Written by 
A photograph captures a scene from a particular viewpoint. Photogrammetry goes further: by combining multiple images and understanding the geometry between them, it is possible to derive measurements and reconstruct three-dimensional information about the world.
This principle is not new. Photogrammetry has been used for generations in mapping and surveying. What has changed dramatically is the technology available to collect and process imagery.
Today, cameras mounted on drones, aircraft, vehicles and even handheld devices can produce detailed 3D models of environments ranging from individual buildings to entire landscapes.
What is photogrammetry?
ASPRS describes photogrammetry as the art, science and technology of obtaining reliable information about physical objects and the environment through recording, measuring and interpreting imagery and patterns of electromagnetic energy. At its core, photogrammetry uses geometry. If the same feature is visible in multiple images taken from different positions, the relationships between those observations can be used to determine its three-dimensional position. This is the same basic principle that allows human vision to perceive depth.
From overlapping images to 3D
Modern photogrammetric workflows often begin with a collection of overlapping photographs. Software identifies corresponding features between images and estimates the position and orientation of the cameras. In Structure-from-Motion (SfM) workflows, this can be achieved without requiring the camera positions to be known in advance. The result is a sparse reconstruction of the scene and estimated camera poses. Further processing can generate a dense point cloud, with millions of 3D points representing surfaces in the scene.
From there, users can create products such as:
- digital surface models;
- textured 3D meshes;
- orthomosaics;
- point clouds;
- cross-sections;
- measurements of distances, areas and volumes.
The workflow therefore transforms a set of ordinary photographs into a spatially referenced dataset.
Why drones changed the game
Uncrewed aerial vehicles have made photogrammetry accessible for a much wider range of applications. A relatively small drone can capture hundreds or thousands of overlapping images across a site in a short period. These photographs can then be processed into high-resolution 3D products. This has applications in surveying, construction, archaeology, agriculture, environmental monitoring, quarrying, forestry and disaster assessment. The advantages are particularly clear where terrain is difficult or dangerous to access. USGS research and data products demonstrate the breadth of modern SfM applications, including monitoring coastal change, landslides, lava emplacement and other geological processes.
Measuring change, not just creating models
One of the most interesting applications of photogrammetry is repeat surveying. Imagine surveying a cliff today and repeating the survey six months later. If the two point clouds are accurately georeferenced and aligned, differences between them can reveal erosion, rockfall or other changes. This converts photogrammetry from a mapping technology into a monitoring technology. The same principle can be applied to glaciers, landslides, river channels, coastlines, construction sites and archaeological sites. Recent USGS work, for example, has used SfM photogrammetry to generate topographic point clouds and analyse coastal change across multiple years.
Accuracy still matters
The ease of producing an attractive 3D model can sometimes hide the complexity of producing an accurate one. Photogrammetric reconstruction depends on image quality, overlap, camera calibration, scene texture, illumination and image geometry. Georeferencing is also critical. If the objective is simply visualisation, a model with arbitrary scale and position might be perfectly adequate. If the objective is measuring centimetre-scale deformation, however, much greater attention must be paid to control, positioning, camera calibration and error assessment. Ground control points and accurate positioning systems can help constrain the model. Independent check points can then be used to evaluate accuracy. This distinction between visual quality and measurement quality is important. A model can look impressive while still containing systematic errors.
Photogrammetry and remote sensing
Photogrammetry and remote sensing are often discussed separately, but in practice they increasingly overlap. A drone can carry a conventional RGB camera, a multispectral sensor or a thermal camera. Aircraft can carry multiple sensors simultaneously. Satellite imagery can be combined with elevation data and 3D information. Photogrammetric methods can also contribute to the interpretation of remotely sensed imagery. The wider geospatial ecosystem is therefore becoming increasingly multi-sensor.
The rise of computer vision
Modern photogrammetry has also been transformed by computer vision. Feature detection and image matching can be automated. Machine learning can assist with image interpretation, segmentation and classification. Large datasets can be processed more efficiently than was possible with traditional workflows. Recent research is beginning to combine SfM-derived point clouds with large vision models for semantic 3D mapping, allowing reconstructed geometry to be linked with meaningful geological or environmental features. This represents an important evolution.
Traditional photogrammetry asks: Where is this point?
Modern geospatial AI increasingly asks: What is this point, where is it, and how has it changed?
Why photogrammetry remains relevant
With LiDAR, satellite radar and increasingly sophisticated sensors available, it might be tempting to think that photogrammetry has become less important. The opposite may be true. Cameras are relatively inexpensive, flexible and capable of collecting extremely detailed information. They can be deployed from many different platforms, from smartphones and drones to aircraft. Photogrammetry also provides a powerful connection between imagery and geometry. The future is unlikely to be a competition between photogrammetry and other sensing technologies. Instead, the most useful systems will increasingly combine them.
A 3D future
Digital twins, infrastructure monitoring, autonomous systems, environmental modelling and heritage documentation all depend on accurate spatial information. Photogrammetry provides one route to acquiring that information. Its enduring strength is simple: photographs contain geometry. With the right acquisition strategy, processing methods and quality control, those photographs can become measurements, maps and three-dimensional representations of the real world. For a discipline increasingly concerned with understanding change in three dimensions, that remains remarkably powerful.