Hi Kim, On Mon, May 11, 2009 at 7:50 PM, Kim C Bale <[email protected]> wrote: >>Is there any chance that we'd be able to get osgOcean working on > geocentric databases? It'd be very cool to be able zoom down from > space to the sea. > > As I don't know a lot about geocentric databases it would be hard for me to > say. But the potential > for animated oceans on very large terrain datasets would be very exciting. > For very large > expanses of ocean the problem I forsee is the time it takes to update the > vertices and primitive sets.
LOD'ing, tiling and local coordinate systems are the tools of the whole earth database trade. In the context of osgOcean I would expect to have several different techniques - one for near water effects like you have which his surface displacement and shaders, then for lower levels of detail techniques that just have static geometry meshes and shaders/textures for the water rendering. You'd also want to try an reuse tiles so that as you move out of range of one set of tiles these then get reused for the new tiles that need to be created. It is sometimes possible to reuse a small set of geometries and just use shaders to map the separate instances - this is what I did for the rain and snow effects - it's an effectively infinite effect, but implementation it's just a single block of geometry that is animated and positioned by uniforms pass into shaders to look like an infinite effect. Perhaps something similar might be possible. The other challange would be the non uniform boundaries between water and land. Is there any support for this? Another little point with geocentric databases is that each tile is slightly curved, the lower the level of detail the large the area covered and the more curvature it will have. For precision one also uses a local MatrixTransform above each tile to ensure that there is a local origin in the center of the tile to maximum precision. One also need to think about the how the tiles boundaries are managed, with a dynamic effect like sea it'll be more difficult to hide than on terrain where even crude skirts can be sufficient. If the boundary is far enough then the gap might be small enough not to matter. > However, since the FFT technique is tileable it would be possible to only > update 1 tile and then translate > them into position using a shader. This would rule out any surface > interactions, but for this type > of application they probably wouldn't be necessary. I'll add it to the list > of things to look at. > >>Is there are plans for following water like in rivers? > > For the moment rivers are out of the scope of the project as I think this > would require a totally differant > approach. The problem is that I'm torn between what I'd like to do and what > I'm bound by > contract to open source. So for the moment I'm concentrating on getting the > code out to keep > the reviewers happy, then once I'm done with that I'll do some research into > the many suggestions that > have been posted here. Understood. Is it possible to do lakes and non flowing rivers though.... >>Sorry for the all the awkward questions, this stuff is just too pretty > not to want to put it everywhere :-) > > Not at all. I'm extremely happy that it has been so well recieved. Ahh we're all just suckers for a nice bit of eye candy ;-) Robert. _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
