4 hours, up to 8 hours depending on load. Locally the time depends on the hardware.

Where does the result go?

In CartaX: when the job finishes, the outputs are uploaded to the project you chose. The orthophoto becomes a map layer; Measurement elevation: Surface / DSM / DTM appears in the layer settings; the point cloud opens in its own viewer; Classes appears on the orthophoto row.
On your computer (local production): product folders under Documents / CartaX Studio — Ortofoto (orthophoto), SYM (DSM/DTM), Nokta Bulutu (point cloud), Sınıf Katmanları (class layers), Rapor (report). Rapor contains the engine report (report.pdf), the accuracy summary (accuracy.json), camera positions (shots.geojson) and the run logs.

Common questions

The orthophoto edges are blurred or warped. Overlap drops at the edges; fly a strip beyond the area of interest.

GCP targets were not found automatically. Targets need to be clearly visible in the photos; mark the missing points by hand on the same screen. On sets without RTK logs the frame estimate can be off by metres; correct the mark's position in the image.

There is a constant offset between the orthophoto and the cadastre/project. If positions come only from EXIF, a metre-level offset is normal; RTK or GCPs close it. Make sure the delivery coordinate system matches your project.

It takes very long. Photogrammetry is the heaviest step. If your graphics card is not RTX 40 series, the dense point cloud step runs on the CPU; you can also produce the job in the cloud.

Orthophoto from Drone Photos: A Photogrammetry Guide

2026-09-14 · 6 min read

What is produced?

Every job produces the same package at the same high quality; there is no output or quality selection:

Product	Format	What it is for
Orthophoto	GeoTIFF (COG)	A scaled, georeferenced aerial image; the base of the map.
Elevation data (DSM + DTM)	GeoTIFF (COG)	Surface and terrain elevation models; elevation readout, slope and volume on the map are calculated from these.
Classified point cloud	LAZ → Octree	A dense 3D point cloud; measurement, sections; with classes such as ground, vegetation and buildings.
Segmentation layers	Vector	Objects recognised on the orthophoto (buildings, roads, trees, water, vehicles…); open in the Classes window.

A textured 3D model (mesh) is not part of the production package today.

Preparing the photos

Result quality comes more from the photos than from the engine:

Overlap: at least 75% forward, at least 65% side. Low overlap = holes in the model and drifting edges.
Photo count: hundreds of photos is normal; both Studio and the CartaX Studio page on the web scan the root folder including its subfolders (JPG/PNG/TIF) and find the RTK .MRK logs by themselves.
Position data: photos should carry EXIF GPS. On RTK drones (e.g. DJI .MRK) centimetre-level positions are used automatically: when an .MRK is present the RTK position is used, otherwise the photos' EXIF position.
Avoid: moving water surfaces, uniform snow/sand, very dark shadows, mixing different lighting conditions in a single run.

Position accuracy: RTK and GCPs

A ground control point (GCP) file is optional. If you provide one, the product is fitted to those points; without it, positions come from the RTK logs or from the photos' own position data.

File format: .txt or .csv; one point per line: name X Y Z. Values may be separated by spaces, tabs, commas or semicolons; the decimal separator is a dot.
No file? In Studio you can type the coordinates in with Enter and mark points.
Lines named sbt, baz, base, ref, cors and similar are treated as base stations; they are listed but do not enter the solution.
Marking: the photos are scanned, targets are found automatically and the frames where each point appears are brought to you; you check the position of each mark and correct it if needed. Marking each point in at least 3 photos, with at least 3 points in total, is recommended. Once a file is given, production does not start until the points are marked; if you prefer, Produce without GCPs drops the file and production continues on the RTK positions.
For a job sent from the web, the photos are uploaded first and marking happens on the same page once the upload has finished.

Every marked point enters the model. The residuals in the accuracy report are the in-solution differences of these points, not an independent accuracy measurement. The accuracy of the result depends on how well the points were surveyed and marked.

Delivery coordinate system

Where should it run: this computer or the cloud?

On this computer (Studio) — no tokens are spent and the raw photos never leave your machine. It needs Windows, an NVIDIA graphics card (at least 8 GB of memory), at least 16 GB of RAM and engines downloaded once; a single job handles at most 800 photos and 20,000 megapixels. Studio makes the call: if the computer is not up to it, it shows why and routes the job to the cloud. Today the dense point cloud step uses GPU acceleration best on the RTX 40 series (on other cards this step runs on the CPU; beta).
In the cloud (Produce in the cloud in Studio, or CartaX Studio page → New production on the web) — the photos are uploaded, production runs on CartaX's cloud machines and the result lands in your project. Billed in tokens: the cost follows the total megapixels processed; as soon as the photo count is known you see the estimated tokens and your balance, and you are warned before the job starts if tokens are short. Up to 2,000 photos in a single job. Once the photos are uploaded you can shut your computer down; once production has started it cannot be stopped.

Time expectations: in the cloud usually 2–4 hours, up to 8 hours depending on load. Locally the time depends on the hardware.

Where does the result go?

In CartaX: when the job finishes, the outputs are uploaded to the project you chose. The orthophoto becomes a map layer; Measurement elevation: Surface / DSM / DTM appears in the layer settings; the point cloud opens in its own viewer; Classes appears on the orthophoto row.
On your computer (local production): product folders under Documents / CartaX Studio — Ortofoto (orthophoto), SYM (DSM/DTM), Nokta Bulutu (point cloud), Sınıf Katmanları (class layers), Rapor (report). Rapor contains the engine report (report.pdf), the accuracy summary (accuracy.json), camera positions (shots.geojson) and the run logs.

Common questions

The orthophoto edges are blurred or warped. Overlap drops at the edges; fly a strip beyond the area of interest.

It takes very long. Photogrammetry is the heaviest step. If your graphics card is not RTX 40 series, the dense point cloud step runs on the CPU; you can also produce the job in the cloud.

What is produced?

Preparing the photos

Position accuracy: RTK and GCPs

Delivery coordinate system

Where should it run: this computer or the cloud?

Where does the result go?

Common questions

Related guides

What is produced?

Preparing the photos

Position accuracy: RTK and GCPs

Delivery coordinate system

Where should it run: this computer or the cloud?

Where does the result go?

Common questions

Related guides