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

Re: [PanoToolsNG] GSV in 3D

Expand Messages
  • mick crane
    ... http://www.iwantoneofthose.com/pad-dock/index.html
    Message 1 of 28 , Apr 1, 2010
    • 0 Attachment
      Bjørn K Nilssen wrote:

      > Aren't we a bit too paranoid today? Why should it be a fake/joke? It
      > apparently works
      http://www.iwantoneofthose.com/pad-dock/index.html
    • Wim Koornneef
      For a April 1 joke it is done very well but not for a real experiment. With a red/cyan anaglyph viewer I noticed that the stereo separation, which should be
      Message 2 of 28 , Apr 1, 2010
      • 0 Attachment
        For a April 1 joke it is done very well but not for a real experiment.

        With a red/cyan anaglyph viewer I noticed that the stereo separation, which
        should be applied to far objects, was set to close by objects resulting in a
        severe stereo window violation (I know this because I made the same huge
        mistake when I started making 3D panos).

        I m pretty sure that this is a plain pano that is manipulated with software
        to show an artificial 3D effect, there are to many things not right in the
        pano, you can see it when you rotate the pano with a viewer on.
        There is another thing that makes it almost for sure a fake 3D pano, for
        shooting 3D panos in a dynamic environment like a streetview you must have
        at least 24 (more is better) static mounted cameras and then use the
        parallax between the images to create the left and right eye panos needed
        for the creation of an anagyph 3D pano.

        BTW, making a 3D pano can be done even with a single camera and a fisheye
        lens but for this you need a "frozen" scene. On the website of NodalNinja I
        posted a tutorial how to do this.
        http://nodalninja.com

        Wim

        --
        View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1748067.html
        Sent from the PanoToolsNG mailing list archive at Nabble.com.
      • ahoeben41
        ... I already had a headache before I read your message and started thinking about it, so please forgive me if I say something stupid. The stereo separation is
        Message 3 of 28 , Apr 1, 2010
        • 0 Attachment
          > With a red/cyan anaglyph viewer I noticed that the stereo separation, which
          > should be applied to far objects, was set to close by objects resulting in a
          > severe stereo window violation (I know this because I made the same huge
          > mistake when I started making 3D panos).

          I already had a headache before I read your message and started thinking about it, so please forgive me if I say something stupid.

          The stereo separation is basically the same thing as parallax "error" between your two eyes, right? And we all know that the closer objects are to the lens, the more parallax errors are an issue. Parallax issues are far less of a problem with far away objects (ie: mountain views are easy to stitch).

          So... shouldn't far away objects have little separation in stereo panoramas? Just like Google does it now?

          Ofcourse this is on the assumption that you are focussing at infinity; if you focus closer than infinity you may get stereo-separation again on distant objects because your eyes are slightly crosseyed. Objects further away than the focuspoint will separate in the other direction than objects nearer than the focuspoint. Or am I mistaken?

          Ouch, headache...

          PS: I am still convinced there is no such thing as a fully "correct" 360x180 spherical stereo panorama. I guess I may have to wait until Plymouth to be really convinced.
        • panovrx
          ... no that is right Stereo games drivers have an autoconvergence feature so that the bulk of the scene is always behind the window -- ie. the apparent plane
          Message 4 of 28 , Apr 1, 2010
          • 0 Attachment
            --- In PanoToolsNG@yahoogroups.com, "ahoeben41" <aldo@...> wrote:

            > Ofcourse this is on the assumption that you are focussing at infinity; if you focus closer than infinity you may get stereo-separation again on distant objects because your eyes are slightly crosseyed. Objects further away than the focuspoint will separate in the other direction than objects nearer than the focuspoint. Or am I mistaken? >

            no that is right

            Stereo games drivers have an autoconvergence feature so that the bulk of the scene is always behind the "window" -- ie. the apparent plane of the screen -- but not too far behind. This would be a handy feature for stereo pano viewers too but it would need a depth map to work I think. You can generate a depth map from the parallax information in a stereo view but I dont know how well depth map generation can be done in real time. If you have a depth map anyway like StreetView would have if they started shooting stereo panos then it would not be an issue.

            There is the concept of the depth budget of stereo displays too viz. the range of reproducible depth impression of a particular display technology. Projection 3d has a large depth budget, lenticular and autostereoscopic displays generally small. Subject depth must be matched to the depth budget of the targeted display otherwise the depth impression will be too small or large. If too small things look flat, if too large your brain wont resolve the parallax. In real world situations when we look at something close we ignore the image doubling in the background but not so much with displays.

            With stereo panoramas you will often have a difficult situation for depth budgets. The nearest thing to the camera will often be 2m away at ground level around the camera, and the furthest infinity. Which is a big range if the main interest is more than a few meters away. The screen depth will have to placed at 2m which means the depth budget for the more distant background is meagre. So there is a big argument for pole panoramas for hyperstereoscopic outdooor scenes.

            Peter M
          • panovrx
            ... apropos which http://www.mediavr.com/rocksana4.jpg .. another one of my experiments with rapid video capture of stereo panos. I think I am going to give up
            Message 5 of 28 , Apr 1, 2010
            • 0 Attachment
              >
              > With stereo panoramas you will often have a difficult situation for depth budgets. The nearest thing to the camera will often be 2m away at ground level around the camera, and the furthest infinity. Which is a big range if the main interest is more than a few meters away. The screen depth will have to placed at 2m which means the depth budget for the more distant background is meagre. So there is a big argument for pole panoramas for hyperstereoscopic outdooor scenes.
              >
              > Peter M
              >

              apropos which
              http://www.mediavr.com/rocksana4.jpg
              .. another one of my experiments with rapid video capture of stereo panos. I think I am going to give up on this btw (video stereo pano capture with the 5D) as the compression artefacts are so strong. A pity as it is nice to be able to make a stereo spherical 360 record in 4 sec. I need two Scarlets or similar I guess.

