Robert
Thanks a lot - I had missed that completely.
Cheers
Alan
Robert Osfield wrote:
Hi Alan,
The screenshot say alot, it says to me that the problem you have is
down to the near clipping plane being too close. The ways to solve
this problem are to enable a finer grained way of computing the
near/far plane based on polygons intersections with the view frustum
and to reduce the near/far distance ratio to allow the near plane to
move in nearer to the eye point. Both these can be set in the
osg::CullSettings structure:
In osgViewer::Viewer you'd do:
viewer.getCamera()->setComputeNearFarMode(osg::CullSettings::COMPUTE_NEAR_FAR_USING_PRIMITIVES);
viewer.getCamera()->setNearFarRatio(0.0002f);
You can drop the COMPUTE_NEAR_FAR_USING_PRIMITIVES if you wish as the
default COMPUTE_NEAR_FAR_USING_BOUNDING_VOLUMES should suffice given
the setNearFarRatio change, the compute using primitives is also more
expensive an op so if your a limited by cull traversal then don't use
it. What you do gain with compute using primitives is better
optimized near/far ratio for each frame, with the OSG pushing out the
near plane as far as it can without clipping polygons. Pushing the
near plane out is important as it has a huge effect on the numerical
precision of the depth buffer.
Robert.
On 4/30/07, *Alan Harris* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
Robert
I tried to find out a little more detail about what is happening.
Attached is a screenshot to illustrate the problem. The viewer is 2m
above the ground height.
I thought it might be culling one node (5x5 quads plus a 256 pixel
texture) but the ground surface is clipped as you can see (not at the
skirts of the database which are also visible) and it is clipped
smoothly as one advances or retreats - it doesn't suddenly include
another node.
It seems to be a clipping problem and not culling. The clipping
generally occurs when looking uphill - but not enough of a slope
that I
expected the near plane to be clipping it.
I am using the following to set the near clipping plane to 0.001m,
but I
suspect I am missing something.
SceneView::setProjectionMatrixAsPerspective(fovy,aspectRatio,znear,zfar)
This happens if the window size is changed. On re-entry znear is
set to
another value, typically 1/2000 of zfar. Is there a control
somewhere on
the ratio or values of znear and zfar?
I would be grateful if you could point me in the right direction.
Regards
Alan Harris
Robert Osfield wrote:
> Hi Alan,
>
> On 4/26/07, Alan Harris < [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
>> Return of an old question!
>>
>> If the corners of the bounding volume (I assume a box) are all
outside
>> the view frustrum would that node be culled? I assume a
bounding box is
>> used as I would expect a bounding sphere to generally interact
with the
>> view frustrum.
>
> The Node's all have bounding sphere's that enclose their children.
>
> The Drawable leaves all have bounding boxes that enclose their
geometry.
>
> The view frustum culls the above such that if any point of the
> bounding sphere or bounding box is in the view frustum then its
> accepted.
>
>> I expect it is from what is happening. Therefore is there a
simple way
>> of ensuring that culling does not occur on objects within, say, a
>> specified distance of the viewer.
>
> There isn't a way to disable culling for near objects, it
shouldn't be
> required any, such a feature would just be a hack around a bug
> elsewhere in the culling set up - in this case its looks like the
> ClusterCullingCallback set up is wrong so this is where the fix
should
> happen - in the set up of the ClusterCullingCallback.
>
> Robert.
> _______________________________________________
> osg-users mailing list
> [email protected] <mailto:[email protected]>
> http://openscenegraph.net/mailman/listinfo/osg-users
> http://www.openscenegraph.org/ <http://www.openscenegraph.org/>
_______________________________________________
osg-users mailing list
[email protected] <mailto:[email protected]>
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/
------------------------------------------------------------------------
_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/
_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/
