Well, if I understand the way it is working, the fbo saves you the time to copy from the framebuffer to a texture, but you still have to transfer from the texture to main memory. But if is right, then rendering to an image should be faster using an fbo, even when there is only one fbo (and no overhead related to binding...), but it seems the performances are the same whatever the method in osgprerender. Anyway, when using a lot of rtts, I found that using fb is really faster than using fbos, is it a problem of osg ? or just normal due to the way fbos works ?
So If i plan on doing a lot of rtts, saving each render to an image (in main memory), should I stick with adding a lot of cameras each rendering to an osg::Image with framebuffer implementation ?? or has anybody else ever tried something else ? thanks On 6 fév, 17:21, nicolas martin <[EMAIL PROTECTED]> wrote: > Well, let's say that i'm adding like 50 cameras so basically 50 > fbos ... > But as I currently don't really understand what is the need for fbo, > if > you're about to copy the texture to an image anyway, can anyone > explain it to me ??? > I'm also really interested in understanding what is exactly osg doing > when it is rendering to an image ... Creating an fbo (with what format > for the texture ??), rendering to it, and then copying it to my > image ?? > > thanks > > On Feb 6, 2:59 pm, "Paul Martz" <[EMAIL PROTECTED]> wrote: > > > How many fbos is "a lot"? Hundreds of BindBuffer calls per frame can not be > > a good thing, perhaps that explains the performance hit. > > -Paul > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] > > > [mailto:[EMAIL PROTECTED] On Behalf > > > Of [EMAIL PROTECTED] > > > Sent: Wednesday, February 06, 2008 10:42 AM > > > To: [EMAIL PROTECTED] > > > Subject: [osg-users] performance fbo > > > > Hi all, > > > > I modified the osgprerender example to add a certain number > > > of cameras, doing exactly the same thing as the one that<s > > > rendering to a texture, but made them render to a texture > > > that is not used. > > > So in the scene, there is the main viewer, the camera that's > > > doing rtt, and a lot of other cameras doing rtt for nothing. > > > What I've experienced is that by using the fbo > > > implementation, the fps is really lower than the one I get by > > > using any of the other implementation ... > > > You may ask why I'm doing this, but I want to do a lot of rtt > > > saving each render to an image, so I thought that just by > > > modifying the osgprerender example, it would do the trick ... > > > but :( Is there some kind of overhead related to using a lot > > > of fbos compared to another method ?? > > > > Thanks all > > > _______________________________________________ > > > osg-users mailing list > > > [EMAIL PROTECTED] > > >http://lists.openscenegraph.org/listinfo.cgi/osg-users-opensce > > > negraph.org > > > _______________________________________________ > > osg-users mailing list > > [EMAIL > > PROTECTED]://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph... > > _______________________________________________ > osg-users mailing list > [EMAIL > PROTECTED]://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph... _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
