Hi Robert,

There are a couple of issues at play actually.  I've been dealing with quite
a few buggy graphics card drivers lately.  This particular one is an HP ATI
Radeon Mobility x600.  It reports that GL_EXT_framebuffer_object is
supported, but when OSG tries to use it, it crashes.

To try to debug this, I was trying to manually set the render target
implementation of the OverlayNode to PIXEL_BUFFER and PIXEL_BUFFER_RTT to
see if they would resolve the problem.  I then noticed that because the
OverlayData is cached that calling "init()" was not propogating the changes
down to the OverlayData objects.

Disabling the extension frame_buffer_object extension using the
setGLExtensionDisableString method as you suggested for my issues with VBO's
with the Intel 945GM also works in this situation, but it is still a problem
that setRenderTargetImplementation doesn't work correctly, which is what my
fix was trying to solve.

Thanks!

Jason

On Fri, Aug 15, 2008 at 12:20 PM, Robert Osfield
<[EMAIL PROTECTED]>wrote:

> Hi Jason,
>
> On Thu, Aug 7, 2008 at 5:40 PM, Jason Beverage <[EMAIL PROTECTED]>
> wrote:
> > I've got a few machines that OSG incorrectly believes support
> > FRAME_BUFFER_OBJECTS so I am manually trying to set the
> > renderTargetImplementation to PIXEL_BUFFER or PIXEL_BUFFER_RTT.
>
> Could you explain this in more detail.  Does the OpenGL driver declare
> it supports but doesn't?  Do you think this is an OSG bug or a driver
> bug?
>
> The reason why I ask this is that it sounds like you are trying to fix
> one problem that also needs fixing elsewhere.
>
>
> > I noticed
> > that this call wasn't setting the camera's overlay data properly because
> the
> > setRenderTargetImplementation simply calls init() which only does
> anything
> > for OBJECT_DEPENDENT_WITH_ORTHOGRAPHIC_OVERLAY.  Any subsequent calls to
> > getOverlayData will simply return the cached OverlayData instead of
> setting
> > it up.
>
> Just reviewing the code I notice the lack of anything been done in
> most modes.  This suggests that perhaps we should be adding more to
> these methods to make sure all parameters are propagated correctly.
>
> Robert.
> _______________________________________________
> osg-submissions mailing list
> [email protected]
>
> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
>
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to