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

Re: Getting started with QTVRs

Expand Messages
  • fierodeval
    Hi Erik, Computationally it s more efficient to use the cubic format, because it is needed only to build internally a cube with a few triangles. This fact is
    Message 1 of 29 , Sep 27, 2007
      Hi Erik,

      Computationally it's more efficient to use the cubic format, because
      it is needed only to build internally a cube with a few triangles.
      This fact is important: the cubic format does not need any distortion,
      for this reason the viewer can use very few polygons to draw every
      face. Equirectangular format requires to deform the image, this
      distortion is more required in zenit and nadir, because in the image
      equirect these parts are very stretched.

      With equirectangular format, the viewer needs to deform the image,
      this distortion is more required in zenit and nadir, because in the
      equirect image these parts are very stretched. It is needed a lot of
      polygons to deform the image to avoid strange effects as straight
      lines broken in zig-zag.

      You can see in this image the number of polygons used in DevalVR, and
      the very high number of polygons used in the nadir (and zenit) of
      equirect, for this reason the player is slower here (when this is in
      software mode): http://www.devalvr.com/fiero/cubic_equirect_triangles.jpg

      When hardware acceleration is used, the viewer always plays smoothly
      (very fast), because for a modern video cards is the same to draw 10
      polygons than 10000 polygons. For hardware acceleration cubic format
      is more efficient too, because equirect requires more VRAM to draw
      zenit and nadir. You can see in DevalVR how VRAM increases when you
      look at nadir.

      regards!
      fiero




      --- In PanoToolsNG@yahoogroups.com, "Erik Krause" <erik.krause@...> wrote:
      >
      > On Thursday, September 27, 2007 at 9:04, Yuval Levy wrote:
      >
      > > I recall Ken Turkowski, the inventor
      > > of QTVR, mentioning the reason for the choice of the cubic format
      at an
      > > online chat about the FreePV GSoC project.
      > >
      > > I don't remember the exact mathematical term, but it is about
      avoiding
      > > exceptional situation (zenith and nadir).
      > >
      > > Ken will most likely be at the GSoC mentors summit, I will ask him.
      >
      > Would be interesting, yes. However, I don't believe that Helmut
      > Dersch choose a format that was not optimal.
      >
      > > > Cubic format has no advantages over equirectangular though.
      > >
      > > Wrong. While the advantages of equirect cited are right, cubic has an
      > > advantage in that the computation of the interpolation is more
      uniform.
      >
      > This is an issue for the developers. Would be interesting to have
      > Fulvio, Helmut or Fiero comment on this.
      >
      > > For example Shockwave turns automatically off anti-aliasing for
      > > equirects (resulting in poorer interpolation) because if
      anti-aliasing
      > > was on, pointing to the zenith or nadir area would massively reduce
      > > performance.
      >
      > Most better players use a different interpolation for panning anyway.
      > For static display the interpolation can be as good as possible. It
      > is true that both FSPViewer and DevalVR pan slower if Zenith or Nadir
      > is inside the viewing window if using equirects.
      >
      > With Deval there is a direct comparison possible, because it reads
      > QTVRs as well, and here the cubics rotate indeed faster for the nadir
      > and zenith. On the other hand: Who does ever look down? When I
      > published my Verdon panorama I had to tell almost anyone to look down
      > explicitely. And who but speed testers ever rotate the nadir ;-)
      >
      > best regards
      > --
      > http://www.erik-krause.de
      >
    • Yuval Levy
      ... Hi Fiero, thank you for sharing these very interesting details. The image says more than a thousands words and to me it explains the choice of the cubic
      Message 2 of 29 , Sep 27, 2007
        fierodeval wrote:
        > Computationally it's more efficient to use the cubic format
        ...
        > http://www.devalvr.com/fiero/cubic_equirect_triangles.jpg

        Hi Fiero,

        thank you for sharing these very interesting details. The image says
        more than a thousands words and to me it explains the choice of the
        cubic format at the time of designing QTVR (when GPU acceleration was
        neither so common nor so powerful as it is today).

