|
Author Topic:   LOD again
Hardy
New Member 		posted November 09, 1999 09:41 AM            
Right here's the plot -
At the moment I am trying to implement a hybrid of Peter Lindstrom and co's
Real-time, continuous lod rendering paper.
From this I am using the quadtree structure, which is broken down into different
levels depending on where the camera is positioned.

The bit I am stuck on is how do I tell where abouts in the mesh I am situated
and how do I go about calculate what is in my FOV or LOS.

Also what method of visability does ROAM use.

Help me please, its probaly a stupid question but my head is hurting from
banging it against my bedroom wall.

Cheers thanks and bye
Matt

IP:

bdiscoe
Member 		posted November 12, 1999 07:49 AM            
I have a working implementation of the Lindstrom-Koller algorithm. It's not great, but you're welcome to try my source code. That would probably explain it better than i could in this forum.

ROAM doesn't go into detail about a lot of things - like what kind of view-volume culling they do, what kind of error metric they use. I'm kinda upset with the authors, actually - they boast about how great their approach is but leave off nearly all discussion of implementation.

Visit my online notes at http://www.washedashore.com/Terrain/

-Ben

IP: