No announcement yet.

Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

  • Filter
  • Time
  • Show
Clear All
new posts

  • Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

    I am getting a different distance for the lower arm for the Canon 40D using the supplied camera plate. The upper arm is at about the 49 mm marking. I have attempted to make the measurement with a set non digital calipers and get 51 mm +-error (I am not able to measure this very accurately) The table has a lower setting of 54.5 mm for the Canon 40D using the camera plate. I am looking through the view finder with the upper arm at 90 degrees and the tripod leveled using the bubble in the NN5L. I also switched on live view and changed to x10 setting to see if they agree. The middle exposure point is shifted slightly to the right

    I have been using my NN5L for a while with my Canon 40D and Sigma 10mm fisheye which has a 167 degree on Canon APS-C and 180 degree FOV on Nikon DX APS-C sensors. Initially I did a quick alignment with a Zantac pill bottle and a pen. I also shot with too many stops at eight detents. I am now doing what I should have done initially and fine tuning my settings. I am going to change to six detents and see how that effects my results.

    I will be calling Sigma again when a particular person is to see how different the entry pupil site is on the 10mm versus the 8mm.
    Focal length, diameter (mm), length (mm), groups, elements
    8 mm, 73.5 mm, 68.6 mm, 6 groups, 11 elements
    10mm, 75.5 mm, 83.1 mm, 7 groups, 12 elements

    EDIT 04Aug2008 10:56am pacific:
    Just rcvd information from Sigma USA support
    This information is measured from image plane when lens is focused at infinity (elements move when focusing) and something about Lens head (lens head term may have translation issue)
    Lens, 1st Nodal Point, 2nd Nodal point (rear, defines focal length), Entry pupil, Exit pupil
    Sigma 8mm f 3.5, NP1 87.4 mm, NP2 8.1 mm, Entry 94.3 mm, Exit 54.0 mm
    Sigma 10mm f 3.5, NP1 88.8 mm, NP2 10.2 mm, Entry 97.5 mm, Exit 72.5 mm

    Tangent: Sigma support said diffraction losses start at lower aperture for wide angle lenses. He says that I need to try different stops empirically and see where I get better results. I have usually been shooting at f11 because I read that that was a sweet spot for the 20D and a particular lens for depth of field and diffraction loss. Of course this is with the caveat that you don't want to shoot very open with a wide angle lens unless necessary to get the shot (low aperture)

  • #2
    Re: Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

    I took a risk and changed many variables, but it payed off with better results so far. I changed my lower arm setting to 49mm according to the markings on the lower arm. I reduced the detent from 8 to 6. I changed my pattern from {8 @ 0 degrees, 8 @ +30 degrees, 8 @ -30 degrees, Zenith, Nadir, Hand held nadir} to {6 @ 20 degrees, 6 @ -20 degrees, zenith, nadir, hand held nadir}. I entered the image parameters from the NN5L, set lens type to circular and applied a crop to all images. I then realized an average error of 1.164 pixels after deleting about 4 control points ranging from 17 pixels error to 12 pixels error for an interior shot. I bracketed my shots in manual with EV0 at ISO 400 (cringe), f11, 1 second with ambient light only.

    That was the most effortless stitch I have ever done. 1.164 pixels after removing 4 control points. I am glad that I didn't change the upper arm setting yet, as I had no basis for changing it except for the 8mm 3.5 lens table entry. (I am shooting with the Sigma 10mm f3.5 fisheye) Is that a good stitch in ptgui for you?

    I also tried to stitch it with autopano pro v1.4.2 with a global RMS of 2.27, but both the nadir and the zenith were not properly linked. There was even a funny crop on my zenith which I have never seen. I didn't even try and stitch it with the 2 beta because it crashed immediately when I used it for previous stitches.

    I will not change the upper arm until I have something to support the change, such as feedback when using the Belgian guys method that I discovered as a link in Autopano's wiki


    • #3
      Re: Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

      An average cp distance of 1.164 is good, but the maximum distance is important too. A maximum of <2 would be very good. It's also important to have the points well spread out along the overlap areas and not clustered at the centre. But why have you specified the lens type as circular, when the 10mm is a fullframe fisheye? You shouldn't need to take 15 shots for 360x180. 6 around + nadir and zenith are surely enough. As for the NN5 setup, you can do the whole thing quite accurately in 10-15 minutes without measuring anything. See



      • #4
        Re: Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

        Thanks for the link. That helps. I had seen your tape method a while back and couldn't remember where I saw it. The first pin method probably helps shorten the amount of time that it takes with the tape by getting you close.

        I am pretty sure that my Sigma 10 mm is a circular. Maybe I don't understand what circular and full frame mean.


        • #5
          Re: Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

          Thanks for posting the image. It explains your confusion perfectly. The lens as you are using it could certainly be termed a circular fisheye, but with an angle of view of only 100 degrees or so. But in fact it's a fullframe fisheye and you have made a very common error. You haven't removed the ring that the lens cap fits into! The ring is a push fit and needs to be pulled off in order for the full fov to be obtained.



          • #6
            Re: Canon 40D w/ CP lower arm setting (54.5 expected - 51 actual) Sigma 10mm

            Thanks John This is one of those humbling moments when all you can do is blush and then laugh. I would have never tried to pull that off as I thought it was part of the lens hood designed to prevent flares, etc. Wow. I even hesitated after you told me to pull too hard doubting it. Once you pull off the cover the lens seems to be naked and exposed

            That also probably explains a question I had about why ptgui was reporting a fov of about 138 degrees instead of 167 degrees.