        Yuv
      • Erik Krause
        ... Ok, this shows to me that I don t know anything about viewer technology and that I hence shouldn t argue about pros and cons of an image format. I had a
        Message 3 of 29 , Sep 28, 2007
          On Thursday, September 27, 2007 at 22:57, fierodeval wrote:

          > You can see in this image the number of polygons used in DevalVR, and
          > the very high number of polygons used in the nadir (and zenit) of
          > equirect, for this reason the player is slower here (when this is in
          > software mode): http://www.devalvr.com/fiero/cubic_equirect_triangles.jpg

          Ok, this shows to me that I don't know anything about viewer
          technology and that I hence shouldn't argue about pros and cons of an
          image format.

          I had a clue that polygons are used in hardware accelerated graphics
          cards, and it seems logical that they are used in pano viewers if
          there is the possibility of using hardware acceleration. But what
          about f.e. PTViewer: Does it use polygons as well? I searched the
          source code and didn't find any occurrence of "polygon"...

          best regards
          --
          http://www.erik-krause.de
        • hd_de_2000
          These format discussions pop up periodically and you can find a lot of info in the qtvr-forum-archives and other panorama related lists. Briefly: - Cylinder is
          Message 4 of 29 , Sep 28, 2007
            These format discussions pop up periodically and you can find a lot of
            info in the qtvr-forum-archives and other panorama related lists.
            Briefly:

            - Cylinder is nice for printing. Vertical proportions are "correct".
            Disadvantage: No full sphere.
            - Sphere (equirectangular) is good for editing: You can work on the
            whole scene at once in Photoshop. Disadvantage: Viewer requires fast
            trig-functions, aliasing problems at nadir/zenith.
            - Cube is good for interactive viewing: you can make use of hardware
            acceleration. No trig-functions are required, only the fundamental
            arithmetic operations (affine transformations).
            - There are more formats (ipix, ...) which are no longer used.

            Historically, all formats are much older than PanoTools and even QTVR.
            Eg the cubic format was implemented in VRML, and the old venerable
            Cosmoviewer is probably the most widely distributed cubic-viewer in
            the world. It was bundled with Windows 95 and 98, don't know about
            later versions. Equirectangular and cubic viewers were available from
            LivePicture and Smoothmove, both litigated by Ipix, before PTViewer
            and QTVR-cubes, etc. Btw: the (now revoked in EU) ipix patent claims
            were independent of panorama format and mainly refer to the creation
            of the panorama by a wide angle lens.

            As to PTViewer and its derivatives: It is fastest with equirectangular
            images. It could be faster in cubic mode but isn't because it (a)
            doesn't make use of hardware acceleration and (b) uses a flexible but
            inefficient way to decode the cubic faces. This is reversed in
            PTViewerME (PTViewer for cellphones).

            Regards

            Helmut Dersch

            http://www.fh-furtwangen.de/~dersch




            --- In PanoToolsNG@yahoogroups.com, Yuval Levy <yahoo06@...> wrote:
            >
            > Erik Krause wrote:
            > > However, there are historcal reasons for the equirectangular format,
            > > too: When panotools architecture was created cubic QTVR was not yet
            > > released. This started with version 5.x in 2001, probably as a
            > > reaction on the ipix patent claim on spherical images. Panotools is
            > > much older.
            >
            > Not a reaction to a patent claim. I recall Ken Turkowski, the inventor
            > of QTVR, mentioning the reason for the choice of the cubic format at an
            > online chat about the FreePV GSoC project.
            >
            > I don't remember the exact mathematical term, but it is about avoiding
            > exceptional situation (zenith and nadir).
            >
            > Ken will most likely be at the GSoC mentors summit, I will ask him.
            >
            >
            > > Cubic format has no advantages over equirectangular though.
            >
            > Wrong. While the advantages of equirect cited are right, cubic has an
            > advantage in that the computation of the interpolation is more uniform.
            > For example Shockwave turns automatically off anti-aliasing for
            > equirects (resulting in poorer interpolation) because if anti-aliasing
            > was on, pointing to the zenith or nadir area would massively reduce
            > performance.
            >
            > Yuv
            >
          • fierodeval
            Hi Erik, Forget hardware acceleration for this discussion, maybe this feature is confusing. I think two types of viewers exist. Viewers that draw the image
            Message 5 of 29 , Sep 28, 2007
              Hi Erik,

              Forget hardware acceleration for this discussion, maybe this feature
              is confusing.

