From the collection of 2D tracks, MatchMover estimates all camera parameter including internal parameters such as focal length and non-linear distortion, camera position and orientation over time, and 3D point coordinates.
Depending on your project, you can provide MatchMover with some specific information about the shot, thus constraining the process by reducing its parameter space.
A keyframe is a frame containing enough parameter data for the camera solving process. Using the data obtained from the 2D tracking process, MatchMover initializes the cameras and creates the 3D points for the sequence. The computation process starts on a solid keyframe pair, called the reference frames (1 and 2). MatchMover automatically selects these frames when the solver is launched or when using the Select keyframes command, if it is not locked by the user. Depending on the camera motion, overriding reference frames may help solve complex shots when the selected references are not well located.
Point relations are used by MatchMover to obtain information about the geometry of the scene. For example, this information can be that several points share a coordinate, for example, all points on a horizontal plane, such as the ground, share the same Y value. Or, that some points have known coordinates. This usually happens when you have measured survey points in your on-set. In this case, it may be simpler to use survey points. Providing MatchMover with information that helps the camera solving process improves the accuracy of the results.
If you know some of the properties of a scene, such as measurements, or you have some constraints, you may know the 3D coordinates of some points of the scene. Instead of letting MatchMover compute their 3D coordinates, you can set them before the computation.
You can either set these coordinates manually or use one of your 3D object vertex coordinates.
Camera constraints tell MatchMover that a parameter is constant over a subset of frames associated with a camera. As it reduces the number of computed parameters, it limits the calculation time and speeds up the process. MatchMover has four types of constraints. Focal length, Nodal Pan, Dolly, and Planar.
Some hardware devices, such as a Scorpio crane or Flair, output what is known as “Motion Control Data”. You can import some of this data into MatchMover using the motion control file parser. This information is directly input in the calibration engine, and used as an initial solution. It can therefore be refined, and additional camera parameters computed. See Importing motion control data.
You can also manually input or override specified aspects of this data to constrain the final solution to a valid, specified range before launching the solver.
