Hi Edgar,

Yup, that will definitely cause a difference.  I submitted a change to
osgTerrain yesterday that takes care of that issue.  Have you had a
chance to look at those changes yet Robert?

Thanks,

Jason

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Edgar Ellis
Sent: Tuesday, May 23, 2006 4:22 P
To: osg users
Subject: Re: [osg-users] osgTerrain elevation accuracy


I tried your changes, but I'm still getting problems as well.
However I did manage to get much better results using a power of 2 + 1 
sized dem and a 65x65 tile size.

One idea why the paged database might be slightly off:
Suppose you have a tile whose extents cover from .5 to 8.5px of a dem.
When readHeightField is called for that tile, the tile will be created 
from the 0-9px of the dem.
So now your tile represents 9px instead of 8px it's extents indicate.
Is that enough to cause a noticeable difference?

Thanks,
- Edgar

Jason Beverage wrote:
> Hi Edgar,
>
> I did a quick substitution of all bounding box code in the DataSet
class with a 2D double bounding box called GeospatialExtents as Robert
suggested.  I've tried it on a few datasets here and things have worked
fine so far, but the issue I'm having with scene graph heights being off
by quite a large amount from the source data is still present.
>
> I added a bool, interpolateTerrain in readHeightField that takes an
alternate code path to attempt to prevent upsampling of the terrain by
GDAL, but my problem still exists.
>
> I figured since I'd made these changes I'd send it out to you to see
if it fixed whatever problems you've been having with the float bounding
boxes.
>
> Thanks,
>
> Jason
>
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Jason
Beverage
> Sent: Tuesday, May 09, 2006 9:10 A
> To: osg users
> Subject: RE: [osg-users] osgTerrain elevation accuracy
>
> I'll try to do this today.  I've written some code to do bilinear
> interpolation on the source terrain data, but it is still off.  Maybe
a
> combination of it and the double bounding box will solve the issues
I'm
> having.
>
> Thanks for your help,
>
> Jason
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Robert
> Osfield
> Sent: Tuesday, May 09, 2006 4:31 A
> To: osg users
> Subject: Re: [osg-users] osgTerrain elevation accuracy
>
> On 5/8/06, Edgar Ellis <[EMAIL PROTECTED]> wrote:
>   
>> I haven't had a chance deal with the bounding box precision, but as
>> Robert said using doubles should solve the problem (perhaps a
>>     
> BoundingBoxd?)
>
> A local double XY box would be more appropriate as it only uses the XY
> dimensions of the osg::BoundingBox.
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://openscenegraph.net/mailman/listinfo/osg-users
> http://www.openscenegraph.org/
>
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://openscenegraph.net/mailman/listinfo/osg-users
> http://www.openscenegraph.org/
>
------------------------------------------------------------------------
>
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://openscenegraph.net/mailman/listinfo/osg-users
> http://www.openscenegraph.org/

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

Reply via email to