The Analyze Motion behavior analyzes and stores tracking data from a video clip, which you can then apply to objects in your project using other tracking behaviors. Use Object mode to recognize and track common subjects such as people or faces, pets, and cars. Use Point mode to track a pixel pattern.
Note: Tracking in one pass rarely yields perfect results. Successful tracking often requires trying different tracking modes or analysis methods. For information on tracking methods and tips, see Tracking guidelines.
Set the Analyze Motion behavior to Object mode to track recognizable subjects in a clip such as faces and people, animals, and cars.
In the Layers list in Motion, select the video clip to track, then click Behaviors in the toolbar and choose Motion Tracking > Analyze Motion.
An object tracker is added to the center of the canvas.
Track an area of the clip: In the canvas, drag the object tracker to the reference pattern you want to track.
Track a suggested subject in the clip: While pressing and holding the Option key, drag the object tracker over elements in the canvas. As you drag, bounding boxes appear over elements, suggesting regions of interest that may be trackable.
Additionally, elliptical shapes appear over faces.
When you release the mouse button over the element you want to track, the object tracker conforms to the subject.
Track a face in the clip: In the Behaviors Inspector, click the Detect pop-up menu, then choose Faces. Onscreen controls appear around any faces detected in the frame. When you click the onscreen control for the face you want to track, the object tracker conforms to the subject.
Note: A subject identified in a clip isn’t necessarily trackable; it’s a potential region of interest in the scene that the machine learning model recognizes.
In the Behaviors Inspector, click the Analysis Method pop-up menu, then choose one of the following options:
Automatic: Automatically chooses the most suitable analysis method. Because the best method is highly dependent on the properties of a clip and each use case, you may need to experiment with different analysis methods to achieve the best result.
Combined: Uses a combination of the Machine Learning and Point Cloud analysis methods (described below) to track position, scale, and rotation.
Machine Learning: Uses a machine learning model trained on a dataset to identify people, animals, and many other common objects, allowing the tracker to follow the subject in a specified region of video. Choose this option when absolute tracking precision isn’t required, such as when attaching titles or graphics to objects or people. This method can overcome moderate occlusion—when an object (such as a tree or car) briefly obscures the subject being tracked.
Point Cloud: Tracks a specific reference pattern and identifies how the pattern transforms from one frame to the next. Choose this option when you need more precise tracking of specific pixels. This method tracks position, scale, and rotation, and excels at tracking regions that are rigid and somewhat flat (from the camera’s point of view).
In the Behaviors Inspector, do any of the following:
Analyze the entire clip: Click Analyze. The clip is analyzed forward from the playhead position to the end of the clip (or to the frame where the reference pattern can no longer be tracked), and then backward from the playhead position to the start of the clip.
Analyze the portion of the clip before the playhead position: Click the left arrows next to the Analyze button.
Analyze the portion of the clip after the playhead position: Click the right arrows next to the Analyze button.
The tracking analysis progress window displays the analysis method used for the track.
Tip: If you’re using the Machine Learning analysis method and observe jitter during the analysis (the onscreen object tracker bouncing or jumping from one size to another), try switching to the Point Cloud analysis method. Its tracker is much less susceptible to quick changes.
To stop the tracking analysis, click the Stop button in the progress window or press Esc.
Tracking keyframes appear in the Keyframe Editor. If the Keyframe Editor is not visible, click the Show/Hide Keyframe Editor button in the lower-left corner of the Motion window.
In addition to the tracking keyframes, a confidence curve is displayed in the Keyframe Editor. This curve provides a visual indication of the tracker’s accuracy relative to its parameter settings in the Inspector. The confidence curve is for reference only and is not used for editing purposes.
The analyzed track (contained in the Analyze Motion behavior) can now be used to match move an object using other tracking behaviors. See Match move an object or Load existing tracking data.
Set the Analyze Motion behavior to Point mode to track specific points or a pattern in a clip.
In the Layers list in Motion, select the video clip to track, then click Behaviors in the toolbar and choose Motion Tracking > Analyze Motion.
By default, an object tracker is added to the center of the canvas.
In the Behaviors Inspector, click the Mode pop-up menu, then choose Point.
The onscreen tracker changes to a point tracker.
Play your clip several times to determine a good reference point, then move the playhead to the frame where you want to start the track analysis.
In the canvas, drag the point tracker to the reference point you want to use.
Note: For more information about adjusting the onscreen point tracker, see Adjust onscreen trackers.
Optional: To add more point trackers, click the Add button in the Behaviors Inspector, then drag them to reference points in the canvas.
To remove a point tracker, click Remove in the track row in the Behaviors Inspector.
In the Behaviors Inspector (or HUD), do one of the following:
Analyze the clip forward from the position of the playhead: Click Analyze.
Analyze the portion of the clip before the playhead position: Click the left arrows next to the Analyze button.
Analyze the portion of the clip after the playhead position: Click the right arrows next to the Analyze button.
When the tracking analysis begins, a progress window opens and displays the analysis method used for the track. Additionally, points appear on a motion path in the canvas. The point at the current playhead position is emphasized.
To stop a track analysis, click the Stop button in the progress window or press Esc.
Tip: If the tracker loses its reference pattern, don’t click the Stop button; allow the analysis to continue for a few seconds. When the tracker fails, the playhead jumps to the frame where the reference pattern was lost.
The points in the canvas correspond to the tracking keyframes that appear in the Keyframe Editor. If the Keyframe Editor is not visible, click the Show/Hide Keyframe Editor button in the lower-left corner of the Motion window.
A confidence curve is also displayed in the Keyframe Editor. This curve provides a visual indication of the tracker’s accuracy relative to its parameter settings in the Inspector. The confidence curve is for reference only and is not used for editing purposes.
The recorded tracking data (contained in the Analyze Motion behavior) can now be used to match move objects using other tracking behaviors. See Match move an object or Load existing tracking data.
