Hi all,
And as far as performance is concerned: how often is this rendering
paradigm used in high-fps applications?
Games like Killzone 2 are using deferred rendering
(http://www.guerrilla-games.com/publications/dr_kz2_rsx_dev07.pdf)
in conjunction with 'Multiple Render Targets' to achieve the desired
performance.
Deferred shading is orthogonal to post-render RTT passes. Its goal is
not to speed up post-render RTT passes, it's to support large number of
lights and decouple shading from geometry (essentially, lights become
constant-cost and each visible pixel is shaded exactly once instead of
possibly multiple times with results being thrown away because of
overdraw). It has disadvantages too of course.
Post-render RTT passes are just one of the tools deferred shading uses
(along with MRT).
http://en.wikipedia.org/wiki/Deferred_shading
Talking strictly about post-render RTT passes, they are not costly at
all. There is no overdraw, practically no geometry, no traversal to
speak of, etc. so they are very cheap. All resources stay on the GPU,
there is no flush of the pipeline, no read-back of textures, and it's
just N calls of a fragment shader (where N is the number of pixels the
quad covers), so as long as your processing shader is not costly it will
be fast.
J-S
--
______________________________________________________
Jean-Sébastien Guay [email protected]
http://www.cm-labs.com/
http://whitestar02.dyndns-web.com/
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