              This is with the cameras quite far apart ~ 20cm with them on a pole 3m from the ground. You can see the wharf edges are a little behind the stereo window and most of the foreground is in front of the window -- but it is not too intrusive here as the contrast is so low there -- but if the contrast/definition was higher there it would look wrong.
              If the camera separation was any less the depth of the buildings would be minimal.



              Peter M
            • Roger D. Williams
              ... I appreciate and agree with all that you wrote earlier in your message but I wonder how you consider zooming in and out should affect things? I was very
              Message 6 of 28 , Apr 1, 2010
              • 0 Attachment
                On Fri, 02 Apr 2010 08:01:25 +0900, panovrx <panovrx@...> wrote:

                > With stereo panoramas you will often have a difficult situation for
                > depth budgets. The nearest thing to the camera will often be 2m away at
                > ground level around the camera, and the furthest infinity. Which is a
                > big range if the main interest is more than a few meters away. The
                > screen depth will have to placed at 2m which means the depth budget for
                > the more distant background is meagre. So there is a big argument for
                > pole panoramas for hyperstereoscopic outdooor scenes.

                I appreciate and agree with all that you wrote earlier in your message
                but I wonder how you consider zooming in and out should affect things?
                I was very surprised and impressed recently to see that when I zoomed
                into a stereo panorama to focus on some background buildings, the
                apparent stereo effect was enhanced so that the buildings looked
                really solid rather than flat, as they did when I was zoomed fully
                out. This made the whole experience very "real" in the sense that I got
                the impression I really was zooming closer to the more remote objects
                of interest, rather than just enlarging a picture of them.

                Roger W.

                --
                Business: www.adex-japan.com
                Pleasure: www.usefilm.com/member/roger
              • panovrx
                ... Depth realism and stereoscopy is a vexed question which I havent got my head around really. And zoomable interactive stereo viewing imagery is not
                Message 7 of 28 , Apr 2, 2010
                • 0 Attachment
                  --- In PanoToolsNG@yahoogroups.com, "Roger D. Williams" <roger@...> wrote:

                  > but I wonder how you consider zooming in and out should affect things?
                  > I was very surprised and impressed recently to see that when I zoomed
                  > into a stereo panorama to focus on some background buildings, the
                  > apparent stereo effect was enhanced so that the buildings looked
                  > really solid rather than flat, as they did when I was zoomed fully
                  > out. This made the whole experience very "real" in the sense that I got
                  > the impression I really was zooming closer to the more remote objects
                  > of interest, rather than just enlarging a picture of them.


                  Depth realism and stereoscopy is a vexed question which I havent got my head around really. And zoomable interactive stereo viewing imagery is not widespread. The idea of a gigapixel stereo panorama where you kept seeing clear depth as you zoomed is very appealing certainly. You could do it with a multicamera array for still or panorama photography I guess where you kept widening the camera separation as you zoomed. In the case of my pano perhaps it was that the anaglyphic depth fringes became more visible as you zoomed.

                  There is the notion of orthoscopy
                  http://www.starosta.com/3dshowcase/ireal.html
                  where you match the lens optical arrangement with the viewing optics so you reproduce the depth impression of reality but this doesnt have much relevance to macro or telephoto stereo. Fuji W owners debate the wisdom of it having wider than eye separation leading to distortions and it certainly does if you get close enough with the camera.

                  Peter M
                • Wim Koornneef
                  Hello Aldo, In his reply Peter perfectly explained about the depth parallax budget (in short budget) that a 3D image should have and the relation of it to the
                  Message 8 of 28 , Apr 2, 2010
                  • 0 Attachment
                    Hello Aldo,

                    In his reply Peter perfectly explained about the depth parallax budget (in
                    short budget) that a 3D image should have and the relation of it to the
                    distance between the viewer and the screen or print. The depth budget can be
                    set totally to the foreground, the background or in between.

                    When you take a look (without a viewer) at Peter's example you see that the
                    budget is set between foreground and background objects.
                    You can see that the position of the cyan and red image edges of the
                    foreground and the background is different and just below the center of the
                    scene the red and cyan edges are merging, this is the zero parallax point.

                    The advantage of this is that parts on the foreground (with a viewer) are
                    "coming out of the screen" so it is just if a part of the scene is floating
                    in front of your monitor, this enhances the 3D experience.
                    If you put all of the budget to front object, like GSV did, you get severe
                    stereo window violation because your left and right eye are seeing more then
                    they should be at the edges of the window.

                    To avoid a long explanation what stereo window violation is and why you
                    should avoid it it is best to take a look here:
                    http://www.vmresource.com/camera/stereowindow.htm

                    As long as the objects in the scene with stereo window violation are not
                    touching a left or right edge in the window and when the violation is not to
                    severe all is fine, but in a 360 pano you turn around and then the violation
                    will kick in as soon as the object hit a window edge and when that happen it
                    will spoil or even ruin the 3D effect and when you are viewing such panos
                    for a longer period of time I can image you will get a headache.
                    For that reason I always set the zero parallax point on close by objects in
                    the scene to avoid any stereo window violation issue.

