- Posts: 98
- Joined: Sat Sep 21, 2013 10:52 pm
I'll answer in general case here, others can provide details about Fusion.
Planar tracker in itself is based on an assumption that the underlying part of image is planar, that is, describable using plane equation. This assumption is then used to solve the plane translation based on simple feature tracks that are generated automatically inside tracked area. This translation can then be expressed as a corner pin, which is the representation of rectangular plane in 2D view. Cornerpin is not just four points, because it also compensates for perspective, otherwise using it for planes that are not perpendicular to view would give unnatural result.
This 2D transform that planar tracker outputs can also be used for moving roto shapes. In this case roto shape in its reference frame is "projected" to planar surface (in Mocha you set the reference plane manually, it is that blue rectangle thingy) and on other frames, that transform is applied to rotoshape to move it accordingly, perspective change and all. Now, if you animate the roto shape on top of that transform, you get the best of both worlds. An automatic track and manual control.
Mocha usually exports roto shapes in baked form, that is, the underlying planar tracker data is baked into rotoshape in screen space. If you want to use the tracking data and do manual roto outside Mocha, export a cornerpin and use it to drive a roto shape.