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

Flash file size, thanks Hans

Expand Messages
  • Ned Chiariello
    Thanks Hans, With your help I found CubicConverter is source of the file size discrepancy. I use CubicConverter and when it creates cube faces the default
    Message 1 of 2 , Jan 21, 2011
    • 0 Attachment
      Thanks Hans,
      With your help I found CubicConverter is source of the file size discrepancy. I use CubicConverter and when it creates cube faces the default setting is 2036X2036 which increases the original 6400X3200 equirectangular image. Pano2VR's cube face size is in relation to the equirectangular image size like you said and Pano2VR duplicates whatever cube face size you import. CubicConverter will take those 2036X2036 cube faces and revert them back to 6400X3200 by default. This default back and forth image size change seems like strange behavior to me maybe you can explain. Ned
    • Hans
      ... It is a choice Thomas made for Pano2VR to not use the standard pi conversion as default. As far as I remember it was discussed at his forum long ago. In
      Message 2 of 2 , Jan 21, 2011
      • 0 Attachment
        --- In PanoToolsNG@yahoogroups.com, Ned Chiariello <panophoto@...> wrote:
        >
        > Thanks Hans,
        > With your help I found CubicConverter is source of the file size discrepancy. I use CubicConverter and when it creates cube faces the default setting is 2036X2036 which increases the original 6400X3200 equirectangular image. Pano2VR's cube face size is in relation to the equirectangular image size like you said and Pano2VR duplicates whatever cube face size you import. CubicConverter will take those 2036X2036 cube faces and revert them back to 6400X3200 by default. This default back and forth image size change seems like strange behavior to me maybe you can explain. Ned


        It is a choice Thomas made for Pano2VR to not use the standard pi conversion as default.
        As far as I remember it was discussed at his forum long ago.

        In theory the pi conversion is correct as it duplicates the exact size from the panorama.
        But it is based on the center resolution and if you use it the actual resolution in the corners will be interpolated.

        By reducing the size to 1/4 of the equirectangular width you get a smaller download with almost same quality.
        You can also say that it partly take advantage of what is called the 70% rule which Ken Turkowski long ago presented. This says that the actual maximum resolution from a Bayer sensor can be reduced to 70% of the initial camera resolution without any loss of details.

        From my experience with sensors we have today I would say that 75-80% would be a better choice.

        You can check it on this page I did a couple of years ago.
        http://www.panoramas.dk/panorama/cubeface-sizes/

        Just load the maximum size 3500 cubeface and zoom in until it matches the 100% image at the top corner.
        Then load the 2750 cubeface (78% ) and see if you can actually see any difference.
        When you have them loaded in cache you can pan around and change fast between the 2 sizes. I bet you will have very difficult to see any difference at all.

        Of course this demands that you make the smaller cubefaces from the largest equirectangular to get as much information as possible in them.


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