                    You are right in thinking that it isn't possible to create a perfect 360x180
                    degree spherical 3D pano.
                    The 3D pano that I will show in the dome theatre at the PanoTools Meeting in
                    Plymouth (I will bring ColorCode3D viewers with me) is definitely not a
                    perfect 3D pano as this is impossible by the nature of seeing 3D.
                    It is possible to create a pano with a perfect 3D around the horizon of the
                    scene but definitely not in nadir or zenith because objects in the left and
                    right images are flipping when the pano is rotated 180 degree.

                    When I think about how it is possible that our visual system is so forgiving
                    for those big errors and how it is possible to get a reasonable 3D
                    experience when viewing down in the pano then I get a headiche, I can't
                    understand how it is possible that our visual system is capable of making a
                    3D experience out of this mess.

                    BTW, I have no idea what we will get on the screen of the dome theatre, I
                    will use a anaglyph that will be projected on a large surface.
                    The 3D experience is based on the different colors for our left and right
                    eye and as you know the colors are fading and loosing contrast when
                    projected on large surfaces.
                    I don't have any experience in projecting a 3D pano on a large screen so it
                    can be that the 3D experience will be fine but it can also be a flop.
                    The proof is in the pudding so we have to wait a few more month to find out
                    ;-)

                    Wim


                    ahoeben wrote:
                    >
                    >
                    >> With a red/cyan anaglyph viewer I noticed that the stereo separation,
                    >> which
                    >> should be applied to far objects, was set to close by objects resulting
                    >> in a
                    >> severe stereo window violation (I know this because I made the same huge
                    >> mistake when I started making 3D panos).
                    >
                    > I already had a headache before I read your message and started thinking
                    > about it, so please forgive me if I say something stupid.
                    >
                    > The stereo separation is basically the same thing as parallax "error"
                    > between your two eyes, right? And we all know that the closer objects are
                    > to the lens, the more parallax errors are an issue. Parallax issues are
                    > far less of a problem with far away objects (ie: mountain views are easy
                    > to stitch).
                    >
                    > So... shouldn't far away objects have little separation in stereo
                    > panoramas? Just like Google does it now?
                    >
                    > Ofcourse this is on the assumption that you are focussing at infinity; if
                    > you focus closer than infinity you may get stereo-separation again on
                    > distant objects because your eyes are slightly crosseyed. Objects further
                    > away than the focuspoint will separate in the other direction than objects
                    > nearer than the focuspoint. Or am I mistaken?
                    >
                    > Ouch, headache...
                    >
                    > PS: I am still convinced there is no such thing as a fully "correct"
                    > 360x180 spherical stereo panorama. I guess I may have to wait until
                    > Plymouth to be really convinced.
                    >
                    >
                    >
                    --
                    View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1748892.html
                    Sent from the PanoToolsNG mailing list archive at Nabble.com.
                  • Roger D. Williams
                    ... Thank you, Peter. I am familiar with the concept of orthoscopy, and think that the Fuji W has got it about right provided you don t try to take closeups
                    Message 9 of 28 , Apr 2, 2010
                    • 0 Attachment
                      On Fri, 02 Apr 2010 16:06:53 +0900, panovrx <panovrx@...> wrote:

                      > There is the notion of orthoscopy
                      > http://www.starosta.com/3dshowcase/ireal.html
                      > where you match the lens optical arrangement with the viewing optics so
                      > you reproduce the depth impression of reality but this doesnt have much
                      > relevance to macro or telephoto stereo. Fuji W owners debate the wisdom
                      > of it having wider than eye separation leading to distortions and it
                      > certainly does if you get close enough with the camera.

                      Thank you, Peter. I am familiar with the concept of orthoscopy, and think
                      that the Fuji W has got it about right provided you don't try to take
                      closeups that are TOO close. It is fine for street scenes and even most
                      kinds of scenery, where it suffers proportionately less from the "flat
                      backdrop" effect. The reverse case is seen with the very expensive
                      3D modification of the Hasselblad/Fuji TX-1, where the two lenses are so
                      close together that the stereo effect is quite lost on more distant
                      objects or indeed anything more than 20 or 30 meters away. Still, I am
                      not buying a Fuji W; I am waiting for them to produce a Mk II with some
                      of the more annoying "features" removed.

                      I was cleaning out the room where I have been storing my library and my
                      negatives/slides/prints taken over the last 50 years and came across lots
                      taken with the Pentax mirrored stereo adapter. I also found the viewer
                      (a minor miracle among all the accumulated junk of half a century)
                      although alas the adapter itself is lost. The stereo effect is generally
                      quite satisfactory, although limited to portrait orientation, of course.

                      I have decided to take up the challenge of stereo panoramic photography.
                      Japan is going crazy over 3D TV, and I am sure that there will be plenty
                      of people wanting to look at 3D panoramas. My own best bet is that a
                      single lens rotated about a point deliberately displaced from the NPP,
                      and stereo images formed from thin slices of multiple images taken at
                      very narrow angular separations, will provide the most realistic results.
                      I do have concerns over orthoscopy and the window-edge effect, though.

                      I am thinking that a video camera may be better than a DSLR that takes
                      video, not least because I do not possess one of the latter, and an HD
                      video camera would be less expensive than an upgrade. I know that in
                      this case the quality will be lower but perhaps good enough for
                      web-based presentations of tourist attractions, accommodations and
                      restaurants, which I see as my best market. I do have some fears that
                      a budget-minded customer may opt for simple video sequences rather than
                      the expense of pan- and zoomable stereo panoramas. <sigh>

                      Roger W.

