Loading ...
Sorry, an error occurred while loading the content.

Re: The Pupil Goes Back To School ...

Expand Messages
  • John Houghton
    ... Stuart, A poorly positioned camera on the pano head may well result in stitching problems, but stitching problems do not necessarily mean you have a poorly
    Message 1 of 8 , Mar 31, 2007
    • 0 Attachment
      --- In PanoToolsNG@yahoogroups.com, stuart mcalister
      <strtmcalister@...> wrote:
      > It then adds that theoretically, 'it is impossible to
      > get a perfectly stitched panorama from a fisheye
      > lens.' The proof of this is that PTGui comes up with
      > the 'there may well be misalignments etc'

      Stuart, A poorly positioned camera on the pano head may well result in
      stitching problems, but stitching problems do not necessarily mean you
      have a poorly positioned camera! To check the camera and head setup,
      you need to test specifically for parallax errors. Once you have
      confirmation that parallax has been eliminated, you can address any
      stitching errors that you may then get in PTGui and look for other
      causes. See:

      http://homepage.ntlworld.com/j.houghton/epcalib.htm

      John
    • stuart mcalister
      Dear John, Many thanks for getting back to me on this subject. I seem to have sorted out the problem but will certainly give your theory a good work-out. The
      Message 2 of 8 , Mar 31, 2007
      • 0 Attachment
        Dear John,

        Many thanks for getting back to me on this subject. I
        seem to have sorted out the problem but will certainly
        give your theory a good work-out.

        The one remaining problem I have is that in order to
        level the 303 on the tripod, I have a heavy Gitzo ball
        joint. To keep an eye on the entire level, I have a
        small 2-way spirit level which I place on the side
        mounting plate - I watch this carefully but although
        the 303 is level, as it passes through 180 degrees,
        the entire kaboodle goes very slightly off tilt. I'm
        sure it's just a case of leveling for every shot or
        the Gitzo ball has a problem. I would have thought
        that if the Gitzo was level, as was the 303, then any
        spin would result in a level spin ...

        It seems to show up more when using the 8mm in
        confined spaces as there's more 'give' in a wider
        photographic area. Is there a chapter anywhere on
        'confined spaces'?

        Many thanks again for your time and suggestion ...

        Regards,

        Stuart


        --- John Houghton <j.houghton@...> wrote:

        > --- In PanoToolsNG@yahoogroups.com, stuart mcalister
        >
        > <strtmcalister@...> wrote:
        > > It then adds that theoretically, 'it is impossible
        > to
        > > get a perfectly stitched panorama from a fisheye
        > > lens.' The proof of this is that PTGui comes up
        > with
        > > the 'there may well be misalignments etc'
        >
        > Stuart, A poorly positioned camera on the pano head
        > may well result in
        > stitching problems, but stitching problems do not
        > necessarily mean you
        > have a poorly positioned camera! To check the
        > camera and head setup,
        > you need to test specifically for parallax errors.
        > Once you have
        > confirmation that parallax has been eliminated, you
        > can address any
        > stitching errors that you may then get in PTGui and
        > look for other
        > causes. See:
        >
        > http://homepage.ntlworld.com/j.houghton/epcalib.htm
        >
        > John
        >
        >




        ___________________________________________________________
        The all-new Yahoo! Mail goes wherever you go - free your email address from your Internet provider. http://uk.docs.yahoo.com/nowyoucan.html
      • John Houghton
        ... Stuart, If you have set up the head so that the entrance pupil is aligned with the axis of rotation of the head to eliminate parallax, then you should have
        Message 3 of 8 , Mar 31, 2007
        • 0 Attachment
          --- In PanoToolsNG@yahoogroups.com, stuart mcalister
          <strtmcalister@...> wrote:
          >
          > I'm sure it's just a case of leveling for every shot or
          > the Gitzo ball has a problem. I would have thought
          > that if the Gitzo was level, as was the 303, then any
          > spin would result in a level spin ...

          Stuart, If you have set up the head so that the entrance pupil is
          aligned with the axis of rotation of the head to eliminate parallax,
          then you should have no problems stitching a panorama shot in a
          confined area. The levelling of the head will have no bearing on the
          stitching quality. If the axis of rotation is not precisely vertical,
          it is usually a very simple task to accurately level the panorama using
          t1 points on vertical edges in PTGui: no more than two minutes work at
          most.

