Hi Lars,

Thanks for the fix and test example.  I've merged most of your changes
to KdTree - in particular the change in parameter Vec3d type to the
KdTree::intersect() method, and in the IntersectKdTree constructor.

I didn't find it necessary to merge the change the
IntersectKdTree::intersect method as it's internal maths is all Vec3
based so I didn't see any advantage in making this change.  Your test
example works fine without merging this.

As a more general issue, I do wonder about whether we should still
instigate a full Vec3d/Vec3 implementation where the later can be used
for speed, and the full Vec3d math version used where precision is
more critical.  There is a limit on precision though as the vertex
data on geometry is in Vec3's.

The combined changes are now checked into svn/trunk.  Could you please
test with your application/datasets to make sure that things are still
working at your end.

Cheers,
Robert.

On Tue, Jan 19, 2010 at 4:49 PM, Nilsson Lars
<[email protected]> wrote:
> Hi Robert
>
>
>
> Attached is a small program doing intersection calculations, both with and
> without KdTree. The geometry is a TRIANGLE_STRIP consisting of five
> vertices, all with the same rather high Z-value. If the intersection
> calculation uses KdTree, it fails. When I changed osg::Vec3 to osg::Vec3d in
> a few places in osg::KdTree it finds the correct intersection point.
>
>
>
> Another way to solve the problem was to increase the expansion of the
> bounding-box in osgUtil::LineSegmentIntersector::intersectAndClip.
>
>
>
> The version of OSG that I use is 2.9.6.
>
>
>
> Because I do not have enough knowledge in KdTree to understand what other
> implications these changes may have, I tried a week ago to start a
> discussion in osg-user mailing list to find out if either of these fixes
> seems to be reasonable. But there was no response. Now it is up to you to
> decide which one, or none of them to be used.
>
>
>
> Regards,
>
> Lars Nilsson
>
>
>
>
>
> _______________________________________________
> 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