                      --
                      Business: www.adex-japan.com
                      Pleasure: www.usefilm.com/member/roger
                    • Wim Koornneef
                      Hello Roger, I think your idea of using a small HD videocam instead of a DSLR is very appealing. With a relatively cheap fisheye adapter lens (Nikon FC-E8) it
                      Message 10 of 28 , Apr 2, 2010
                      • 0 Attachment
                        Hello Roger,

                        I think your idea of using a small HD videocam instead of a DSLR is very
                        appealing.
                        With a relatively cheap fisheye adapter lens (Nikon FC-E8) it will be
                        possible to capture the footage needed for 3D.
                        There is just one major drawback of using such a setup and that is the fact
                        that a one camera system is not suited to shoot in a dynamic scene, even a
                        fast rotation of 4 seconds is way to much to avoid huge errors.
                        From experience I can tell that you really need a "frozen" scene.
                        So I think a dual HD videocam solution is needed in most scenes but even
                        with twice the costs it will be still a lot less expensive then 2 video
                        capable DSLR's with fisheye lenses.
                        The good part is that you can start with a single videocam and single
                        fisheye adapter lens and upgrade when needed ;-)
                        Please keep us informed about any progress.

                        Best,
                        Wim
                        --
                        View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1749001.html
                        Sent from the PanoToolsNG mailing list archive at Nabble.com.
                      • prague
                        Peter, where do you learn this stuff if I may ask?
                        Message 11 of 28 , Apr 2, 2010
                        • 0 Attachment
                          Peter, where do you learn this stuff if I may ask?

                          --- In PanoToolsNG@yahoogroups.com, "panovrx" <panovrx@...> wrote:
                          >
                          >
                          >
                          >
                          >
                          >
                          > --- In PanoToolsNG@yahoogroups.com, "ahoeben41" <aldo@> wrote:
                          >
                          > > Ofcourse this is on the assumption that you are focussing at infinity; if you focus closer than infinity you may get stereo-separation again on distant objects because your eyes are slightly crosseyed. Objects further away than the focuspoint will separate in the other direction than objects nearer than the focuspoint. Or am I mistaken? >
                          >
                          > no that is right
                          >
                          > Stereo games drivers have an autoconvergence feature so that the bulk of the scene is always behind the "window" -- ie. the apparent plane of the screen -- but not too far behind. This would be a handy feature for stereo pano viewers too but it would need a depth map to work I think. You can generate a depth map from the parallax information in a stereo view but I dont know how well depth map generation can be done in real time. If you have a depth map anyway like StreetView would have if they started shooting stereo panos then it would not be an issue.
                          >
                          > There is the concept of the depth budget of stereo displays too viz. the range of reproducible depth impression of a particular display technology. Projection 3d has a large depth budget, lenticular and autostereoscopic displays generally small. Subject depth must be matched to the depth budget of the targeted display otherwise the depth impression will be too small or large. If too small things look flat, if too large your brain wont resolve the parallax. In real world situations when we look at something close we ignore the image doubling in the background but not so much with displays.
                          >
                          > With stereo panoramas you will often have a difficult situation for depth budgets. The nearest thing to the camera will often be 2m away at ground level around the camera, and the furthest infinity. Which is a big range if the main interest is more than a few meters away. The screen depth will have to placed at 2m which means the depth budget for the more distant background is meagre. So there is a big argument for pole panoramas for hyperstereoscopic outdooor scenes.
                          >
                          > Peter M
                          >
                        • prague
                          i don t have any red/blue glasses on me right now - http://www.360cities.net/image/reisperbachtal-1-anaglyph-austria what happens when you look down on that
                          Message 12 of 28 , Apr 2, 2010
                          • 0 Attachment
                            i don't have any red/blue glasses on me right now -

                            http://www.360cities.net/image/reisperbachtal-1-anaglyph-austria


                            what happens when you look down on that and spin around?


                            here are others if anyone's interested
                            http://360cities.net/search/anaglyph

                            (kudos to martin kneth :-)