          John
        • Serge Maandag (yahoo)
          ... It s true that with a fisheye the NPP depends of the angle at which light enters the lens surface. I wrote a page about that:
          Message 4 of 8 , Mar 31, 2007
          • 0 Attachment
            > It then adds that theoretically, 'it is impossible to
            > get a perfectly stitched panorama from a fisheye
            > lens.' The proof of this is that PTGui comes up with
            > the 'there may well be misalignments etc' ... when
            > using a 14mm, it is more or less a perfect match.

            It's true that with a fisheye the NPP depends of the angle at which light
            enters the lens surface. I wrote a page about that:
            http://pano.maandag.com/index.php?id=57

            But it shouldn't give you unstitchable results if you are a few mm off the
            NPP.
            Be sure to optimize horizontal and vertical shift though. That usually
            makes a lot of difference.

            And as for the moving NPP: it moves forward and backward. So if you shoot
            in a confined space, it might help to optimize individual FoV for each
            image. Be sure to have lots of control points, though.

            Serge.
          • stuart mcalister
            Serge, Many thanks ... I ll give your paper a read. Regards, Stuart ... ___________________________________________________________ The all-new Yahoo! Mail
            Message 5 of 8 , Mar 31, 2007
            • 0 Attachment
              Serge,

              Many thanks ... I'll give your paper a read.

              Regards,

              Stuart
              --- "Serge Maandag (yahoo)" <yahoo@...> wrote:

              > > It then adds that theoretically, 'it is impossible
              > to
              > > get a perfectly stitched panorama from a fisheye
              > > lens.' The proof of this is that PTGui comes up
              > with
              > > the 'there may well be misalignments etc' ... when
              > > using a 14mm, it is more or less a perfect match.
              >
              > It's true that with a fisheye the NPP depends of the
              > angle at which light
              > enters the lens surface. I wrote a page about that:
              > http://pano.maandag.com/index.php?id=57
              >
              > But it shouldn't give you unstitchable results if
              > you are a few mm off the
              > NPP.
              > Be sure to optimize horizontal and vertical shift
              > though. That usually
              > makes a lot of difference.
              >
              > And as for the moving NPP: it moves forward and
              > backward. So if you shoot
              > in a confined space, it might help to optimize
              > individual FoV for each
              > image. Be sure to have lots of control points,
              > though.
              >
              > Serge.
              >




              ___________________________________________________________
              The all-new Yahoo! Mail goes wherever you go - free your email address from your Internet provider. http://uk.docs.yahoo.com/nowyoucan.html
            • Erik Krause
              ... Usually (as a beginner) one pays far too much attention to the NPP and less or none to other causes of stitching errors. The NPP plays a role *only* if
              Message 6 of 8 , Mar 31, 2007
              • 0 Attachment
                On Saturday, March 31, 2007 at 13:41, Serge Maandag (yahoo) wrote:

                > It's true that with a fisheye the NPP depends of the angle at which light
                > enters the lens surface. I wrote a page about that:
                > http://pano.maandag.com/index.php?id=57
                >
                > But it shouldn't give you unstitchable results if you are a few mm off the
                > NPP. Be sure to optimize horizontal and vertical shift though. That usually
                > makes a lot of difference.

                Usually (as a beginner) one pays far too much attention to the NPP
                and less or none to other causes of stitching errors. The NPP plays a
                role *only* if there are subjects at different distances in the
                overlap region, and they probably are evident only if the distance
                between near and far subjects is relatively large.

                The usual causes of stitching errors are:
                - not allowed to optimze all necessary parameters for a task.
                - wrongly or not calibrated lens
                - too few control points for the amount of parameters to optimize
                (see: http://wiki.panotools.org/Perspective_correction )
                - sideways or up-down displacement of the rotation axis.


                best regards

                --
                Copyright (c) 2007 Erik Krause
                Verbatim copying and distribution strictly forbidden
                except those allowed in wiki.panotools.org/User_Guidelines
              Your message has been successfully submitted and would be delivered to recipients shortly.