On 29 November 2012 02:32, Greg 'groggy' Lehey <[email protected]> wrote:
>
> Where do you find that? I've tried this in the "Camera and Lens" tab
> with a photo taken vertically with the 9 mm lens. If I select
> "Rectilinear" it tells me 71.5° vertically, which presumably ignores
> the fact that it's mounted vertically, and doesn't say anything about
> HFOV. That tallies relatively well with my program above. But when I
> select "circular fisheye" it comes up with (only) 92.8°. That's some
> way from my estimate of 108°. Is there some other place you can get
> similar results?
As far as I remember hugin (or panotools) only refer to HFOV
throughout, which depends on the orientation of the image data (so
lens corrections values will also depend on image data orientation). I
always input my images in portrait and fix any rotation by assigning
roll values. Nevertheless you can play with the crop and focal length
values and the projection and hugin will calculate an an "estimate"
for the hfov, that's how i came up with the cited figures.
Diagonal fov on the other hand will be influenced by a,b,c lens parameters.
> I also get the same results whether I select "full frame fisheye" or
> "circular fisheye". What's the difference?
Basically only the kind of crop that can be applied to the image,
circular vs. rectangular. The very same lens can be a full frame
fisheye or a circular fisheye depending on the sensor size of your
camera.
> In any case, I can't see any mathematical correspondence between the
> focal length and the angle of view for fisheyes. Can anybody point me
> at some background information?
There is no single mathematical correspondence. It all depends on the
form of the projection function, of which the focal length typically
is the derivative at the lens center. The Samyang is special in that
regard as it features a ("stereographic") fisheye projection that is
different than that of most (all ?) other fisheyes.
>> Back to the original question:
>> For me the step from a wide angle rectilinear to a fisheye was very
>> well worth doing. The noticable but bearable quality difference is
>> outweighted by the much easier stitching with the lower number of
>> shots, especially so for 360x180s.
>
> That's interesting. I've asked on the German Olympus forum and got a
> reply from Reinhard Wagner (the author mentioned above) that confirmed
> my suspicion that the distortion would be less, since a 360x180°
> panorama is a form of fisheye image anyway, and the image needs to be
> distorted less. If you're interested, the thread is at
> http://oly-e.de/forum/e.e-system/135036.htm
Having read that thread, I think both views (fisheyes do not need that
much distortion or fisheyes need a lot more distortion) aren't exactly
right.
Typically the final format for a spherical pano is an equirectangular
file. And whether your input images are fisheye or rectilinear doesn't
matter that much, as there is always going to be a lot of mapping (or
distortion if you want) going on, especially near the poles. The main
quality loss here is caused by interpolation, but interpolation
routines are good enough (if don't choose bilinear or similar) that
you will not be able to tell a difference by eye even if you do
interpolation back and forth a few times.
Another issue is the compression at the extreme edges of the fisheye.
But this can easily be dealt with by sufficient overlap and crop.
Felix
--
You received this message because you are subscribed to the Google Groups
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx
