Jan Ciger wrote: > I was thinking about the same issue because we would like to move over > to a COLLADA-based production pipeline. What would be good enough is a > possibility to register a callback to be called, whenever an object is > loaded, passing the root node of the corresponding DOM subtree. Then it > would be easy to parse out the required parts, either manually or using > the COLLADA DOM and feed them into the physics engine. > Yes, I agree. It should mainly be a matter of modifying the collada reader slightly so that it allows for the callbacks and hooks necessary for integration with a physics engine. > I do not believe that the Collada loader for OSG will ever support > physics - it doesn't really make sense, OSG is a scene graph and physics > could be implemented in many different ways. > No, and it shouldn't. The question is how it can be best adapted for efficient integration with a physics engine - and if this adaption can be submitted to the cvs. It shouldn't be a problem I think, but might require some slight changes in the design of the plug-in strategy to open for these types of hooks/callbacks. Will look into the actual code next week... >> Of course the same questions apply for a collada exporter.. >> > > For example, the Blender exporter for Collada exports physical > properties too (e.g. the collision bounding volumes and masses). > However, that is likely because Blender includes its own physics engine > already. > Yes, but I meant something else. A physics/graphics based simulator should be able to dump its state into a collada file (like osg is able to dump a state into an osg-file). So, if one would develop a Collada exporter for OSG, one should also be able to use hooks/callbacks to the same exporter so that the physics engine can dump the state of the physics scene.
Of course, the above strategy means that the physics engine will depend on osg for it's collada reader/writer, rather than having it's own support for collada. This is also slightly frustrating, since a physics engine should be open for integration with any graphics library - but I guess I can live with it, since it would still solve most of our problems. /Kenneth -- Kenneth Bodin, HPC2N Director VRlab Umeå University, Sweden [EMAIL PROTECTED] Philosophy of science is about as useful to scientists as ornithology is to birds. -- Richard P. Feynman _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
