Solving Single Photo Geometry with External Data

Photogrammetry is the science of extracting geometric data—measurements and 3D models—from photographs. This reconstruction can be done with multiple photos of a scene or, under the right conditions, from a single photo. In a single-photo project, you can still obtain accurate measurements, positions, and even full 3D models, as long as certain conditions are met.

Single-photo projects work differently from multi-photo projects because you cannot rely on multiple viewpoints to solve the 3D geometry and camera parameters. Instead, PhotoModeler uses extra information you provide about the scene, called “constraint data,” to recover scale, orientation, and shape.

Constraint data tells PhotoModeler something you already know or assume—for example, 3D control points with known X,Y,Z coordinates from a survey, laser scan, or another PhotoModeler project; or geometric relationships such as parallel and perpendicular lines in the scene. In multi-photo projects, PhotoModeler can often solve the camera’s internal settings (like focal length and format size) and external pose (position and angles) automatically, using the redundancy of multiple overlapping views. With a single photo, these parameters become under‑determined, so constraints are what make reliable single-photo reconstruction possible.

Once the camera position and internal parameters are solved, you can extract the information you need. This is done by either defining points on a plane (such as the ground plane solved as part of the initial solution), or matching 3D models (e.g. of vehicles) to the photographic data to get position and orientation.

Single-Photo Projects in PhotoModeler

PhotoModeler provides specialized tools for the single-photo reconstruction, including workflows that use known XYZ control points and the Axes Constraint system to recover a usable 3D coordinate system from just one image. The axes-constraint system is based on the geometry of vanishing points. The resources linked below show how to set up these constraints in practice and how far you can go with single-photo modeling, measurement, and analysis in real projects.

Not every single photograph will work: you need sufficient resolution, clear geometry, and either known XYZ data or good perspective with axis‑aligned features. For the Axes Constraint method in particular, very wide‑angle photos with high lens distortion may not be suitable.

Also see the Surveillance Cam, Dash Cam, Body Cam Measurement with PhotoModeler page with links to video examples.

Typical Use Cases

The most significant application of working with single photographs, images, and videos is forensic reconstruction—extracting important information for a court case or investigation. Examples include determining the height of a person or object, locating key evidence on the ground, or determining the speed and path of a vehicle from a video or a set of frames.

Single‑photo techniques are also useful in areas such as crash analysis, site documentation, and measurements from historical or surveillance imagery when new photography is not possible.

Lastly, these methods can be used to extract accurate photo-textures, ortho-photos, and planar measurements where a surface can be defined in the single photograph.

Single images commonly originate from surveillance systems, bystander mobile phones, dash cameras, body‑worn cameras, and historical photo or video archives.

The Two PhotoModeler Workflows

There are two main workflows in PhotoModeler for working with single photographs: using known XYZ control points when external 3D data is available, and using Axes Constraints when you only have the photograph but the scene has strong, axis-aligned geometry.

Known XYZ control points: Known XYZ control points come from survey data, LiDAR point clouds (scenes or objects), or other PhotoModeler projects. This data is loaded using the Imports and Coordinates system. You mark the control points in the photograph and, when processed, the camera internal and external parameters are solved. With sufficient XYZ control points, lens distortion can also be solved, which improves accuracy for wide‑angle lenses. Here are some resources:

Axes Constraints: For photographs containing strong perspective and recognizable horizontal/vertical or axis-aligned features, Axes Constraints can recover camera orientation and metric 3D from a single image. This workflow is especially useful when known XYZ control points are not available but the scene has straight, orthogonal structures such as building edges, road markings, or vehicle edges.

Summary

Single-photo projects in PhotoModeler turn otherwise “one-off” images into measurable, 3D-aware reconstructions. By adding constraint data—either XYZ control points from survey, LiDAR, or other projects, or geometric information via Axes Constraints—you solve the camera, establish a usable 3D coordinate system, and extract the measurements, positions, and orientations you need from a single image.