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

Re: Gigapixels might get a new blast

Expand Messages
  • Hans
    ... A preview in PTGui fullscreen 2700 pixels is generated within just 10-20 sec for a 1 gigapixel pano. It all depends on what you start your timer from. Try
    Message 1 of 13 , Oct 31, 2010
    • 0 Attachment
      --- In PanoToolsNG@yahoogroups.com, Trausti Hraunfjord <trausti.hraunfjord@...> wrote:
      >
      > Well, it is only a "preview". What kind of a preview can be seen today in
      > seconds? That is a genuine and sincere question, since I have never done
      > any of those huge works.
      >
      > As I understand it, no preview can be made unless hours of processing have
      > been put on the clock.

      A preview in PTGui fullscreen 2700 pixels is generated within just 10-20 sec for a 1 gigapixel pano. It all depends on what you start your timer from.

      Try the speedtest pano it is 0.8 GB

      On my iMac the original 300 images take 90 sec to load but remember this is also generating full size tempfiles. The article says "By sampling only a fraction of the pixels in a massive image" I sounds like the people in UTAH just use some small tumbnails for this preview.

      I can after the images are loaded apply my template which sets all the images in place that takes 3 sec.
      When you load the preview editor I now get a perfect preview in 7 sec.
      And I just have 8 Gb Ram.

      Of course if you do not have a template you have to generate controlpoints and optimize it first. That takes me 5 minutes. Generating the final pano in 100000x80000 takes around 12 minutes on my iMac as far as I remember but the fastest ones like a MacPro with 32GB ram takes around 4 - 6.

      Hans


      >
      > Trausti
      >
      >
      > On Sat, Oct 30, 2010 at 2:23 PM, prague <360cities@...> wrote:
      >
      > >
      > >
      > >
      > > So they made a 1/3600 size preview image of a gigapixel in a few seconds...
      > > Ok :-)
      > >
      > > "It looks almost as good as the full, 3.27-gigapixel panorama, even though
      > > it contains only one-3,600th as much image data."
      > >
      > > LOL!
      > >
      > > That toy ferrari is almost as fun as driving a real one, except that it's
      > > 5cm long and has no engine. :-D
      > >
      > > Anyway, maybe it will really result in some breakthrough. But for now I
      > > can't help laughing.....
      > >
      > >
      >
      >
      > [Non-text portions of this message have been removed]
      >
    • Erik Krause
      ... Fractal is still a buzzword and it guarantees for attention especially from those who don t exactly understand what it means. However, a file format that
      Message 2 of 13 , Oct 31, 2010
      • 0 Attachment
        Am 31.10.2010 11:02, schrieb johnmyrstad:
        > The interesting part is the new (patented? )file system which lets
        > the software access image data in a fractal way and use these
        > "fractal views" in realtime.

        "Fractal" is still a buzzword and it guarantees for attention especially
        from those who don't exactly understand what it means.

        However, a file format that stores an image pyramid in a non-redundant
        way would be very good. But I doubt you need a patent for that. The old
        Kodak Photo CD format PCD had this feature and I bet there are some more
        (and more flexible) formats. But the market accepts new file formats
        very slowly if at all. See PNG or Jpeg2000...

        --
        Erik Krause
      • Joost Nieuwenhuijse
        AFAIK Tiff supports a pyramid image representation. The MrSID file format (used for satellite imagery) uses wavelet pyramids. Then there s progressive jpeg.
        Message 3 of 13 , Oct 31, 2010
        • 0 Attachment
          AFAIK Tiff supports a pyramid image representation. The MrSID file
          format (used for satellite imagery) uses wavelet pyramids. Then there's
          progressive jpeg. And Zoomify, krpano and pano2vr basically use pyramid
          images when displaying multi resolution panoramas.

          I guess there must be some novelty here but the article doesn't really
          reveal it..

          Joost

          On 31-10-2010 17:51, Erik Krause wrote:
          > Am 31.10.2010 11:02, schrieb johnmyrstad:
          >> The interesting part is the new (patented? )file system which lets
          >> the software access image data in a fractal way and use these
          >> "fractal views" in realtime.
          >
          > "Fractal" is still a buzzword and it guarantees for attention especially
          > from those who don't exactly understand what it means.
          >
          > However, a file format that stores an image pyramid in a non-redundant
          > way would be very good. But I doubt you need a patent for that. The old
          > Kodak Photo CD format PCD had this feature and I bet there are some more
          > (and more flexible) formats. But the market accepts new file formats
          > very slowly if at all. See PNG or Jpeg2000...
          >
        • johnmyrstad
          Erik: Yes, fractal is a buzzword, but thats what they say. I believe ou may read details about it the upcoming Computer Graphics (CG) Quarterly:
          Message 4 of 13 , Oct 31, 2010
          • 0 Attachment
            Erik: Yes, "fractal" is a buzzword, but thats what they say. I believe ou may read details about it the upcoming Computer Graphics (CG) Quarterly: www.siggraph.org/publications/newsletter/volume-44-number-4/

            John Myrstad
          • prague
            did they start with no known camera positions or lens parameters? or are both known already? if so, i wouldn t really call it much of a breakthrough, more like
            Message 5 of 13 , Nov 1, 2010
            • 0 Attachment
              did they start with no known camera positions or lens parameters? or are both known already? if so, i wouldn't really call it much of a breakthrough, more like a significant step forward, but not a breakthrough. creating a thumbnail preview from known camera positions/parameters isn't a huge leap forward, i guess, but those more knowledgeable than me about such stuff can correct me if i'm wrong.

              forgive me for being skeptical, but it's hard to tell what is linkbait and what is a genuine breakthrough from a lot of these "scientific" articles, which rarely give enough information for the reader to discern which is which.

              --- In PanoToolsNG@yahoogroups.com, "johnmyrstad" <johnmyrstad@...> wrote:
              >
              >
              > > "It looks almost as good as the full, 3.27-gigapixel panorama, even though it contains only one-3,600th as much image data."
              > >
              > > LOL!
              > >
              > > That toy ferrari is almost as fun as driving a real one, except that it's 5cm long and has no engine. :-D
              >
              > The interesting part is the new (patented? )file system which lets the software access image data in a fractal way and use these "fractal views" in realtime.
              >
              > This might open for real-time stitching previews as you shoot, and that would be cool when on top of a tower in Prague, would it not Martin ?
              >
            • Erik Krause
              ... Yes, but those are redundant. A part of the pixels in the larger resolution is also present in the lower one. -- Erik Krause http://www.erik-krause.de
              Message 6 of 13 , Nov 1, 2010
              • 0 Attachment
                Am 31.10.2010 18:22, schrieb Joost Nieuwenhuijse:

                > And Zoomify, krpano and pano2vr basically use pyramid
                > images when displaying multi resolution panoramas.

                Yes, but those are redundant. A part of the pixels in the larger
                resolution is also present in the lower one.

                --
                Erik Krause
                http://www.erik-krause.de
              • Erik Krause
                ... - To access this part of the site, you need to log in with your user name and password. ... -- Erik Krause http://www.erik-krause.de
                Message 7 of 13 , Nov 1, 2010
                • 0 Attachment
                  Am 31.10.2010 21:18, schrieb johnmyrstad:

                  > I believe ou may read details about it the upcoming Computer
                  > Graphics (CG) Quarterly:
                  > www.siggraph.org/publications/newsletter/volume-44-number-4/

                  -> To access this part of the site, you need to log in with your user
                  name and password.

                  :-(

                  --
                  Erik Krause
                  http://www.erik-krause.de
                Your message has been successfully submitted and would be delivered to recipients shortly.