Hello Alexej, Robert
I think the reason for Alexej to think that disabling the culling would
improve render time in any way, is that he measuring the time of the
cull traversal.
Indeed the cull traversal time is culling + preparing the drawing of the
geometries. I've fallen for this in the beginning.
So maybe the problem lies with the number of drawables (i.e. there are
too many too small batches). So being clever at this point would rather
mean to make bigger yet fewer sets.
cheers
Sebastian
Hi Alexej,
On 24 August 2012 14:41, Alexej Fink <[email protected]> wrote:
The point is: At the group node it can be already decided, if the underlying
nodes are visible or not. Thus no need to compare against each single
Drawable's bbox for refinement.
So I would like to tell the cull-visitor: Take all underlying drawables, if the
group-node bbox passes (or skip all, if not).
The OSG will automatically stop do the cull tests against the frustum
sides once a frustum side totally encloses a subgraph, so there really
isn't much to be gained by trying to be clever, it's already doing
everything it can to be efficient w.r.t cull tests.
Culling is import even down at the leaf level so there is still
performance to be gained by leaving culling on right down to the leaf
level.
Not only does the OSG using the bounding box of the Drawable for
culling it also uses to help compute the near/far ranges required to
encompass the scene.
So.... stop trying to be clever, you actually just outwitting yourself
for no reason, you'll end up with crappy solution because you are over
thinking things. JUST set the bounding box on the Drawable and let
the OSG do it's stuff, it written this way for a reason.
Robert.
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
