On Fri, 13 Feb 2004 07:00:55 +0000 Qkano <wildimages@xxxxxxxxxxx> wrote: > > Quite amazing! I guess they are using two blended sets of off-set > panoramas? > > AZ > > That's what I still don't understand. > When you take a normal panorama presumably the "lens" rotates about > the entrance pupil (?). Supposed to be the front nodal point, I believe. > If you had two cameras side by side where would the axis of rotation > be? > > Mmm. Maybe it's OK for a "scanning" type of shot but I could see > problems for assembling these from a sequence of static frames. Well, I've seen agonizing over finding the axis of rotation when making a panorama, but it isn't very convincing. Of course if you rotate the camera about the tripod axis (which is what happens by default) there is a parallax effect. But calculate its magnitude: the worst case will be that a close object moves with respect to something at infinity (like a cloud - fat chance of them being stationary!) In practice the parallax error will be a very small number of centimetres. So if the closes object is around 5m away, this will be appreciable. But provided this is a "landscape" view, a resolution of 2cm *at the subject* is likely to be perfectly adequate. So when you rotate your pair of cameras, do it about the mid-point between them. Getting the transverse movement of the camera between two shots to be not more than a cm or 2 just isn't going to be difficult. So I think it's a non-issue. Brian Chandler ---------------- Jigsaw puzzles from Japan http://imaginatorium.org/shop/ imaginatorium@xxxxxxxxxxxxx