              I think two types of viewers exist. Viewers that draw the image using
              image transformations and interpolations (I don't know how to explain
              this better), like PTviewer and FSPviewer; and 3D viewers that draw
              the image using 3D polygons (triangles), like QuickTime, DevalVR,
              SPI-V, FreePV, PangeaVR, Inmervision? and Flash viewers.

              In the past, Ipix used equirects, and I think its viewer use the same
              image transformations than PTviewer, not 3D polygons.

              3D viewers use textured triangles to represent the image. Only it's
              required to create a 3D model composed by triangles with the form of
              the projection, a cube for cubic, a sphere for equirect, a cylinder
              for cylindrical.

              For a 3D engine like QuickTime (remember, QuickTime does not use
              hardware acceleration), it's more efficient to use a cube, because a
              cube does not require any deformation, you see all cube faces as it's,
              squared. You can build the cube with only 2 triangles in each face and
              the quality will be perfect. This is a good reason to choose cubic
              format in QT5. Nevertheless, if you build a 3D sphere with few
              triangles, you can see the result:
              http://www.devalvr.com/fiero/equirect_few_polygons.jpg

              Of course, I'm speaking in the case that the viewer does not modify
              the projection when image is loaded.


              Hardware accelerated viewers, like SPI-V, FreePV, PangeaVR or DevalVR
              are apart, because they could show any number of triangles without
              problems.

              regards!
              fiero


              --- In PanoToolsNG@yahoogroups.com, "Erik Krause" <erik.krause@...> wrote:
              >
              > On Thursday, September 27, 2007 at 22:57, fierodeval wrote:
              >
              > > You can see in this image the number of polygons used in DevalVR, and
              > > the very high number of polygons used in the nadir (and zenit) of
              > > equirect, for this reason the player is slower here (when this is in
              > > software mode):
              http://www.devalvr.com/fiero/cubic_equirect_triangles.jpg
              >
              > Ok, this shows to me that I don't know anything about viewer
              > technology and that I hence shouldn't argue about pros and cons of an
              > image format.
              >
              > I had a clue that polygons are used in hardware accelerated graphics
              > cards, and it seems logical that they are used in pano viewers if
              > there is the possibility of using hardware acceleration. But what
              > about f.e. PTViewer: Does it use polygons as well? I searched the
              > source code and didn't find any occurrence of "polygon"...
              >
              > best regards
              > --
              > http://www.erik-krause.de
              >
            • Serge Maandag (yahoo)
              ... Thanks for the explanations and very clear photographic examples, Fiero. And I d also like to say thank you bigtime! for your stand alone viewer. I use
              Message 6 of 29 , Sep 28, 2007
                > Hardware accelerated viewers, like SPI-V, FreePV, PangeaVR or DevalVR
                > are apart, because they could show any number of triangles without
                > problems.

                Thanks for the explanations and very clear photographic examples, Fiero.

                And I'd also like to say "thank you bigtime!" for your stand alone viewer.
                I use it in hardware acceleration mode exclusively to view my panoramas.

                It's so smooth, responsive and easy to use, no other viewer comes close to it.
                The down side to it is that I've gotten used to it. Nowadays it hurts to see a
                stuttering QTVR hop around in my browser.. :)

                Serge.
              • Erik Krause
                ... That is exactly what I suspected. F.e. Immervision is relatively fast, even in fullscreen, but only if you don t set the quality parameter higher than 100.
                Message 7 of 29 , Sep 28, 2007
                  On Friday, September 28, 2007 at 10:32, fierodeval wrote:

                  > I think two types of viewers exist. Viewers that draw the image using
                  > image transformations and interpolations (I don't know how to explain
                  > this better), like PTviewer and FSPviewer; and 3D viewers that draw
                  > the image using 3D polygons (triangles), like QuickTime, DevalVR,
                  > SPI-V, FreePV, PangeaVR, Inmervision? and Flash viewers.

                  That is exactly what I suspected.

                  F.e. Immervision is relatively fast, even in fullscreen, but only if
                  you don't set the quality parameter higher than 100. But in this case
                  it produces exactly the kind of deformations you show in
                  http://www.devalvr.com/fiero/equirect_few_polygons.jpg
                  If you increase the quality parameter it gets much slower - perhaps
                  slower than PTViewer. This is most likely due to the increased number
                  of polygons.

                  Many thanks for that information.

                  best regards
                  --
                  http://www.erik-krause.de
                • Yuval Levy
                  ... the same happens with flashpanoramas too Yuv
                  Message 8 of 29 , Sep 28, 2007
                    Erik Krause wrote:
                    > If you increase the quality parameter it gets much slower - perhaps
                    > slower than PTViewer. This is most likely due to the increased number
                    > of polygons.

                    the same happens with flashpanoramas too

                    Yuv
                  • ahoeben41
                    ... To put it in your words: Wrong. First, it is not Shockwave doing this, it is SPi-V. Second, and more importantly, the reason SPi-V does this has nothing
                    Message 9 of 29 , Sep 28, 2007
                      > For example Shockwave turns automatically off anti-aliasing for
                      > equirects (resulting in poorer interpolation) because if anti-aliasing
                      > was on, pointing to the zenith or nadir area would massively reduce
                      > performance.

                      To put it in your words: Wrong.

                      First, it is not Shockwave doing this, it is SPi-V. Second, and more
                      importantly, the reason SPi-V does this has nothing whatsoever to do
                      with performance; When bilinear filtering ('anti-aliasing') is turned
                      on for the spherical mesh, there are rendering artifacts at the zenith
                      and nadir because the vertices are 'crumpled' together and mipmapping
                      kicks in in a wrong way.
                    • Yuval Levy
                      ... thank you for reminding me of the nitty gritty details you already shared with me in the past and that I will most likely forget again until the image
                      Message 10 of 29 , Sep 28, 2007
                        ahoeben41 wrote:
                        > mipmapping kicks in in a wrong way.

                        thank you for reminding me of the nitty gritty details you already
                        shared with me in the past and that I will most likely forget again
                        until the image format discussion comes up again.

                        rinse, repeat.

                        Yuv
                      • ptgroup
                        Erik: I set my Quality Parameter in Immervision to 200 and my resolution to 3072 x 1536. Seems to me fast- much faster than PTViewer. They wrote s.th. about
                        Message 11 of 29 , Sep 28, 2007
                          Erik:
                          I set my "Quality Parameter" in Immervision to 200 and my
                          resolution to 3072 x 1536. Seems to me fast- much faster than
                          PTViewer.
                          They wrote s.th. about additional settings in their forum - this I did with
                          good results.
                          Not all of my panos are set to this requirement by now - but alls new ones.

                          Ciao
                          Mike
                          -------------------------


                          -----Urspr�ngliche Nachricht-----
                          Von: PanoToolsNG@yahoogroups.com [mailto:PanoToolsNG@yahoogroups.com]Im Auftrag von Erik Krause
                          Gesendet: Freitag, 28. September 2007 14:38
                          An: PanoToolsNG@yahoogroups.com
                          Betreff: [PanoToolsNG] Re: Viewer technology (was: Getting started with QTVRs)


                          On Friday, September 28, 2007 at 10:32, fierodeval wrote:

                          > I think two types of viewers exist. Viewers that draw the image using
                          > image transformations and interpolations (I don't know how to explain
                          > this better), like PTviewer and FSPviewer; and 3D viewers that draw
                          > the image using 3D polygons (triangles), like QuickTime, DevalVR,
                          > SPI-V, FreePV, PangeaVR, Inmervision? and Flash viewers.

                          That is exactly what I suspected.

