Hi Robert,

I am back from my holidays, now trying to sketch new RayIntersector according 
to our discussion. Sorry, the topic seems difficult:

- RayIntersector (start and direction) is good and easy to understand by 
users. Finally, using it in CameraManipulators would make Manipulators 
compatible with shadow volumes and would make them robust for any homogeneous 
transformation. Potential problems: Somebody might want to limit the length of 
the ray. Solution - use LineSegmentIntersector, but this one is not an option 
with shadow volumes as it is not robust. Solution 2 - maybe, we can append 
length class member that would specify the length of the ray (in the current 
coordinate frame) with the default value of +INF. 

Does it seems a good solution for you?
Or you prefer to not have the length member? (This would force the people to 
use LineSegmentIntersector, probably.)

I still think about LineSegmentIntersector, as it will became unusable for me, 
as it is not robust against all types of transformations that can appear in 
the scene graph. In the presence of homogeneous transformations, line with one 
point at infinity is just a special case of line for me. The same with point, 
it is just a special case of line. Just thinking if LineSegmentIntersector can 
be made robust for such cases. Maybe LineSegmentIntersectorRobust, or 
LineSegmentIntersector4, or some other approach, or we can leave the problem 
without solving it.


Quick question: For very long time, I do not understand following line of 
LineSegmentIntersector:
        inline Intersections& getIntersections() { return _parent ?
                    _parent->_intersections : _intersections; }

It returns the active list of intersections (used for appending and 
intersection retrieval). If there is stack of Intersectors, the intersections 
should be put into the most bottom one, isn't it? Why are they put to the 
second most top intersector, that will be popped and intersections destroyed 
before the end of traversal? Please, what am I missing on this point?

John
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to