I was looking at experimenting with the terrain engine a bit. I'm not sure a fractal terrain like Outerra is the best design for something like FG. Notice that Outerra's design goals are to show the earth before the effects of man. Although they are planning on developing other landclass textures, I'll need to see what kind of results they get when trying to model something like suburbs and urban areas.
The biggest problem with FG's terrain system, is the lack of a level of detail algorithm. Looking at the planet from high altitude, or even increasing visibility to get Thorsten's local weather to really shine is going to require using less detailed polys in the distance.
Because FG uses an irregular triangle mesh to represent the scenery, there really isn't any way to mathematically (with acceptable frame rates) lessen the detail for tiles in the distance.
I had started implementing the cdlod implementation
http://www.vertexasylum.com/downloads/cdlod/cdlod_latest.pdf in OSG. I was just going to use a single texture, and see if I could get it into a new tile format that was 'wrapped' in the BTG format simgear uses.
This was going to take an enormous amount of effort, and it would need to be done in such a way as to not break the existing scenery. Many people have a large investment in the current engine, so any new engine would need to live side by side, until the merits of a new engine are obvious enough to make a switch. All I was hoping to get out of it was to see how difficult it would be to decouple the terrain engine from the other systems. If it could be done, many other lod algorithms could be tried with a lot less pain.
The problem with regular mesh scenery, is that to get irregular texture landclass, you have to generate textures for each piece of the planet. I was reading up on how FSX handled this with a combination of pre-made very low level of detail textures, and using Anti-Grain to generate the highly detailed textures on the fly.
Interesting stuff, and it could be a lot of fun. But it would take many man months to even get to a proof of concept.