                          F.e. Immervision is relatively fast, even in fullscreen, but only if
                          you don't set the quality parameter higher than 100. But in this case
                          it produces exactly the kind of deformations you show in
                          http://www.devalvr.com/fiero/equirect_few_polygons.jpg
                          If you increase the quality parameter it gets much slower - perhaps
                          slower than PTViewer. This is most likely due to the increased number
                          of polygons.

                          Many thanks for that information.

                          best regards
                          --
                          http://www.erik-krause.de






                          [Non-text portions of this message have been removed]
                        • Erik Krause
                          ... Have a look at http://www.erik-krause.de/immervision.jpg This is a screenshot of the upper right corner of an immervision player window. Quality was set to
                          Message 12 of 29 , Sep 28, 2007
                            On Friday, September 28, 2007 at 17:39, ptgroup wrote:

                            > I set my "Quality Parameter" in Immervision to 200 and my
                            > resolution to 3072 x 1536. Seems to me fast- much faster than
                            > PTViewer.

                            Have a look at http://www.erik-krause.de/immervision.jpg This is a
                            screenshot of the upper right corner of an immervision player window.
                            Quality was set to 200. PTViewer doesn't show this kind of artifacts,
                            the structure is straight there...

                            best regards


                            --
                            http://www.erik-krause.de
                          • hd_de_2000
                            ... Wrong: Drawing polygons is nothing else than image transformations and interpolations . PTViewer in cubic mode also draws polygons. Quicktime in cylinder
                            Message 13 of 29 , Sep 28, 2007
                              --- In PanoToolsNG@yahoogroups.com, "fierodeval" <fierodeval@...> wrote:
                              >
                              >
                              > I think two types of viewers exist. Viewers that draw the image using
                              > image transformations and interpolations (I don't know how to explain
                              > this better), like PTviewer and FSPviewer; and 3D viewers that draw
                              > the image using 3D polygons (triangles), like QuickTime, DevalVR,
                              > SPI-V, FreePV, PangeaVR, Inmervision? and Flash viewers.
                              >

                              Wrong: Drawing polygons is nothing else than
                              "image transformations and interpolations". PTViewer in cubic mode
                              also draws polygons. Quicktime in cylinder mode does not draw polygons.

                              The difference you are talking about exists between viewers for the
                              equirectangular format: Some (eg PTViewer, but also the older viewers
                              like Livepicture) apply the spherical transformation to each point in
                              the image, while others (the ones using polygons) apply the
                              transformation only to a few points which are the corners of the
                              triangles. These viewers then interpolate the region in between using
                              straight lines. This is faster, but less accurate ( like building a
                              sphere using flat surfaces).

                              > In the past, Ipix used equirects, and I think its viewer use the same
                              > image transformations than PTviewer,

                              ipix did not use equirectangular format. Their panoramas consisted of
                              two hemispheres.

                              Regards

                              Helmut Dersch
                            • ptgroup
                              What res is the image ? As said : Immervision runs best with multiples of 1024- so the forum says. I made some Immervision testing some time ago at:
                              Message 14 of 29 , Sep 28, 2007
                                What res is the image ?
                                As said : Immervision runs best with multiples of 1024- so the forum says.
                                I made some Immervision testing some time ago at:
                                http://www.netzserver1.de/ivtest/

                                Quality means PS CS2 quality setting.
                                Ciao
                                Mike
                                -------------------------


                                -----Urspr�ngliche Nachricht-----
                                Von: PanoToolsNG@yahoogroups.com [mailto:PanoToolsNG@yahoogroups.com]Im Auftrag von Erik Krause
                                Gesendet: Freitag, 28. September 2007 19:01
                                An: PanoToolsNG@yahoogroups.com
                                Betreff: [PanoToolsNG] Re: Viewer technology (was: Getting started with QTVRs)


                                On Friday, September 28, 2007 at 17:39, ptgroup wrote:

                                > I set my "Quality Parameter" in Immervision to 200 and my
                                > resolution to 3072 x 1536. Seems to me fast- much faster than
                                > PTViewer.

