buggy HFOV calculation (hugin_2003_10_28-10_55_win32.zip)

Mike Runge mike at trozzreaxxion.net
Thu Oct 30 11:42:33 GMT 2003


Hello everybody,

today I struggled a bit working with the the lens settings.
My camera is a Fujifilm S602Z and PTGui extracts the following parameters out
of the exif data:

Crop factor/focal length multiplier: 4.52
Size of sensor: 7.65x5.74
That matches with the crop factor I calculated by myself (35mmKB ~ 7.8mm).

I usually use a focal length of 9.7mm what should be 43.5mm in KB speaking.
Therefore PTGui calculates a value of 43.0416 for HFOV using a landscape
image and HFOV of 32.9645 if the images are done in portrait mode.

Hugin reports nearly the same values, but the shown focal length (KB size) is
wrong what tooks me some time to verify if the values are correct or not.
Here's what hugin calculates for my pics with focal length of 9.7mm
(43.mmKB)

Landscape:
HFOV (Degrees) = 43.029
Focal length   = 40.588

Portrait:
HFOV (Degrees) = 32.941
Focal length   = 54.118

It's not disturbing me creating panos even the HFOV value is correct, but it
was quite irritating?!

Even more irritating is another bug, that hurts me while trying to keyin the
values by hand:

You need to fill in the focal length value you got directly from the camera
exif - not the KB-value. This is fine for digicam users because you don't
need to calculate the corresponding KB value.
To apply the value, you have to tab into another field and both fields
(degrees and Focal length) will be calculated. 
You better should not tab several times between Degrees and Focal length!!!
Everytime you tab into one of this fields the values will be calculated newly
and after several tabs (Degrees, than Focal length, than again Degrees, etc.
)you will end up with some stupid values.

Quite irritating. Somebody who never thougth about the corresponding Field of
View of the used Focal Length will probably not believe that the basic
calculation of hugin is correct.

Sorry,
I hope it was understandable enough to reproduce it.

best, mike 
    


More information about the ptX mailing list