Hi Alex,

I've done a review of your changes and don't feel that the approach
you have taken is appropriate, first up the default size quadruples
the number of vertices per tile, and secondly the accounting for
curvature seems based on some that roughly works for your purposes but
for what I can make out isn't based on any actual error metric.

What I would have thought that would be appropriate would be to have
an error metric by tested against the vertical offsets that the
faceted mesh would have from the curved surface of the ellipsoid.  The
maximum error would be found at the mid of each triangle on the mesh,
the smaller the triangles the smaller the error.  You could probably
simplify it so just compute the error based on separate lat/long
divisions for the number of columns and rows.  The error metric would
be best done as a pixel error, and have the LOD range and a default
screen resolution (say 1280x1024) used as the basis to compute the
acceptable error in world coordinates, from this one could then
compute the lat/long delta which is the largest tolerable delta, then
from this compute the number of rows and columns.

Robert.

On Tue, Nov 16, 2010 at 5:16 PM, Alexander Irion <[email protected]> wrote:
> Hi!
>
> I created a dataset with osgdem, where the whole earth is covered by 
> textures, but height values are only given for one country.
>
> One half of the whole earth was looking very "polygonal", while the other 
> half - where the height values are given - was looking round. This happened 
> on level 1, while level 0 was looking round on both sides. Deeper levels were 
> also looking good again.
>
> I found the reason to reside in the method 
> DestinationTile::optimizeResolution(), which computed the value 8 for 
> numColumns and numRows for tile L1_X0_Y0, while L1_X1_Y0 is generated with 32 
> x 32 colums and rows.
>
> The method did not take the angular extends into account for computing the 
> numbers of colums and rows.
>
> Please find the attached fix.
>
> --Alex
>
> ------------------
> Read this topic online here:
> http://forum.openscenegraph.org/viewtopic.php?p=33782#33782
>
>
>
>
> Attachments:
> http://forum.openscenegraph.org//files/destination_909.zip
>
>
> _______________________________________________
> osg-submissions mailing list
> [email protected]
> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
>
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to