                                Have a look at http://www.erik-krause.de/immervision.jpg This is a
                                screenshot of the upper right corner of an immervision player window.
                                Quality was set to 200. PTViewer doesn't show this kind of artifacts,
                                the structure is straight there...

                                best regards

                                --
                                http://www.erik-krause.de






                                [Non-text portions of this message have been removed]
                              • Fulvio Senore
                                ... Other people have already talked about cubic and spherical file format. In short, cubic format is easier to handle (requires less computations) and a cubic
                                Message 15 of 29 , Sep 28, 2007
                                  Erik Krause ha scritto:
                                  > I had a clue that polygons are used in hardware accelerated graphics
                                  > cards, and it seems logical that they are used in pano viewers if
                                  > there is the possibility of using hardware acceleration. But what
                                  > about f.e. PTViewer: Does it use polygons as well? I searched the
                                  > source code and didn't find any occurrence of "polygon"..
                                  Other people have already talked about cubic and spherical file format.
                                  In short, cubic format is easier to handle (requires less computations)
                                  and a cubic image is handled in the same way for every viewing direction.
                                  An equirectangular image requires more computations and it has problems
                                  at the zenith and at the nadir. Near those points the image is very
                                  heavily distorted (the zenith is a single point in the viewer but it
                                  expands to the full upper border of the equirectangular image, same for
                                  the nadir) so there is less room for optimization.

                                  About polygons, older versions of PTViewer (up to 2.5) did not use them
                                  at all. Each point was computed with the full mathematical transform so
                                  the viewer was rather slow (to say the truth the same transformation was
                                  shared by two pixels, but it is not a great difference).
                                  PTViewer 2.8 uses an approach similar to polygons, dividing the viewer
                                  in squares and computing the full transformation only for the four
                                  vertices of the squares. For internal points the transform is computed
                                  in a faster way: if the image is not very distorted the error is minimal.
                                  The less the image is distorted, the larger can be the squares, the
                                  faster is the viewer. Near the horizon it is possible to use larger
                                  squares, at the poles the image is very distorted so it is only possible
                                  to use progressively smaller squares and the viewer slows down. The same
                                  holds for FSPViewer.

                                  Fulvio Senore
                                • Erik Krause
                                  ... It was 6000x3000 but I repeated the test with 4096x2048 with same result. ... The source image size surely dos not change the used geometry. ... Here is a
                                  Message 16 of 29 , Sep 28, 2007
                                    On Friday, September 28, 2007 at 19:16, ptgroup wrote:

                                    > What res is the image ?

                                    It was 6000x3000 but I repeated the test with 4096x2048 with same
                                    result.

                                    > As said : Immervision runs best with multiples of 1024- so the forum says.

                                    The source image size surely dos not change the used geometry.

                                    > I made some Immervision testing some time ago at:
                                    > http://www.netzserver1.de/ivtest/

                                    Here is a screenshot of your
                                    http://www.netzserver1.de/ivtest/html/4096-10.html
                                    example:
                                    http://www.erik-krause.de/immervision2.jpg

                                    best regards
                                    --
                                    http://www.erik-krause.de
                                  • ptgroup
                                    To me, the DEVALVR is better - but these erratics are there also. Ciao Mike ... Von: PanoToolsNG@yahoogroups.com [mailto:PanoToolsNG@yahoogroups.com]Im Auftrag
                                    Message 17 of 29 , Sep 28, 2007
                                      To me, the DEVALVR is better - but these erratics are there also.
                                      Ciao
                                      Mike
                                      -------------------------


                                      -----Urspr�ngliche Nachricht-----
                                      Von: PanoToolsNG@yahoogroups.com [mailto:PanoToolsNG@yahoogroups.com]Im Auftrag von Erik Krause
                                      Gesendet: Freitag, 28. September 2007 21:27
                                      An: PanoToolsNG@yahoogroups.com
                                      Betreff: [PanoToolsNG] Re: Viewer technology (was: Getting started with QTVRs)


                                      On Friday, September 28, 2007 at 19:16, ptgroup wrote:

                                      > What res is the image ?

