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

Opinion - quality - 6 or 4 shots?

Expand Messages
  • Rick Drew
    I m shooting with a Canon 7D and Sigma 8mm 3.5 lens. I have shot both 6 photos and 4 photos (not including the Zenith and Nadir - always shoot those) - I
    Message 1 of 10 , Oct 4, 2010
    • 0 Attachment
      I'm shooting with a Canon 7D and Sigma 8mm 3.5 lens.



      I have shot both 6 photos and 4 photos (not including the Zenith and Nadir -
      always shoot those) - I really don't see a quality difference in the final
      pano. I've often found that with 6 shots there tends to be stitching errors
      that don't occur in 4 shots. My head is aligned perfectly (spent a lot of
      time on it.)



      Just curious what others are doing and if you have the same results as mine.



      Thanks



      Rick Drew



      [Non-text portions of this message have been removed]
    • Fabio Bustamante
      I don t think it d make much of a difference, Rick. Instead of using 90 degrees of the image circle you ll be using 60. Of course the Sigma 8mm has image
      Message 2 of 10 , Oct 4, 2010
      • 0 Attachment
        I don't think it'd make much of a difference, Rick.

        Instead of using 90 degrees of the image circle you'll be using 60. Of
        course the Sigma 8mm has image degradation towards the borders but my
        guess is that it's not important in that area. I used to shoot panos
        with this lens on a full frame with 3 horizontal shots only and still it
        didn't bother me. However you are in a better position to judge it since
        I used to shoot this lens on a 13mp full frame while you're using it on
        a 18mp 1.6x crop factor. You're much closer to the limit of the
        resolution of this len's than I was.

        My suggestion is making this test: choose a tree or whatever as a
        sharpness reference and place it 45 degrees away from the center and
        take a shot. Then turn the camera 15 degrees so the tree is now 30
        degrees from the center and shoot again. Compare how's it's sharpness at
        both positions. It think it won't be worth taking 6 over 4.

        Don't forget that no matter how many horizontal shots you're making - 4,
        6, 12, 36 or whatever - you'll be only improving the horizontal quality,
        since vertically you'll still be using the more critical areas of this
        lens, i. e., away from the sweet spot.

        Now, these stitching errors shouldn't be happening...

        Fabio

        Em 04/10/2010 18:57, Rick Drew escreveu:
        > I'm shooting with a Canon 7D and Sigma 8mm 3.5 lens.
        >
        >
        >
        > I have shot both 6 photos and 4 photos (not including the Zenith and Nadir -
        > always shoot those) - I really don't see a quality difference in the final
        > pano. I've often found that with 6 shots there tends to be stitching errors
        > that don't occur in 4 shots. My head is aligned perfectly (spent a lot of
        > time on it.)
        >
        >
        >
        > Just curious what others are doing and if you have the same results as mine.
        >
        >
        >
        > Thanks
        >
        >
        >
        > Rick Drew
        >
        >
        >
        > [Non-text portions of this message have been removed]
        >
        >
        >
        > ------------------------------------
        >
      • Edson Maiero
        Rick, When I am using the 8mm (EOS 5d) to shot a place with few people I shot 3 times, but I like to shot 6 photos when it is a place with lot of people
        Message 3 of 10 , Oct 4, 2010
        • 0 Attachment
          Rick,

          When I am using the 8mm (EOS 5d) to shot a place with few people I shot 3
          times, but I like to shot 6 photos when it is a place with lot of people
          walking. 6 photos help me to fix any person that had moved between the
          shots.

          I think that the problem with 6 shot is that when you have only 4 shot (in
          your set up) the photo 1, is aligned with photo 2, that is aligned with
          photo 3, etc... But when you have more photos than necessary the photo 1, is
          aligned with photo 2 and 3...and this can make some problem to the process.



          Maiero




          On Mon, Oct 4, 2010 at 6:57 PM, Rick Drew <rick@...> wrote:

          >
          >
          > I'm shooting with a Canon 7D and Sigma 8mm 3.5 lens.
          >
          > I have shot both 6 photos and 4 photos (not including the Zenith and Nadir
          > -
          > always shoot those) - I really don't see a quality difference in the final
          > pano. I've often found that with 6 shots there tends to be stitching errors
          > that don't occur in 4 shots. My head is aligned perfectly (spent a lot of
          > time on it.)
          >
          > Just curious what others are doing and if you have the same results as
          > mine.
          >
          > Thanks
          >
          > Rick Drew
          >
          > [Non-text portions of this message have been removed]
          >
          >
          >



          --
          Edson Maiero
          http://phototravel.wordpress.com/
          http://twitter.com/emaiero


          [Non-text portions of this message have been removed]
        • Keith Martin
          ... The no-parallax-point for the Sigma may be a little different for a six-shot-around workflow than for a four-around one. Fisheye lenses are optically
          Message 4 of 10 , Oct 5, 2010
          • 0 Attachment
            Sometime around 4/10/10 (at 16:57 -0500) Rick Drew said:

            >I've often found that with 6 shots there tends to be stitching errors
            >that don't occur in 4 shots. My head is aligned perfectly (spent a lot of
            >time on it.)

