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

47316Re: What is the actual Field of View?

Expand Messages
  • prague
    Jan 7, 2011
      Thanks Michel, I always learn so much from your posts. Fascinating!

      --- In PanoToolsNG@yahoogroups.com, Michel Thoby <thobymichel@...> wrote:
      > Le 6 janv. 2011 à 22:37, Erik Krause a écrit :
      > > Am 06.01.2011 21:18, schrieb engstrom_henrik:
      > >
      > > > This gives a nominal hfov of 56.3 deg, not 57.5. It is still not
      > > > exactly the same as the measured 55.5, but closer. The difference in
      > > > focal length has dropped to 1.7%, compared to the previous 4%.
      > > The datasheet states a certain amount of barrel distortion, which
      > > changes the actual FoV compared to the theoretical value you get from
      > > focal length.
      > Fully agreed. However, I am not sure to understand correctly the signification of the all concerned variables that are present on the whole lens datasheet. Is there a page of abbreviations definition somewhere in the manufacturer web site?
      > BTW, the relative distortion (< 0 = barrel) is presented on a chart as a function of the image radial field position(u'/u'max). As distortion is a change of the magnification in the image, Schneider-Kreuznach could have equivalently presented the real (i.e., measured) focal length profile in the image field. I MO this alternate innovative way of presenting data very clear, informative and attractive. DxOAnalyzer for instance, offers such a presentation option to the user.
      > >
      > >
      > > > I have considered using zoom lenses so one can fine-tune the hfov
      > > > before mounting them. But the zoom lenses I have seen have so much
      > > > more lens distortion then fix focal lenses so this will be a last
      > > > resort.
      > >
      > > Will the images of this lenses actually be stitched using one of the
      > > panotools based stitchers (hugin, PTGui...) or will they be used
      > > otherwise? Panotools lens correction model can very well deal even with
      > > strong and unusal distortions. See
      > > http://wiki.panotools.org/Lens_correction_model for details.
      > I concur: if distortion on the video image could be corrected on-the-fly using the stitcher or a dedicated software, distortion should consequently not be a major selection criterion.
      > IMO discarding a lens only because of higher distortion might not yield the best selection result at all in these modern times. Even for normal photograph (single, not stitched), image processing (on-board or post-treatment) can correct raw image flaws and modern technologies (hardware and software) have changed the paradigm: the best corrected output image IQ doesn't necessarily come from the less-distorted initial image.
      > The topic of "potential image quality of a digital camera" is the subject of current intense research in part because it is of upmost importance for the low-end lucrative photographic segment (e.g., camera phone "SMIA" and "CPIQ" initiatives).
      > You may read overviews on this subject in papers such as "Information capacity: a measure of potential image quality of a digital camera" by DxoLabs or "Image quality quantification in camera phone application" by Micron Technology for instance.
      > Let's take examples: the Nikkor 14-24 mm (i.e., a zoom lens introduced with the FX DSLR in 2007) isn't by far with the lowest distortion in its class amongst the competition, but it might be the best lens ever made by Nikon and is coveted by most Nikonists despite the price. On the lowest end of the price scale, the Samyang 14 mm f2.8 happens to be probably the worst rectilinear lens as distortion is concerned. It is however one of the best "overall" performer when adequate post-processing is applied. Even better than the Nikkor 14-24 mm: it has thus become my own favorite lens despite the lack of "automatism"!
      > Finally and as Erik suggests, when stitching of individual images is considered then distortion correction, warping and therefore interpolation shall be always involved, thus making all lenses nearly equal when distortion is only concerned.
      > Regards,
      > Michel Thoby
      > [Non-text portions of this message have been removed]
    • Show all 23 messages in this topic