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

47269Re: [PanoToolsNG] Re: What is the actual Field of View?

Expand Messages
  • Michel Thoby
    Jan 5, 2011
      Hi Erik,

      Le 5 janv. 2011 � 14:46, Erik Krause a �crit :

      > "Physical vignetting should be avoided because it may cause the lens to
      > have only a collection of �least-parallax� points instead of a single no
      > parallax point."
      >
      > For some lenses if you use them wide open other parts of the lens
      > housing might "select the rays which form the image". In this case the
      > center of perspective moves to the (aparent) location of those parts.
      > Consequentially the field of view changes as well. This is independent
      > from object distance and hence applies to long-range applications same
      > as close-up.
      >
      > --
      > Erik Krause

      I believe that for most (if not all) wide-angle lenses, the movement of the entrance pupil is not basically due to the "selection of rays" by lens components mountings that act as a secondary stops for oblique ray bundles (as you seem to suggest). Even if the diaphragm is closed down to let pass the narrowest possible ray and without interception, the pupil shall moves. And with some lenses it may move a lot! An animated illustration is shown for example on Fig.16 at:
      http://www.pierretoscani.com/echo_fisheyes_english.html
      You may see that the physical vignetting and the subsequent interception of rays start to happen only at about 30� when the pupil had already moved backward and away from the axis. BTW in a fisheye the same happens but in a forward fashion.

      The "normal" lateral distortion of a lens varies with the object distance (and varies more by focus distance setting) but practically not by very much. Whereas the movement of the pupil causes an additional distortion that is negligible at large object distance that becomes preponderant at close object distance and that compounds the "normal" distortion. The height of the image is consequently changed simultaneously by these factors. In short: distortion and FOV depend on distance: the closer the distance is, the stronger is the dependance.
      For a good illustration you may see http://toothwalker.org/optics/distortion.html
      Paul van Walre did not take explicitly into account the entrance pupil movement for close range object imaging, but IMHO his photo clearly demonstrates the resulting spectacular effect on figure 5!

      OK, this has been known for long as fisheyes are concerned, but extensive tests have recently convinced me that it applies also for real-life standard wide-angle lenses: stitching of close-up photographs may then be problematic because of inevitable parallax induced errors.

      To come back to the OP topic and question: Panotools is a good and accurate way to measure HFOV when (in addition to shift d and e) it is associated with adequate distortion correction coefficients, but they must be put in relation with object distance if it's close.

      Regards,

      Michel Thoby



      [Non-text portions of this message have been removed]
    • Show all 23 messages in this topic