PanoTools:
Re: Would 10.5 pano join better if image is defished?
Roger D. Williams 2006-Jun-06 07:34:29
On Tue, 06 Jun 2006 11:10:49 +0900, smarfingerfeulcher
<#removed#> wrote:
> I've been more or less following discussion on the true nodal point of
> the 10.5. As I currently understand this, the nodal point varies with
> the angle, thus pano alignment is best within a doughnut shape
> depending on where the camera sat on the pano head. All this because
> it's a fisheye.
>
> Graphs at michel.thoby.free.fr
>
> If I defish the image in Capture, will this facilitate better joins
> for the pano?
These are the precise issues I was pondering some time ago and which
were answered quickly and unambiguously here on the list. Definitely
the right place to ask. Let's see if I remember the answers...
The reason the nodal point/entrance-pupil/non-parallax point varies
with the angle is the different degree of "bending" that straight lines
not passing through the center of the image undergo. This means that
the point about which you will need to swing the lens will not be
the same if you swing it through 90 degrees between each shot (four
shots to make 360 degrees--fine if you don't want full zenith to nadir
coverage--as if you swing it through 60 degrees between each shot
(six shots to make 360 degrees--fine for leaving smallish "holes" at
zenith and nadir which you will probably fill with one each additional
shot for zenith and nadir). The former would use your camera in landscape
orientation, the latter in portrait. I am assuming that your 10.5mm
Nikon is functionally equivalent to my 16mm fisheye used with full-frame
35mm film.
It may not be easy to imagine WHY this should be, but a quick check of
any way you know of locating such a non-parallax point soon reveals
that this IS so. A line of objects that stands precisely one behind
the other when plus or minus 45 degrees to left or right of center will
require the camera to rotate about ONE point. When the line is 30 degrees
off axis, the center of rotation needs to be displaced. Try it! <g>
This does mean that once you determine the angular separation between
the shots you stitch, you CAN fix the non-parallax axis and there will
be no more variation, and stitching will go very smoothly with no more
trouble from parallax between images.
So much for that.
The quick answer to your second question is that you do NOT need to
defish images before stitching. The joins can be perfectly well made
with PTgui (it may also be true of other software, but this is what I
am most familiar with). The reason is that even perfectly rectilinear
images need to be "warped" to achieve perfect stitches, and it is no
more difficult for the software to warp to the necessary form from
an equiangular fisheye image than it is from a rectilinear image.
It is quite difficult to put these things clearly, at least I find it
so. I hope this hasn't created any confusion.
Roger W.
--
Work: www.adex-japan.com, Play: www.usefilm.com/member/roger
------------------------ Yahoo! Groups Sponsor --------------------~-->
You can search right from your browser? It's easy and it's free. See how.
http://us.click.yahoo.com/_7bhrC/NGxNAA/yQLSAA/.Cr1lB/TM
--------------------------------------------------------------------~->
<*> PanoTools.Info Wiki
http://wiki.panotools.org
<*> PanoTools.Info List Archives
http://lists.panotools.info/pipermail/ptml/
<*> PanoTools List Reader @ GMane:
http://news.gmane.org/gmane.comp.graphics.panotools
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/PanoTools/
<*> To unsubscribe from this group, send an email to:
#removed#
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/