Re: pfLayer Culling Problems
John Rohlf (jrohlf++at++tubes)
Mon, 22 May 95 9:32:50 PDT
>
> Hi,
>
> I've noticed a strange behaviour of pfLayers. In my application I create
> pfLayers after loading Flight format file. i.e. I traverse through the scene
> graph and create new layers in certain places. Those layers replace other nodes
> which are removed from the scene graph. The strange thing is that after I
> update the graph, the sub-graphs, which are under the layer nodes are not
> cull-tested properly. When I look on the polygon count in the chan stats, I see
> that they are tested against the far clipping plane, but not against the other
> faces of the viewing frustum. i.e. I can direct my eyepoint to the sky, or to
> the end of the world, and those nodes are still drawn, and, ofcourse affect my
> precious hertzes ...
> I've tried to modify all bounding volumes for everything after I add the
> layers, to no avail.
>
> However, when I create my own 'layers', meaning that I implement decaling with
> draw callbacks in the appropriate nodes, everything becomes normal again. the
> addinional nodes (now pfGroups instead of pfLayers) are culled properly and my
> hertzes are almost intact.
>
> Is there something I'm missing here ?
> If its indeed a bug, will it be fixed in Performer 2.0 ?
In 1.2, pfLayer children were not cull-tested if the pfLayer
was partially visible. This was a cull optimization based on
the presumed spatial coherency of layers but one which might draw
unnecessary geometry. This "feature" has been removed from 2.0.
This archive was generated by hypermail 2.0b2
on Mon Aug 10 1998 - 17:51:30 PDT