I've had my Liteplacer up and running for a few weeks now. Since I'm doing final testing before I put a board into low volumne production I haven't been too worried about the placement accuracy I've had.
What I noticed was that some components would be placed perfectly fine, and some components would be quite off. The same component at the same location would always be placed the same (accounting for any inaccuracy caused by the part being in the tape slightly differently). I eventually realized it was related to rotation.
What was happening is that if a component does not need to be rotated, the nozzle to camera calibration doesn't matter very much. If you pick up a metal box 6 inches off center to the right with a magnet, move it two feet to the left, and put it down, the box is 2 feet from where you put it. However, if you pick up the metal box 6 inches off center to the right, rotate it 180 degrees, move it two feet to the left, and put it down, the box is only 18 inches to the left, instead of 24 inches.
The moral of the story is that the Nozzle to Camera calibration doesn't impact all part placements, only ones that are rotated.
That being said, when placing 0402 components you need to have the Nozzle to Camera calibration pretty spot on in order to pick the components up without undesirable behavior like the component jamming itself in the nozzle at an angle.