                            --- In PanoToolsNG@yahoogroups.com, Wim Koornneef <wim.koornneef@...> wrote:
                            >
                            >
                            > Hello Aldo,
                            >
                            > In his reply Peter perfectly explained about the depth parallax budget (in
                            > short budget) that a 3D image should have and the relation of it to the
                            > distance between the viewer and the screen or print. The depth budget can be
                            > set totally to the foreground, the background or in between.
                            >
                            > When you take a look (without a viewer) at Peter's example you see that the
                            > budget is set between foreground and background objects.
                            > You can see that the position of the cyan and red image edges of the
                            > foreground and the background is different and just below the center of the
                            > scene the red and cyan edges are merging, this is the zero parallax point.
                            >
                            > The advantage of this is that parts on the foreground (with a viewer) are
                            > "coming out of the screen" so it is just if a part of the scene is floating
                            > in front of your monitor, this enhances the 3D experience.
                            > If you put all of the budget to front object, like GSV did, you get severe
                            > stereo window violation because your left and right eye are seeing more then
                            > they should be at the edges of the window.
                            >
                            > To avoid a long explanation what stereo window violation is and why you
                            > should avoid it it is best to take a look here:
                            > http://www.vmresource.com/camera/stereowindow.htm
                            >
                            > As long as the objects in the scene with stereo window violation are not
                            > touching a left or right edge in the window and when the violation is not to
                            > severe all is fine, but in a 360 pano you turn around and then the violation
                            > will kick in as soon as the object hit a window edge and when that happen it
                            > will spoil or even ruin the 3D effect and when you are viewing such panos
                            > for a longer period of time I can image you will get a headache.
                            > For that reason I always set the zero parallax point on close by objects in
                            > the scene to avoid any stereo window violation issue.
                            >
                            > You are right in thinking that it isn't possible to create a perfect 360x180
                            > degree spherical 3D pano.
                            > The 3D pano that I will show in the dome theatre at the PanoTools Meeting in
                            > Plymouth (I will bring ColorCode3D viewers with me) is definitely not a
                            > perfect 3D pano as this is impossible by the nature of seeing 3D.
                            > It is possible to create a pano with a perfect 3D around the horizon of the
                            > scene but definitely not in nadir or zenith because objects in the left and
                            > right images are flipping when the pano is rotated 180 degree.
                            >
                            > When I think about how it is possible that our visual system is so forgiving
                            > for those big errors and how it is possible to get a reasonable 3D
                            > experience when viewing down in the pano then I get a headiche, I can't
                            > understand how it is possible that our visual system is capable of making a
                            > 3D experience out of this mess.
                            >
                            > BTW, I have no idea what we will get on the screen of the dome theatre, I
                            > will use a anaglyph that will be projected on a large surface.
                            > The 3D experience is based on the different colors for our left and right
                            > eye and as you know the colors are fading and loosing contrast when
                            > projected on large surfaces.
                            > I don't have any experience in projecting a 3D pano on a large screen so it
                            > can be that the 3D experience will be fine but it can also be a flop.
                            > The proof is in the pudding so we have to wait a few more month to find out
                            > ;-)
                            >
                            > Wim
                            >
                            >
                            > ahoeben wrote:
                            > >
                            > >
                            > >> With a red/cyan anaglyph viewer I noticed that the stereo separation,
                            > >> which
                            > >> should be applied to far objects, was set to close by objects resulting
                            > >> in a
                            > >> severe stereo window violation (I know this because I made the same huge
                            > >> mistake when I started making 3D panos).
                            > >
                            > > I already had a headache before I read your message and started thinking
                            > > about it, so please forgive me if I say something stupid.
                            > >
                            > > The stereo separation is basically the same thing as parallax "error"
                            > > between your two eyes, right? And we all know that the closer objects are
                            > > to the lens, the more parallax errors are an issue. Parallax issues are
                            > > far less of a problem with far away objects (ie: mountain views are easy
                            > > to stitch).
                            > >
                            > > So... shouldn't far away objects have little separation in stereo
                            > > panoramas? Just like Google does it now?
                            > >
                            > > Ofcourse this is on the assumption that you are focussing at infinity; if
                            > > you focus closer than infinity you may get stereo-separation again on
                            > > distant objects because your eyes are slightly crosseyed. Objects further
                            > > away than the focuspoint will separate in the other direction than objects
                            > > nearer than the focuspoint. Or am I mistaken?
                            > >
                            > > Ouch, headache...
                            > >
                            > > PS: I am still convinced there is no such thing as a fully "correct"
                            > > 360x180 spherical stereo panorama. I guess I may have to wait until
                            > > Plymouth to be really convinced.
                            > >
                            > >
                            > >
                            > --
                            > View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1748892.html
                            > Sent from the PanoToolsNG mailing list archive at Nabble.com.
                            >
                          • Roger D. Williams
                            Hello, Wim. On Fri, 02 Apr 2010 19:38:51 +0900, Wim Koornneef ... Well, I ve never been QUITE so insistent on getting the full vertical 180 degrees included as
                            Message 13 of 28 , Apr 2, 2010
                            • 0 Attachment
                              Hello, Wim.

                              On Fri, 02 Apr 2010 19:38:51 +0900, Wim Koornneef
                              <wim.koornneef@...> wrote:

                              > I think your idea of using a small HD videocam instead of a DSLR is very
                              > appealing.
                              > With a relatively cheap fisheye adapter lens (Nikon FC-E8) it will be
                              > possible to capture the footage needed for 3D.

                              Well, I've never been QUITE so insistent on getting the full vertical 180
                              degrees included as the true 360 x 180 afficionados. There's seldom much of
                              interest at zenith or nadir. I was thinking of keeping to cylindrical
                              panoramas.

                              > There is just one major drawback of using such a setup and that is the
                              > fact
                              > that a one camera system is not suited to shoot in a dynamic scene, even
                              > a
                              > fast rotation of 4 seconds is way to much to avoid huge errors.

                              Are you thinking of the errors you get with any rotating "slit" type
                              camera--the elongation of things moving with the camera and the fore-
                              shortening of those moving in the opposite direction?

                              > From experience I can tell that you really need a "frozen" scene.
                              > So I think a dual HD videocam solution is needed in most scenes but even
                              > with twice the costs it will be still a lot less expensive then 2 video
                              > capable DSLR's with fisheye lenses.

                              I don't see how having two video cameras scanning together can produce a
                              static scene. Could you explain?

                              > The good part is that you can start with a single videocam and single
                              > fisheye adapter lens and upgrade when needed ;-)
                              > Please keep us informed about any progress.

                              I'll do my best. But I made the same decision last year but was unable to
                              make anything of it... we shall see. At least if I succeed I will have
                              something to wow the local stereo photography club members, who have never
                              seen anything like the stereo panoramas we are talking about... And it
                              would wow them even more if I could do it with a single camera!

                              Roger W.

