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

Re: [PanoToolsNG] Re: PTGui/PTMac Lens Database?

Expand Messages
  • Milko K. Amorth
    Hi Sacha, I think we might have totally scared off Kristin with these tech answers to a relatively simple question. I hope she is not discouraged by this. Let
    Message 1 of 8 , Jan 6, 2008
    View Source
    • 0 Attachment
      Hi Sacha,
      I think we might have totally scared off Kristin with these tech answers
      to a relatively simple question.
      I hope she is not discouraged by this. Let me just say that she can use
      any data of lens profiles provided with either software.
      The import however has to be manually. Of course the calibration data
      for the lens is only for a, b and c parameters.
      Fov and d and e are determined by the crop you apply and how the
      lens/sensor relation pertains to your particular combo.
      D an E always varies. I haven"t seen a lens which does not shift in its
      mount. For repeatable results or scripting you need a bracket or a strong
      rubber strap to pull the lens into a registered position.
      >
      > Right, but I think most agree, if they use ptgui, templates are far better
      > than a lens db, simply for the fact that the cropping is correct.
      >
      Depending on the lens/cam registration, yes. If the lens is not
      registered there will always be a shift you have to optimize for.
      As i have said, I have yet to see a lens mount stay fixed and be
      registere each and every time out of the box.
      >
      >
      > If you use only a lens db, and recreate the cropping, you might as well
      > start from zero.to get a good stitch.
      >
      I disagree. a b and c values of a calibration target are better than
      from just any subject. Often there are not enough good features to be
      correct.
      It is hard to tell if you are correct if the image fits and ultimately
      that is the goal. However, if you take these achieved values and apply
      it to a
      perfect grid image to test the lens, you will see that the values are
      never quite right.
      Have you ever ended up with every pano from scratch with the exact same
      values?

      Calibrated values in the lens database are a treasure. Especially to
      correct single shots. Thats essentially what PTlens, Lensfix or Radcor
      do. For each parameter you know is correct you do not have to optimize
      for and dont need a cp for it.

      >
      > The rotation for each shot with a fisheye is incredibly important.
      > For instance, same lens, same camera dimensions, but the NPP is almost
      > half
      > an inch different going from 72 degrees rotation to 45 degrees rotation.
      >
      Its always been that way. The NPP shifts depending on aperture and angle
      of view of your segment used in the pano (i.e. Sigma 8 npp range is 13
      mm from beta 0 to 90 ). That shift applies to pitched (vertical)
      segments as well.
      >
      >
      > Which also means, if you do creative composition shots, where you
      > alter your
      > rotations for the subjects, your stitching and optimization is going to be
      > all over the place.
      >
      Yes. And you can not expect correct abc lens values from that can you?
      Never will.
      >
      >
      > As is evident by my experience with hundreds of projects using a template
      > and the difference in optimization when I freed it to get best parameters.
      >
      Thats exactly it. Its always an approximation depending on features given.
    Your message has been successfully submitted and would be delivered to recipients shortly.