Re: Panorama Tools

[Glenn Barry <glenn@xxxxxxxxxxx>]

Hi Helmut,

I found the error that has been plaguing me for days now, a small typo that
was incorrectly setting my image sizes in the optimizer script. I am now in
total awe of your work. I did manage to read everything in sight trying to
solve the problem, except re-reading the script carefully!!! I was becoming
despondent at the number of possibly incorrect variables and errors. Since
having shot the test frames I have found the rear nodal point of the lens, so
things should become simpler from this point forward.

>From the results that I have achieved, the angular coverage comes in at
183.007 degrees, and it has a usable  image circle of just over 43mm, which is
not spectacular. The advantage lies in the fact that I don't have a 35mm
scanner, I use a flatbed, and I have to do less scans.

I am going to do some tests with my 28mm Nikkor to use it for panoramas that
don't merit full spherical views, for memory it has an image circle in the
region of 52mm

Thanks for your tips

Glenn

Helmut Dersch wrote:

> Glenn Barry wrote:
> >
> > Hi All,
> >
> > I am having problems determining correction co-efficients (Panorama
> > tools correct) and Field of view values (Panorama Tools Remap) for a
> > 15mm F2.8 AF Sigma fisheye. When I run multiple optimizations my values
> > seem to fluctuate and the between actual and desired values between
> > correction points don't decrease.
> > Has anybody else had any success determining these values that they
> > could share?
> >
>
> Hi Glenn,
>
> I am using this lens on 35mm format and have an example on-line:
> <http://www.fh-furtwangen.de/~dersch/example.zip>
> It includes an optimizer script with correction values.
> However, this can not be directly applied to 120 film images
> since the width of the image (field-of-view) matters.
>
> Correction values usually scatter somewhat for different sets
> of images and/or feature points since not only the radial lens
> distortions but also other geometric errors (nodal point alignment,
> scanner errors, non-radial distortions etc) contribute.
>
> For your set-up, you need to crop the images to the circle,
> and then use one set with ample overlap (eg 3-4 for a 360 degree pano)
> with lots of feature points. Try to be as accurate as possible
> with the alignment and the estimates for yaw/pitch/roll. Then
> optimize position (yaw/pitch/roll), radial coefficients (a,b,c)
> and field of view (v). You probably don't need to optimize
> the position of the optical center (d,e) as in the example
> on my site, since it should be close to the center of the circle.
>
> > One thing to bear in mind is that I am shooting full circle fisheye with
> > this lens on 120.
> >
>
> Could you share some details on this setup. How large is the
> image circle, what is the angular coverage?
>
> Regards
>
> Helmut Dersch

--
Glenn Barry Photography
Studio 1, Level 1
2-14 Mountain St
Ultimo N.S.W. 2007
Australia
Ph (612) 9211 3080
Mobile 0415 279 366
E-Mail: mailto:glenn@xxxxxxxxxxx
Web: http://www.acay.com.au/~glenn