                              --
                              Business: www.adex-japan.com
                              Pleasure: www.usefilm.com/member/roger
                            • Roger D. Williams
                              ... I always keep some near at hand just in CASE someone posts a panorama anaglyph... ... The ground appears to be far to close for the eyes to be able to
                              Message 14 of 28 , Apr 2, 2010
                              • 0 Attachment
                                On Sat, 03 Apr 2010 07:41:23 +0900, prague <360cities@...> wrote:

                                >
                                > i don't have any red/blue glasses on me right now -

                                I always keep some near at hand just in CASE someone posts a panorama
                                anaglyph...

                                > http://www.360cities.net/image/reisperbachtal-1-anaglyph-austria
                                >
                                >
                                > what happens when you look down on that and spin around?

                                The ground appears to be far to close for the eyes to be able to
                                resolve the two images into a single stereo image. Try it and your
                                eyes just hurt. This kind of thing is going to turn a lot of people
                                off 3D TV unless the broadcasters are both aware of this and
                                careful to avoid it.

                                > here are others if anyone's interested
                                > http://360cities.net/search/anaglyph
                                >
                                > (kudos to martin kneth :-)

                                Thanks for the link. I, for one, am most keenly interested.

                                Roger W.

                                --
                                Business: www.adex-japan.com
                                Pleasure: www.usefilm.com/member/roger
                              • Wim Koornneef
                                Hello Jeffrey, When you find your glasses I am sure your findings will the same as mine, the nadir area looks terrible. The main reason is that all the the
                                Message 15 of 28 , Apr 3, 2010
                                • 0 Attachment
                                  Hello Jeffrey,

                                  When you find your glasses I am sure your findings will the same as mine,
                                  the nadir area looks terrible.
                                  The main reason is that all the the available amount of parallax between the
                                  left and right image is applied to the foreground, even without glasses on
                                  you can see that the zero parallax point (the point where the red and cyan
                                  colored edges are merging) is set to the background and that means that the
                                  nadir is having all the parallax. In a previous posting in this thread I
                                  stated that our visual system is very foregiving but obviously not that
                                  much.
                                  You can imagine that when the parallax is set the other way around that in
                                  the nadir area the parallax is much less and then our visual system is
                                  capable of making a reasonable 3D experience out of the mess there.

                                  Wim


                                  Jeffrey Martin | 360Cities.net wrote:
                                  >
                                  >
                                  > i don't have any red/blue glasses on me right now -
                                  >
                                  > http://www.360cities.net/image/reisperbachtal-1-anaglyph-austria
                                  >
                                  >
                                  > what happens when you look down on that and spin around?
                                  >
                                  >
                                  > here are others if anyone's interested
                                  > http://360cities.net/search/anaglyph
                                  >
                                  > (kudos to martin kneth :-)
                                  >
                                  >
                                  >
                                  > --- In PanoToolsNG@yahoogroups.com, Wim Koornneef <wim.koornneef@...>
                                  > wrote:
                                  >>
                                  >>
                                  >> Hello Aldo,
                                  >>
                                  >> In his reply Peter perfectly explained about the depth parallax budget
                                  >> (in
                                  >> short budget) that a 3D image should have and the relation of it to the
                                  >> distance between the viewer and the screen or print. The depth budget can
                                  >> be
                                  >> set totally to the foreground, the background or in between.
                                  >>
                                  >> When you take a look (without a viewer) at Peter's example you see that
                                  >> the
                                  >> budget is set between foreground and background objects.
                                  >> You can see that the position of the cyan and red image edges of the
                                  >> foreground and the background is different and just below the center of
                                  >> the
                                  >> scene the red and cyan edges are merging, this is the zero parallax
                                  >> point.
                                  >>
                                  >> The advantage of this is that parts on the foreground (with a viewer) are
                                  >> "coming out of the screen" so it is just if a part of the scene is
                                  >> floating
                                  >> in front of your monitor, this enhances the 3D experience.
                                  >> If you put all of the budget to front object, like GSV did, you get
                                  >> severe
                                  >> stereo window violation because your left and right eye are seeing more
                                  >> then
                                  >> they should be at the edges of the window.
                                  >>
                                  >> To avoid a long explanation what stereo window violation is and why you
                                  >> should avoid it it is best to take a look here:
                                  >> http://www.vmresource.com/camera/stereowindow.htm
                                  >>
                                  >> As long as the objects in the scene with stereo window violation are not
                                  >> touching a left or right edge in the window and when the violation is not
                                  >> to
                                  >> severe all is fine, but in a 360 pano you turn around and then the
                                  >> violation
                                  >> will kick in as soon as the object hit a window edge and when that happen
                                  >> it
                                  >> will spoil or even ruin the 3D effect and when you are viewing such panos
                                  >> for a longer period of time I can image you will get a headache.
                                  >> For that reason I always set the zero parallax point on close by objects
                                  >> in
                                  >> the scene to avoid any stereo window violation issue.
                                  >>
                                  >> You are right in thinking that it isn't possible to create a perfect
                                  >> 360x180
                                  >> degree spherical 3D pano.
                                  >> The 3D pano that I will show in the dome theatre at the PanoTools Meeting
                                  >> in
                                  >> Plymouth (I will bring ColorCode3D viewers with me) is definitely not a
                                  >> perfect 3D pano as this is impossible by the nature of seeing 3D.
                                  >> It is possible to create a pano with a perfect 3D around the horizon of
                                  >> the
                                  >> scene but definitely not in nadir or zenith because objects in the left
                                  >> and
                                  >> right images are flipping when the pano is rotated 180 degree.
                                  >>
                                  >> When I think about how it is possible that our visual system is so
                                  >> forgiving
                                  >> for those big errors and how it is possible to get a reasonable 3D
                                  >> experience when viewing down in the pano then I get a headiche, I can't
                                  >> understand how it is possible that our visual system is capable of making
                                  >> a
                                  >> 3D experience out of this mess.
                                  >>
                                  >> BTW, I have no idea what we will get on the screen of the dome theatre, I
                                  >> will use a anaglyph that will be projected on a large surface.
                                  >> The 3D experience is based on the different colors for our left and right
                                  >> eye and as you know the colors are fading and loosing contrast when
                                  >> projected on large surfaces.
                                  >> I don't have any experience in projecting a 3D pano on a large screen so
                                  >> it
                                  >> can be that the 3D experience will be fine but it can also be a flop.
                                  >> The proof is in the pudding so we have to wait a few more month to find
                                  >> out
                                  >> ;-)
                                  >>
                                  >> Wim
                                  >>
                                  >>
                                  >> ahoeben wrote:
                                  >> >
                                  >> >
                                  >> >> With a red/cyan anaglyph viewer I noticed that the stereo separation,
                                  >> >> which
                                  >> >> should be applied to far objects, was set to close by objects
                                  >> resulting
                                  >> >> in a
                                  >> >> severe stereo window violation (I know this because I made the same
                                  >> huge
                                  >> >> mistake when I started making 3D panos).
                                  >> >
                                  >> > I already had a headache before I read your message and started
                                  >> thinking
                                  >> > about it, so please forgive me if I say something stupid.
                                  >> >
                                  >> > The stereo separation is basically the same thing as parallax "error"
                                  >> > between your two eyes, right? And we all know that the closer objects
                                  >> are
                                  >> > to the lens, the more parallax errors are an issue. Parallax issues are
                                  >> > far less of a problem with far away objects (ie: mountain views are
                                  >> easy
                                  >> > to stitch).
                                  >> >
                                  >> > So... shouldn't far away objects have little separation in stereo
                                  >> > panoramas? Just like Google does it now?
                                  >> >
                                  >> > Ofcourse this is on the assumption that you are focussing at infinity;
                                  >> if
                                  >> > you focus closer than infinity you may get stereo-separation again on
                                  >> > distant objects because your eyes are slightly crosseyed. Objects
                                  >> further
                                  >> > away than the focuspoint will separate in the other direction than
                                  >> objects
                                  >> > nearer than the focuspoint. Or am I mistaken?
                                  >> >
                                  >> > Ouch, headache...
                                  >> >
                                  >> > PS: I am still convinced there is no such thing as a fully "correct"
                                  >> > 360x180 spherical stereo panorama. I guess I may have to wait until
                                  >> > Plymouth to be really convinced.
                                  >> >
                                  >> >
                                  >> >
                                  >> --
                                  >> View this message in context:
                                  >> http://n4.nabble.com/GSV-in-3D-tp1747596p1748892.html
                                  >> Sent from the PanoToolsNG mailing list archive at Nabble.com.
                                  >>
                                  >
                                  >
                                  >
                                  >
                                  --
                                  View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1749967.html
                                  Sent from the PanoToolsNG mailing list archive at Nabble.com.
                                • Wim Koornneef
                                  Hello Roger, I will try to answer your questions. ... Among others yes. There are at least 4 issues involved when shooting 3D with a single camera and single
                                  Message 16 of 28 , Apr 3, 2010
                                  • 0 Attachment
                                    Hello Roger,