                                      It was 6000x3000 but I repeated the test with 4096x2048 with same
                                      result.

                                      > As said : Immervision runs best with multiples of 1024- so the forum says.

                                      The source image size surely dos not change the used geometry.

                                      > I made some Immervision testing some time ago at:
                                      > http://www.netzserver1.de/ivtest/

                                      Here is a screenshot of your
                                      http://www.netzserver1.de/ivtest/html/4096-10.html
                                      example:
                                      http://www.erik-krause.de/immervision2.jpg

                                      best regards
                                      --
                                      http://www.erik-krause.de






                                      [Non-text portions of this message have been removed]
                                    • Yili Zhao
                                      Hi Helmut, ... Could PTViewer also make use of hardware acceleration in cubic mode? Because Java Applet is also can use OpenGL for hardware acceleration
                                      Message 18 of 29 , Sep 29, 2007
                                        Hi Helmut,
                                        Helmut wrote:
                                        > As to PTViewer and its derivatives: It is fastest with equirectangular
                                        > images. It could be faster in cubic mode but isn't because it (a)
                                        > doesn't make use of hardware acceleration and (b) uses a flexible but
                                        > inefficient way to decode the cubic faces. This is reversed in
                                        > PTViewerME (PTViewer for cellphones).

                                        Could PTViewer also make use of hardware acceleration in cubic mode?
                                        Because Java Applet is also can use OpenGL for hardware acceleration
                                        through JOGL ( Java Binding for the OpenGL API,
                                        https://jogl.dev.java.net/).
                                        --
                                        Yili Zhao
                                      • hd_de_2000
                                        ... Hi Yili Zhao this is probably quite easy to implement, especially if you start with PTViewerME. This version is based on m3g which is basically a subset of
                                        Message 19 of 29 , Sep 29, 2007
                                          --- In PanoToolsNG@yahoogroups.com, "Yili Zhao" <panovr@...> wrote:
                                          >
                                          >
                                          > Could PTViewer also make use of hardware acceleration in cubic mode?
                                          > Because Java Applet is also can use OpenGL for hardware acceleration
                                          > through JOGL ( Java Binding for the OpenGL API,
                                          > https://jogl.dev.java.net/).
                                          > --

                                          Hi Yili Zhao

                                          this is probably quite easy to implement, especially if you start
                                          with PTViewerME. This version is based on m3g which is basically
                                          a subset of opengl.

                                          Regards

                                          Helmut Dersch
                                        • yi_li_zhao
                                          ... mode? ... acceleration ... Hi Helmut, as considering M3G which is based on Scene Graph structure above low level OpenGL API, I think maybe Java3D
                                          Message 20 of 29 , Sep 30, 2007
                                            --- In PanoToolsNG@yahoogroups.com, "hd_de_2000" <der@...> wrote:
                                            >
                                            > --- In PanoToolsNG@yahoogroups.com, "Yili Zhao" <panovr@> wrote:
                                            > >
                                            > >
                                            > > Could PTViewer also make use of hardware acceleration in cubic
                                            mode?
                                            > > Because Java Applet is also can use OpenGL for hardware
                                            acceleration
                                            > > through JOGL ( Java Binding for the OpenGL API,
                                            > > https://jogl.dev.java.net/).
                                            > > --
                                            >
                                            > Hi Yili Zhao
                                            >
                                            > this is probably quite easy to implement, especially if you start
                                            > with PTViewerME. This version is based on m3g which is basically
                                            > a subset of opengl.
                                            >
                                            > Regards
                                            >
                                            > Helmut Dersch
                                            >
                                            Hi Helmut,
                                            as considering M3G which is based on "Scene Graph" structure above
                                            low level OpenGL API, I think maybe Java3D
                                            (https://java3d.dev.java.net, is also Scene Graph based) is more
                                            suitable. And thanks for the release of PTViewerME.

                                            Sincerely
                                            Yili Zhao
                                          Your message has been successfully submitted and would be delivered to recipients shortly.