< Previous | Contents | Manuals Home | Boris FX | Next >
Typically it is best to track the camera first, then the object(s), and maybe only then test for small amounts of lens distortion. With both tracked, you can then apply axis locks as needed.
There is some subtlety associated with the camera field of view (for non-360VR cameras), since the field of view affects both camera and object.
So here we're going to point out a common subtle mistake , and what to do about it.
If you solve the camera, disable the camera, then solve the object, you will create an subtle mismatch, because without further action, the object solve will change the field of view: typically you will have left the lens set to Fixed, Unknown (or maybe Zooming). SynthEyes will compute the object solve and field of view that produces the lowest error for the object. As the camera field of view is changed to optimize the object solve, the camera solve becomes incorrect. The stated error may be low, but will no longer match the situation exactly, because of the field of view change.
Instead, if you disable the camera, you should also set the lens to Known mode, and accept the question to copy the solved field of view to be the seed field of view.
Of course, this means that the field of view can no longer be changed to optimize the object solve, and it may be the case that the object solve has more and better trackers than the camera solve, such that you want to have the object solve impact the ultimate field of view.
To do that, instead you can change the camera to Refine, instead of Disabled, when beginning the object solve. Then the field of view will be chosen to optimize both.
An alternative is to Disable the camera, complete the object tracking, then set both camera and object to refine mode, to optimize them together. Note that for small differences, typically the field of view will stick at its current value.
©2024 Boris FX, Inc. — UNOFFICIAL — Converted from original PDF.