                                    I will try to answer your questions.

                                    >Are you thinking of the errors you get with any rotating "slit" type
                                    >camera--the elongation of things moving with the camera and the fore-
                                    >shortening of those moving in the opposite direction?

                                    Among others yes.

                                    There are at least 4 issues involved when shooting 3D with a single camera
                                    and single lens:
                                    * the compression/expanding effect of slow moving objects that will deform
                                    moving objects, the "slit" effect you mentioned,
                                    * the total absence of fast moving objects in one image, this is due to the
                                    fact that there is a time difference of approx. 0.5-1 second between the
                                    left and right eye pano.
                                    * the vertical hicks you will get in branches, flags etc. when there is wind
                                    * a visible 360 seem when the light is changing, this is especially an issue
                                    when the strips are connected when they are blended this is a much less
                                    issue.

                                    >I don't see how having two video cameras scanning together can produce a
                                    >static scene. Could you explain?

                                    I think I was not clear in this.
                                    What I mean is that with a single camera all of the issues I mentioned above
                                    makes it necessary to only shoot 3D panos in a static, or "frozen" scene.
                                    When there is no movement in the scene and the light is at a fixed level
                                    then there is no problem at all but in all other scenes you have to deal
                                    with at least 2 or 3 issues and then a fast rotating 2 cam system can avoid
                                    most, but definitely not all, issues.

                                    > I was thinking of keeping to cylindrical panoramas.

                                    I am sure that with a wide angle lens you will get a fine result, the
                                    advantage of the cylindrical pano will be a larger resolution for objects
                                    around the horizon due to the fact that you will use all of the available
                                    1920 px frame height of the HD videocam for a relatively small vertical
                                    angle of the panorama.

                                    Best,
                                    Wim
                                    --
                                    View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1749989.html
                                    Sent from the PanoToolsNG mailing list archive at Nabble.com.
                                  • Roger D. Williams
                                    Thank you for your answer, Wim. Several things are clearer to me now. The fact that the left- and right-eye images are inevitably not quite synchronized when
                                    Message 17 of 28 , Apr 3, 2010
                                    • 0 Attachment
                                      Thank you for your answer, Wim. Several things are clearer to me now.

                                      The fact that the left- and right-eye images are inevitably not quite
                                      synchronized when shooting with a single, rotating lens, had quite
                                      escaped me.

