Tracking

From RMIT Visual Effects
Jump to: navigation, search

By tracking a moving object the motion parameters of that object is acquired. Matching the motion from one source to another is one of the 'big magic' items of compositing. This motion information can be used in a number of ways:

  • Stabilize To remove the motion entirely.
  • Remove jitter To keep the global motion but to remove the local variations in motion.
  • Add jitter To pass on the local motion variations to another item.
  • Match-move To apply the same global motion to another item.

Tracker

This is the simplest of Nuke's trackers, but still surprisingly powerful. It can track an unlimited number of points, though commonly it is used to track either 1 (one point tracking) or 4. In the case of the later, the results are passed off onto the four points of a CornerPin. Commonly, the tracks last the entire duration of the clip. More information on the Tracker page.

PlanarTracker

The PlanarTracker tracks areas, unlike the Tracker and the CameraTracker, which both track points. The PlanarTracker requires first a region of the footage to be defined using a Roto node. hence, when you open the PlanarTracker, the first thing that you see is something that looks like a Roto node. Most of the important controls for this node are in the Viewer window.

The PlanarTracker is produces a result that sticks like glue to regions, perfect for non-organic flat planes moving in 3D space (e.g. buildings, walls etc). However, it can also be used on organic things, like faces, so long as the shot is not too complex.

We will not cover this tracker too much in this course. For those interested, here is a page of tutorials detailing the use of the PlanarTracker node.

CameraTracker

Though it is in the 3D menu, not the Transform menu, the CameraTracker node should also be considered as part of the tracker family.

Whist the Tracker commonly tracks just a few points, the CameraTracker can track hundreds. However each track can be as short as three frames. Once this data is gathered, it is translated into 3D information. This specifies: the camera and its optical properties, and the 3D attributes of the scene.

Effectively, the CameraTracker serves to recreate the same 3D world that the camera captured. Into this world can be added new elements, to be composited back into the captured footage. More info in the CameraTracker page.