            The no-parallax-point for the Sigma may be a little different for a
            six-shot-around workflow than for a four-around one. Fisheye lenses
            are optically complex beasties.

            Additionally, I trust you're not allowing control points to go
            between non-adjacent shots? Control point linking should only be done
            between immediately adjacent images, not ones further away in the
            arrangement. If you have 50% or more overlap you might see this being
            done by the automatic control point generation process.

            Take a pano that's given you problems and try setting control points
            manually. Just assign four or maybe five points per image pair, and
            try to have them spread out as much as you can; no clustering
            together. See if this optimises and stitches any better...

            k
          • Rick Drew
            I ll have to try that - I always made it a point to cross link - image 1 to image 2 and 3. 2 to 3 and 4, etc. I m not sure if the autolink creates links
            Message 5 of 10 , Oct 5, 2010
            • 0 Attachment
              I'll have to try that - I always made it a point to cross link - image 1 to
              image 2 and 3. 2 to 3 and 4, etc. I'm not sure if the autolink creates
              links across non-adjacent images or not., but I was doing it manually. I'll
              stop that. Guess that's why I have to add so many points!



              One think I've been curious about - Ptgui always assigns a "roll" value -
              albeit small. If I set these to 0 I don't see any stitching errors. Is the
              "roll" needed?



              Most problems occur when I'm really close to an object. For example, I shoot
              next to and above a lattice hand-railing. When you look down you can see
              the profile of the hand-rail. In the distance are some curved features.
              Part of the curved features and bottom of the handrail had a minor stitch
              error, all related to the same shot. It was also an evening photo with
              little contrast. It's my recent WWP entry -



              http://www.worldwidepanorama.org/worldwidepanorama/wwp910/html/RichardCDrew-
              6331.html



              I finally gave up and just Photoshopped the errors (and darn if I did not
              just see one I missed!)



              I'm a perfectionist and take time to look for any errors, even minor ones.
              First I'll try adding more control points, and if that does not work, I'll
              fix it in Photoshop.



              Thanks



              Rick Drew



              From: PanoToolsNG@yahoogroups.com [mailto:PanoToolsNG@yahoogroups.com] On
              Behalf Of Keith Martin
              Sent: 2010-10-05 3:05 AM
              To: PanoToolsNG@yahoogroups.com
              Subject: Re: [PanoToolsNG] Opinion - quality - 6 or 4 shots?





              Sometime around 4/10/10 (at 16:57 -0500) Rick Drew said:

              >I've often found that with 6 shots there tends to be stitching errors
              >that don't occur in 4 shots. My head is aligned perfectly (spent a lot of
              >time on it.)

              The no-parallax-point for the Sigma may be a little different for a
              six-shot-around workflow than for a four-around one. Fisheye lenses
              are optically complex beasties.

              Additionally, I trust you're not allowing control points to go
              between non-adjacent shots? Control point linking should only be done
              between immediately adjacent images, not ones further away in the
              arrangement. If you have 50% or more overlap you might see this being
              done by the automatic control point generation process.

              Take a pano that's given you problems and try setting control points
              manually. Just assign four or maybe five points per image pair, and
              try to have them spread out as much as you can; no clustering
              together. See if this optimises and stitches any better...

              k





              [Non-text portions of this message have been removed]
            • John Houghton
              ... Rick, PTGui doesn t assign roll values. The optimizer will evaluate yaw, pitch and roll values in the alignment process. It s not uncommon for roll to
              Message 6 of 10 , Oct 5, 2010
              • 0 Attachment
                --- In PanoToolsNG@yahoogroups.com, "Rick Drew" <rick@...> wrote:
                > One think I've been curious about - Ptgui always assigns a "roll"
                > value - albeit small. If I set these to 0 I don't see any stitching
                > errors. Is the "roll" needed?

                Rick, PTGui doesn't "assign" roll values. The optimizer will evaluate yaw, pitch and roll values in the alignment process. It's not uncommon for roll to be non-zero, owing to the tendency for the weight of the camera to bend/twist the panohead arms very slightly. One expects the optimizer to find the correct roll values if it's controlled properly. If you get improved results by setting the roll values at 0 (and not reoptimizing), then there is something odd about your processing.

                It's difficult to diagnose the cause of the stitching errors without access to the images and your project file. If you can make these available, it should be possible to get a better idea of what's going on.

                John
              • Michel Thoby
                Hi Rick, ... I believe that you have encountered the fisheye bulging problem. It is a direct consequence of the inherent characteristics of the fisheye
                Message 7 of 10 , Oct 6, 2010
                • 0 Attachment
                  Hi Rick,

                  Le 5 oct. 2010 à 21:07, Rick Drew a écrit :

                  > Most problems occur when I'm really close to an object. For example, I shoot
                  > next to and above a lattice hand-railing. When you look down you can see
                  > the profile of the hand-rail. In the distance are some curved features.
                  > Part of the curved features and bottom of the handrail had a minor stitch
                  > error, all related to the same shot. It was also an evening photo with
                  > little contrast. It's my recent WWP entry -
                  I believe that you have encountered the fisheye "bulging" problem. It is a direct consequence of the inherent characteristics of the fisheye lenses where the entrance pupil moves when the angle of entering light changes.

