Hi Alexandre,
In my resize implementation I directly update the projection matrix from the slave camera and I don't update the _projectionOffset wich for me is useless.
Why don't you update the projection matrix of the master camera? If both slave cameras render to the same GraphicsContext (i.e. are in the same window) then it's the right thing to do. You shouldn't touch the slave cameras' projection matrices, only the master camera's projection matrix and the slave cameras' projection offsets.
I still think that future osg version should correct the default resizeimplementation because this behaviour seems very strange to me, maybe am I the only one to think so ;-)
Well if you mean that it doesn't do the right thing when you have one master camera and one slave camera, both with graphics contexts, then you're right, but that's expected because it's not a valid use case. You should use a master camera with no graphics context, and two slave cameras with graphics contexts (perhaps the same one). See the code for osgViewer::View::setUpViewOnAllScreens() for an example of this.
Perhaps a test could be added to catch that kind of utilization and warn the user that their view's camera setup is invalid.
Go back to my previous message, really it makes sense. You have the master camera which in this case is only there to represent the viewer's position. So that's what the CameraManipulator acts upon, but it has no GraphicsContext so it doesn't render anything by itself. And then you have two slave cameras which have viewpoints and projections offset from the master camera, and which do the actual rendering, hence they have GraphicsContexts.
Hope this helps, J-S -- ______________________________________________________ Jean-Sebastien Guay [email protected] http://www.cm-labs.com/ http://whitestar02.webhop.org/ _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
