Hi Robert,
> It sounds like you've been rendering all the cameras to a single texture? > I am still weighing up the whether multiple textures or a shared texture > would be appropriate. Multiple textures, i.e. one per eye camera would be > the most scalable. > Yes, I don't remember exactly the numbers but when I was setting up this configuration I tested both configurations and I got better performance rendering all the cameras (as viewports) and then doing only one RTT to screen and apply the interleaving shader knowing the grid layout and each viewport size. Also, I had some problems with texture arrays in glsl for the interleaving shader so I ended up using one RTT. I think multiple textures or a single final RTT, depends on the application, but the multiple texture would be for sure the most scalable. > >> So answering your question, Yes, a callback per slave computing each >> frame the matrix offsets should be enough. I was using a similar approach. >> > > osg::View::Slave's already have an update callback that can be used to > compute the offsets. I'm in the process of writing such a callback to > handle standard stereo, but we can probably use this same callback class > for multiple cameras. Even if we can't it's still quite lightweight a > callback class. The key thing right now is getting the general approach > right so that it's easy to add various alternative setups. > Perfect, I will be checking svn, and I will try to set up a multiview display with the new configuration. Thank you! Rafa. > > Robert. > > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > -- Rafael Gaitán Linares CTO at Mirage Technologies S.L - http://www.mirage-tech.com
_______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