                  When a straight object (or edge) is VERY close to the lens, it shall be partially, and at this close distance, more strongly distorted on the image than any other straight line that is more distant. If this happens on the seam of two adjacent images it makes stitch errors. This may also yield ugly bulging distortion (away from any seam) at the viewing final stage.
                  I wrote a specific study of this problem:
                  http://michel.thoby.free.fr/Scholtes/Parallaxe/Plague%20on%20claustro.html

                  Regards,

                  Michel

                  [Non-text portions of this message have been removed]
                • Keith Martin
                  ... This is also why the perfect head alignment for 6 shots isn t necessarily the same as the perfect head alignment for four shots with the same lens. The
                  Message 8 of 10 , Oct 6, 2010
                  • 0 Attachment
                    Sometime around 6/10/10 (at 11:05 +0200) Michel Thoby said:

                    >I believe that you have encountered the fisheye "bulging" problem.
                    >It is a direct consequence of the inherent characteristics of the
                    >fisheye lenses where the entrance pupil moves when the angle of
                    >entering light changes.

                    This is also why the perfect head alignment for 6 shots isn't
                    necessarily the same as the perfect head alignment for four shots
                    with the same lens. The critical area of overlap will be in a
                    different part of the shot (okay, slightly), so the effective
                    no-parallax-point could be slightly different. We're talking about
                    small differences, but it is worth considering.

                    But I'd say that the issue with near objects isn't one that's been a
                    problem for me (10.5mm fisheye). Either you have some fine-tuning to
                    do with your head setup or I've been lucky and not had very near
                    objects in the overlap area.

                    k
                  • Michel Thoby
                    ... Yes, I agree. You should consider setting the longitudinal position of the lens accordingly to the angle of rotation between successive shots. If the
                    Message 9 of 10 , Oct 6, 2010
                    • 0 Attachment
                      Le 6 oct. 2010 à 11:30, Keith Martin a écrit :

                      > >I believe that you have encountered the fisheye "bulging" problem.
                      > >It is a direct consequence of the inherent characteristics of the
                      > >fisheye lenses where the entrance pupil moves when the angle of
                      > >entering light changes.
                      >
                      > This is also why the perfect head alignment for 6 shots isn't
                      > necessarily the same as the perfect head alignment for four shots
                      > with the same lens. The critical area of overlap will be in a
                      > different part of the shot (okay, slightly), so the effective
                      > no-parallax-point could be slightly different. We're talking about
                      > small differences, but it is worth considering.
                      >
                      > But I'd say that the issue with near objects isn't one that's been a
                      > problem for me (10.5mm fisheye). Either you have some fine-tuning to
                      > do with your head setup or I've been lucky and not had very near
                      > objects in the overlap area.

                      Yes, I agree. You should consider setting the longitudinal position of the lens accordingly to the angle of rotation between successive shots. If the subject is really at close range from the fisheye lens, this is mandatory.

                      BTW, I have observed the very same problem with a recent extreme wide angle standard lens (Samyang 14 mm) and reported my result:
                      http://tinyurl.com/2u2judu

                      Michel



                      [Non-text portions of this message have been removed]
                    • Hans
                      ... The problem is that if you adjust for 6 shots / at the centre) you get larger parallax at the + -45 degrees areas which is actually where most people have
                      Message 10 of 10 , Oct 6, 2010
                      • 0 Attachment
                        --- In PanoToolsNG@yahoogroups.com, Keith Martin <keith@...> wrote:
                        >
                        > Sometime around 6/10/10 (at 11:05 +0200) Michel Thoby said:
                        >
                        > >I believe that you have encountered the fisheye "bulging" problem.
                        > >It is a direct consequence of the inherent characteristics of the
                        > >fisheye lenses where the entrance pupil moves when the angle of
                        > >entering light changes.
                        >
                        > This is also why the perfect head alignment for 6 shots isn't
                        > necessarily the same as the perfect head alignment for four shots
                        > with the same lens. The critical area of overlap will be in a
                        > different part of the shot (okay, slightly), so the effective
                        > no-parallax-point could be slightly different. We're talking about
                        > small differences, but it is worth considering.
                        >

                        The problem is that if you adjust for 6 shots / at the centre) you get larger parallax at the + -45 degrees areas which is actually where most people have the problems.

                        In my opinion adjusting for the number of shots without doing zenith is just wrong.
                        The blending point will increase all the way up to the zenith and if you shoot 6 you have a very large parallax difference.

                        Also if you shoot zenith the critical area is where the zenith blends the 4/6 around and that will be at 90 degrees both for zenith and the 4 around but if you shoot 6 you have a mismatch with the 6 around blending at 60 while the zenith blends at 90

                        You can use the new feature in the editor in PTgui 9 beta to see exactly where you blend.

                        Hans
                      Your message has been successfully submitted and would be delivered to recipients shortly.