User Manual

Table Of Contents
Green tracker marks are added automatically to the features in an image.
Bi-Directional Tracking
When performing a track, you can enable the Bidirectional Tracking checkbox, which first tracks
forward from the start of the clip, and then tracks a second pass in reverse. This two-pass
approach can potentially extend the duration of any given point by re-analyzing points initially
identified in the forward pass. There is very little reason not to have this enabled unless you are
very short on time. Bidirectional tracking takes longer, but it’s usually worth it, and the process
is reasonably quick considering the benefit.
Choosing a Tracking Algorithm
There are three available choices for the algorithm to use when tracking. The three options in
the New Track Defaults section of the Inspector include:
Optical Flow: Usually your best choice, unless you have a great deal of criss-crossing
objects in a clip.
Tracker: A good second choice when Optical Flow can’t be used due to motion
estimation errors like criss-crossing objects.
Planar: Mostly used in simpler clips, where the majority of the image consists of planar
surfaces such as the facades of buildings.
Masking Out Objects
When tracking a clip, the Camera Tracker automatically generates trackers on feature details.
However, not all features that stand out in a clip are appropriate for camera tracking. You only
want to track features that are “nailed to the set.” In other words, objects that move
independently of the camera motion, like moving cars and people, cause inaccuracies when
camera tracking. You must eliminate these types of objects from the analysis.
The primary way of avoiding these problem areas is by masking. You connect a mask to the
Camera Tracker node’s Track Mask input to identify areas of a scene that the Camera Tracker
can analyze. For example, if you have a clip of an airport runway along a shoreline, the waves of
the water and moving clouds in the sky must be masked since they move independently of
the camera.
1594Chapter – 77 3D Camera Tracking