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

Re: Precision of the Peleng

Expand Messages
  • Juergen Schrader
    Boy, you really made my day. These excellent figures rose my envy yesterday and my panogear had to undergo a meticulous revision. Especially for the Nikon 10,5
    Message 1 of 2 , Mar 26, 2007
    • 0 Attachment
      Boy, you really made my day.

      These excellent figures rose my envy yesterday and my panogear had to
      undergo a meticulous revision. Especially for the Nikon 10,5 which I
      always suspected not to deliver the exactest possible results.

      I found some slight offsets of the optical axis which didn't affect
      the results of my rectilinear lenses but after fixing these the
      results with the fisheye became signifcantly better :)

      So if we ever get into handshaking distance be prepared to have a
      free beer ;)

      Cheers
      J├╝rgen


      --- In PanoToolsNG@yahoogroups.com, "dmgalpha" <dmgerman@...> wrote:
      >
      > Here is an example of how precise the Peleng can be. This is
      > a pano with objects very close (60cm or so, the closest).
      >
      > This is the set up:
      >
      > * Canon 20d with Peleng 8mm
      > * Pinnacle VR head
      > * Hugin
      > * 8 images around + 1 zenith. The tripod is left in the pano.
      >
      > As we know we would ideally like to make the seams of one image
      > exactly fall into the next image. MOst of the time they don't quite
      > match, but if we mask the edges of the images (creating a stitching
      > mask) we get decent results.
      >
      > In this image, the seams of one image almost always match the next
      > image (or are off by very few pixels). To show you I have created a
      > photoshop file, where each layer
      > is the peleng image, with no cropping.
      >
      > Be warned, these files are large.
      >
      > http://turingmachine.org/~dmg/temp/labRolled.psd
      >
      > This JPG is created by rolling each image on top of each other, with
      > NO stitching masks, no blending and no editing of masks. It is just
      a
      > call to PTroller with the mapped images
      >
      > http://turingmachine.org/~dmg/temp/labRolled.jpg
      > If it wasn't for the vignetting of the lens, and the edges of the
      > circle it would be difficult to detect the edges.
      >
      > and this is the version after enblend has been run on the images:
      >
      > http://turingmachine.org/~dmg/temp/labBlended.jpg
      >
      > Mark Fink has asked what I think about the Pinnacle VR. Well, it is
      > very precise. This is are the actual yaw, pitch, and roll for each
      of
      > the 8 images:
      >
      > p-11.908 r0.327964 y0
      > p-10.6091 r-0.993962 y45.9353
      > p-9.70911 r-0.589997 y90.7391
      > p-8.80028 r0.338645 y135.48
      > p-8.79255 r1.69837 y-179.343
      > p-9.69431 r2.61333 y-134.485
      > p-10.9817 r2.63146 y-90.1969
      > p-11.9368 r1.67491 y-44.4717
      > p87.7206 r44.5658 y-45.4099
      >
      > as you can see the yaws are all within less than 1 degree off. This,
      > of course, is no indication exactly how accurate the head is, but
      it gives
      > you an idea of how valuable it is (for indoors panos with close
      > objects). the pitch was
      > supposed to be 10 degrees, but I did not level the head, so their
      > variance doesn't really tell me much.
      >
      >
      > dmg
      > http://silvernegative.com
      > http://turingmachine.org
      >
    Your message has been successfully submitted and would be delivered to recipients shortly.