Combining Solves

Sometimes, you might have multiple footage sources of the same scene or content available. For example, you might have footage from witness cameras, or someone has taken detailed still photos of the scene. CameraTracker provides a way to solve each of your sources and then register them all in the same world. This allows you to leverage a high-quality camera track from a secondary source that is good to solve from, and use it in another source that's difficult to solve. You can also use this technique to tie detailed close-up stills to wide-angle shots.

The idea is that you pick one of your sources as the 'reference' solve. This should be footage that has good parallax, for example a set of wide-angle stills taken on set. You also need some geometry created from the solve data. The geometry is to establish 3D survey points, which are then used to tie together the reference and the satellite camera solves.

The workflow is something like this:

1.   Pick one of your sources as the reference solve and use CameraTracker to track and solve the camera.
2.   Create a linked camera using CameraTracker's Export menu. See Creating Camera Nodes.
3.   Use the camera data to create some reference geometry. Either:

• connect the Camera straight into the ModelBuilder node to create the geometry, or

• export the camera data, create the geometry in an external application, then import the geometry back into Nuke.

See Using ModelBuilder or Exporting Geometry, Cameras, Lights, Axes, or Point Clouds .

4.   Set some 3D survey points on the geometry in the reference solve. See Assigning 3D Survey Points.
5.   Create a separate CameraTracker for each of your satellite cameras, and track the footage as you see fit.
6.   Create User Tracks in the satellite cameras, corresponding to the 3D Survey points in the reference solve, making sure to also mark them as survey points. See Assigning 3D Survey Points.
7.   Solve the satellite cameras to align them consistently with the reference solve.

Tip:  You can copy the X, Y, Z position from a track in one CameraTracker node to another by right-clicking on a track and selecting copy > translate.