                                      For years I used a Voyageur rotary camera using 120/220 film to make
                                      cylindrical panoramas, and I recall having problems at the (single)
                                      seam because of light changes in the seconds it took for the camera
                                      to make a single rotation. But that this would sometimes be true for
                                      a stereo panorama made from shots taken by a rotating single lens
                                      had also escaped me. Obviously I have a lot to learn.

                                      I appreciate the fact that you see the potential advantages of
                                      cylindrical panoramas in the stereo context... It seems sensible to
                                      me to concentrate resolution in areas where there is likely to be
                                      much of visual interest, rather than waste on sky or grass.

                                      Thanks again...

                                      I wish I could be in Tucson. Unfortunately I will not even be able to
                                      get to England this summer...

                                      Roger W.

                                      --
                                      Business: www.adex-japan.com
                                      Pleasure: www.usefilm.com/member/roger
                                    • Wim Koornneef
                                      Hello Roger, What a pity that you can t attend the meeting in Plymouth, I was hoping to meet you there in person. If there is any change that there will be a
                                      Message 18 of 28 , Apr 3, 2010
                                      • 0 Attachment
                                        Hello Roger,

                                        What a pity that you can't attend the meeting in Plymouth, I was
                                        hoping to meet you there in person.
                                        If there is any change that there will be a change of plans then that
                                        would be great.

                                        Best,
                                        Wim

                                        ..
                                        > I wish I could be in Tucson. Unfortunately I will not even be able to
                                        > get to England this summer...
                                        ..

                                        --
                                        View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1750072.html
                                        Sent from the PanoToolsNG mailing list archive at Nabble.com.


                                        [Non-text portions of this message have been removed]
                                      • erik_leeman
                                        That s truly unfortunate! I am far from certain that I will be able to attend the meeting in England myself, but the opportunity to meet you in person really
                                        Message 19 of 28 , Apr 3, 2010
                                        • 0 Attachment
                                          That's truly unfortunate!
                                          I am far from certain that I will be able to attend the meeting in England myself, but the opportunity to meet you in person really was something to look forward to : (

                                          Erik Leeman

                                          --- In PanoToolsNG@yahoogroups.com, "Roger D. Williams" wrote:
                                          > ...snip...
                                          > Unfortunately I will not even be able to get to England this summer...
                                          >
                                          > Roger W.
                                        • Roger D. Williams
                                          On Sat, 03 Apr 2010 21:15:03 +0900, Wim Koornneef ... Well it s not absolutely certain, but a major project that I always handle myself promises to land on my
                                          Message 20 of 28 , Apr 3, 2010
                                          • 0 Attachment
                                            On Sat, 03 Apr 2010 21:15:03 +0900, Wim Koornneef
                                            <wim.koornneef@...> wrote:

                                            >
                                            > Hello Roger,
                                            >
                                            > What a pity that you can't attend the meeting in Plymouth, I was
                                            > hoping to meet you there in person.
                                            > If there is any change that there will be a change of plans then that
                                            > would be great.

                                            Well it's not absolutely certain, but a major project that I always handle
                                            myself promises to land on my desk shortly before the Plymouth meeting. I
                                            am checking with the guy responsible right now, but the prospect of a
                                            change are dim. I, too, was looking forward to meeting you and many others
                                            who are just names to me at the moment... I would also like to introduce
                                            my young wife...

                                            Roger W.

                                            --
                                            Business: www.adex-japan.com
                                            Pleasure: www.usefilm.com/member/roger
                                          • Trausti Hraunfjord
                                            Everyone who thought Google 3D SV was april fools joke... the joke is upon us. It is real enough. Brought to my attention by Freddy Discocandy Stapersma:
                                            Message 21 of 28 , Apr 3, 2010
                                            • 0 Attachment
                                              Everyone who thought Google 3D SV was april fools joke... the joke is upon
                                              us. It is real enough.

                                              Brought to my attention by Freddy "Discocandy" Stapersma:

                                              http://www.viasana.nl/Basis.aspx?Tid=3838&Sid=3839&Hmi=3839&Smi=0&Lid=266&Lit=TEKST&STIJL=1

                                              ..... unless this is a bigger and better prepared joke that got extended?

                                              Click the guy with the glasses in the screen to switch between 3D and non
                                              3D.

                                              Trausti


                                              [Non-text portions of this message have been removed]
                                            • Wim Koornneef
                                              I just got a report from someone who wanted to see GSV 3D yesterday afternoon and 3D wasn t available anymore. I just checked and indeed the little yellow
                                              Message 22 of 28 , Apr 8, 2010
                                              • 0 Attachment
                                                I just got a report from someone who wanted to see GSV 3D yesterday afternoon
                                                and 3D wasn't available anymore. I just checked and indeed the little yellow
                                                fellow with his red/cyan viewer is gone from the GSV example that I used for
                                                my posting on PanotoolsNG/Nabble "About the quality of Google Street View
                                                3D"

                                                Perhaps the folks of GSV read the posting ;-)

                                                BTW, I appreciate the efforts of Google because if they implement GSV 3D
                                                again, but of course with a better quality, it will give a swing forwards to
                                                the acceptance by the public of 3D panos and we can all benefit from that.

                                                Wim

                                                --
                                                View this message in context: http://n4.nabble.com/GSV-in-3D-tp1747596p1773877.html
                                                Sent from the PanoToolsNG mailing list archive at Nabble.com.
                                              Your message has been successfully submitted and would be delivered to recipients shortly.