Some stills sequences are inevitably going to cause problems. There are a number of pre-tracking checks and post-tracking refinement controls to assist CameraTracker.
You can improve tracking data by adding User Tracks. See Working with User Tracks for more information.
• For standard sequences, avoid too much or too little redundancy between still frames, that is, frames that contain large portions of the previous frame. As a guide, try to capture every part of the scene in 3 to 4 images, aiming for a maximum change of 20-25% in content between frames.
• Try to capture the same object in at least 3, but preferably 4 photos.
For example, when moving around an object, a photo every 15-25 degrees would be adequate (approximately 16 shots for a full 360).
You can refine a set of feature tracks using the following controls and then retrack to improve your chances of getting a good solve:
• Minimum Length - sets a threshold value for the minimum acceptable track length. Tracks that fail to stick to the associated feature for this amount of frames are rejected.
In long slow camera moves, it is best to solve from long tracks generated in the shot. If there are lots of short tracks, this can lead to noise on the calculated camera. Try removing the short tracks before solving.
• Track Threshold - CameraTracker’s tolerance to change along the track is determined by this control.
Reducing this threshold makes tracking more tolerant to image changes, potentially producing longer tracks.
TIP: Press F in the curve display to fit the selected track(s) to the available screen space.
After refining your feature points, you may not need to retrack the entire sequence. You can use Update Track to analyze a specific frame range.
| 1. | Refine your features using the controls described above. |
| 2. | Click Update Track. |
A dialog displays allowing you to set the frame range to update.
| 3. | Set the required frame range and click OK. |
CameraTracker retracks the selected range and combines the tracking data with any existing tracks.
NOTE: You may notice that CameraTracker re-detects features on existing frames before re-tracking. This can happen when the cached feature data has been purged and needs to be rebuilt.