Camera matching

Camera matching recreates, in a 3D scene, the position, height, viewing direction and focal length of the camera that took a real photograph. The goal is for the rendered model to sit in the photo with exactly the right perspective. It is the foundation of every photomontage and of any image that shows a design in its real surroundings.

How camera matching is done

The most reliable method is to measure the camera’s position and height when the photo is taken, and to record a few points visible in the image whose coordinates are known, such as a kerb corner, the corner of an existing building or a survey marker. Those points are marked in the model and the camera is adjusted until they line up with the photograph. Many modeling and render programs also have tools that estimate the camera from vanishing points found along parallel lines in the photo. The focal length stored in the image metadata makes the job easier.

Practical points

Lens distortion, especially in wide-angle shots, bends straight lines and makes matching harder, so the photo is usually corrected first. Even with a perfect camera match, the result will not convince if the sun direction disagrees with the photo; noting the date and time of the shot lets you set the sun system accordingly. A small angular error can shift a distant building by several meters. For images that blend into real settings, see our architectural visualization services.