David,

 

You described exactly what I was experiencing.  However, I tried making my far clip plane much farther than the extent of my ephemeris sky dome.  For some odd reason, I still had the “large black polygon” problem.

 

My solution to the problem was a little weird.  When I set the sky dome radius to 10000m and made my far clip plane larger than that, the problem went away.  But for some reason when I used the sky dome default radius and then set a larger far clip plane, the problem came back.  I’m really not sure why.

 

Andrew

 


From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of David Spilling
Sent: Saturday, June 10, 2006 2:05 PM
To: osg users
Subject: Re: [osg-users] osgEphemeris Culling

 

"Large black polygons" is exactly what I got when I was experimenting with sky spheres, and was completely related to the z-range, and not at all to GLSL or ATI hardware. I'll post some screenshots Monday, if I get the time.

 

For me, the polygons were "centred" (if that's the phrase) over the vertices of the sky dome's triangles. The reason was that the z near/far  was very close to the dome radius - where the vertices were - but because the distance to the "dome" surface was closer to the viewer in between the vertices (because the dome vertices are connected by flat triangles) the z-range would 'clip' (not the ClipPlane sense of clip) the dome into bits. These bits would pop around if I moved, mainly due to the rotation of the dome.

 

My recommendation would be to experiment with your z-range. It's a pity the code I posted doesn't work; I don't know what goes on "under the hood" on osgEphemeris so there could be all kind of things going on - e.g. transforms etc. A technique based on fixing your projection matrix for some subgraph, along with suspending osg's autocalculation of zrange for that subgraph, will work, I reckon.

David

 

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

Reply via email to