Is initial camera position and orientation information significant?

[wangli]

Is there any evidence that a better alignment result can be generated with initial camera position and orientation information than that without initial values?

[stihl]

As far as I knoknow, because Photoscan uses the camera coordinates when selecting reference for the alignment pre-selection parameters during the alignment, to know which cameras are located near eachother so Photoscan can more easily and thus faster detect tie-points in overlapping images.

This is unrelated to the accuracy of the alignment. I believe this only has to do with the photo quality (motion blur, sharpness, lighting conditions etc).