No worries - sorted. Wasn't calling dirtyBound() in the right places... Sorry for the waste of bandwidth.
David On 20/12/2007, David Spilling <[EMAIL PROTECTED]> wrote: > > Dear OSGers, > > I'm kind of stumped and I was hoping that somebody might tell me where to > look... > > I have a HUD (simple geometry underneath a Ortho projection node > underneath a Post-render camera) exactly as per the osgpick example . I pick > exactly as per osgpick, although I parse mouse button events (push, release, > drag) rather than FRAME events. I have several "windows", which consist of a > matrix transform, and various bits of geometry - rectangles, text etc. - > underneath. I have code that enables DRAG to move the window by updating the > transform. Some of the window elements occasionally are changed ( e.g. > text updated etc.). > > Now the problem : usually, view->computeIntersections correctly returns a > number of hits based on what is under the mouse pointed. However, if nothing > is changing in the window, then view->computeIntersections() _usually_ fails > to intersect with _any_ of the underlying geometry of that window. > > I say usually because to start with, all is fine, but after a few seconds > of moving windows around, the intersection visitor no longer returns hits > from them. > > If any geometry in the window changes, then view->computeIntersections > will then return valid geometry intersections for a while, until it stops > returning anything. The geometry is still there, on screen, just unpickable > with a LineSegmentIntersector. > > What is particularly wierd is the fact that it works for a while and then > stops working. I have tried to work my way through the > LineSegmentIntersector, but got lost. Is there some cache of objects for > intersection that isn't being updated? All the objects have their data > variance set as DYNAMIC, so not sure it can be that. Am I missing something > akin to calling dirty() somewhere, in order to "refresh" the intersector's > cache of objects? > > Is this a problem with LineSegmentIntersector and Ortho2D projections? I > presume not, because osgpick works... > > I can modify osgpick to try to demonstrate, if that helps. > > Has anybody seen anything like this? All pointers appreciated... > > David >
_______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org