[ptx] Re: My comments, questions and enhancements

[Pablo d'Angelo]

Hi,

> > >Hmm, what do you call 'correct'?
> > >
> > The correct focal length is determined by a factor that differs from 
> > camera to camera. As I understood hugin reads the (real) focal length 
> > from the EXIF data, converts it to a focal length for a 35 mm film and 
> > calculates the degrees of view. Is this right?
> 
> Yes, and this conversion is based on the size of the CCD chip as
> compared to the standard 35mm film.
> Next, the HFOV can be computed from focal length and chip size.

Actually I'm computing the HFOV directly from the EXIF data, without the additional
confusing step to the 35 mm equivalent focal length, which can be defined in
several ways.

> This is most probably done assuming an error-free optics.  Real optics are
> not error-free, and, e.g., cussion- or barrel distortion lead to a
> slightly larger/smaller FOV when compared to the 'ideal' one.

jep, simple pinhole camera model.

> Therefore I prefer to compute the actual FOV from fitting a 360 deg
> pano, as this is the important value for stitching.
> Ideally you do this for all lenses / focal length settings and store
> them in a library (together with the distortion parameters, of course).

A lens database would be nice to have.. but.. no time...

ciao
  Pablo
--
http://wurm.wohnheim.uni-ulm.de/~redman/
Please use PGP