I have a cylindrically mapped panorama of about 5 shots and both Ptgiu and Autopano are giving me an overall FOV of 116 and 117 degrees respectively. However when I calculate FOV from the GPS camera bearings it comes out at 141 degrees and when I mark the limits of the panorama in autocad and measure the angle it gives 138 degrees.
I would have written it off as inaccurate software on Autopanos behalf but it's curious how both it and ptgui are both so consistently off, have I done a booboo or is software FOV calculation generally unreliable?
Calculating the fov accurately on a partial panorama is really complicated as a slight shift in the non parallax point will give 20% shift in the focal. Workflow to get a perfect evaluation of the field of view for a partial panorama: - with the same lens/camera/head setup, do a full row panorama covering 360° with a lot of overlapping. Detect it, and write down the calculated fov this lens/camera couple. - on the partial panorama, enter their the calculated fov from previous panorama and say to the optimizer 'do not optimize fov' ( v3.5 only ), so it will keep that focal. Then, you'll get a perfect estimation of the fov
Unfortunately the panorama is being supplied by the client (I'm a 3D artist) and he has not invested in a head that allows for parallax adjustment so I can be pretty certain the camera is pivoting on it's own mount and not around the lens's nodal point. Would this account for the FOV being consistently out on both software packages?
One other question about your suggested workflow, if I do a 360 degree pano, surely the FOV will be 360 degrees?
Pivoting a camera on its own mount like a ball head will not produce accurate stitching.. In fact almost impossible to stitch using a fisheye lens... Not so sure about a standard prime lens etc.. I thought about trying it once ...
Yes, I knew it would cause parallax errors but wasn't aware it would cause FOV errors too. Would you (or anyone else here) be strong on the maths behind FOV calculations? I'm trying meet some presentation standards for a photo-montage and it's not working, their numbers work by my calculation but the image isn't fitting the numbers so I'm trying to find the cause. I'd love to be able to run them by someone to get a second opinion.
the fov of the panorama is 360°, but the real value you need to calculate is the fov of an individual image ( which is a function of the focal lenght / lens distortion ). Knowing that, in autopano, we can calculate any fov of any partial panorama.