You should not have to manually move the virtual cameras. If the calibration data is good, the software will automatically position the cameras with reasonable accuracy after calibration is complete.
When calibration is done, you need to set the scale of the environment by entering the actual height of one camera. For example, if Camera #1 is 1.8288 m up, you need enter this exact value for this camera. Assuming the calibration footage was shot correctly and successfully processed, the entire scene should now more or less reflect 'real world' scale. You might notice that the other cameras are not at exact heights but they should be pretty close.
You can't track successfully without good calibration data, so I would first make sure the calibration data is good before proceeding. Otherwise, you'd be wasting your time.
If you're interested, I've posted how I do my calibration video here:
http://www.youtube.com/user/LGDTestTube ... lG1_65s2_cThis pattern works every time for me.
For an example of an outdoor calibration session, I highly recommend viewing 1k0's Making of Danse Kabyle video:
http://vimeo.com/22036677Hope this helps.