I'm trying to add multisampling capabilities to my floating-point FBOs, but
I've found what I think is a bug in osgUtil/RenderStage.cpp. Basically, when
I specify a number of samples greater than 0, I actually get a
non-floating-point framebuffer.
See line 380 in RenderStage.cpp:
GLenum internalFormat = attachment._internalFormat;
if the attachment is actually a RTT target rather than a rendertarget, the
internalFormat field will always be zero. This will make RenderStage pick
the default value which is GL_RGBA, thus ignoring the target texture's
internal format (GL_RGBA16F_ARB in my application). Replacing the above code
with the following, fixes the problem for me:
GLenum internalFormat;
if (attachment._texture.valid())
internalFormat = attachment._texture->getInternalFormat();
else
internalFormat = attachment._internalFormat;
Now, I'm pretty new to all this multisampling stuff, and I'm not sure
whether this fix is reliable and/or correct, so I'm not filing a submission
for now. Robert, I can send the modified file if you think it's appropriate.
Cheers,
Marco
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
