I was advised by Wilbur’s author to use basin fill but you can’t really control it very well because it tends to make major changes to large area so I ended up not using it. BUt, our problems may not be alike. I’m working quite cloesly to a topo map reference, where rivers are strictly marked, but, if you had a bit more flexibility in your map, you might fair better with that approach.
One limiting factor with Wilbur is map size. It doesn’t use tiling so you’re limited by your system much more. I’ve used 10K terrains fine, but I think (with 6GB) I could probalby get up to may 12K if I pushed it.
World Machine has probably the best erosion tools around but they work best on small to medium scales. You might turn your sediment carry right up and thinking about it, I do that to solve the same problem- it fills in basins created by the channelling erosion. But they are very localised. I doubt very much if it will extend to large basins. Geologic time enhancment produces some great eorsion but it’s pretty destructive and unpredictable.
Regards rivers, this is what I’d advise you to do. The single most important thing about large landscapes is rivers. If you are not going to be seeing the landscape from a great height, then ignore this, but if you can see it, then this is what I’d do.
You need to create a river mask whether the rivers are real or not. What I’m saying is you need the eye to see them, but they might not necessarily have accurate hydrology (if you ran a physics sim over them). The eye doesn’t see the rivers (not the smaller ones) but it sees the terrain implying them. So you need to get as many good long flow networks from your initial river mask + the incise flows feature in Wilbur. Incidentally there are only 3 programs (as far as I know) that can give you long river flows: Wilbur, GeoControl, GTS (search
Viewing Dale). (GeoControl produces great rivers but is limited to 4096 as it’s still 32 bit).
Create upland areas where water can flow from (a hills mask) in WM. Doesn’t have to be complex terrain even. Build, export. Cut channels into the terrain in Wilbur along your initial river mask to encourage flow along and into them. Noise the terrain a bit still masking. Then run incise flows and precipiton repeatedly. Once you have results you’re hapy with, generate a river mask. Then take your new river mask, combine it with your original river mask to patch any holes, and apply that to the original hills mask in order to break the hilly areas up (ie, generate the hills around the new mask). You could repeat that cycle. I think if you did it enough times you’d have great looking flows- but you may end up changing the coastline too much with the Wilbur erosion. I think it could be done though by using a mask on the erosion step which reduced the erosion towards the coastline. Another thing is, you might not even use the output terrain from Wilbur since you can use the river mask for both terrain placement and cutting channels along rivers to the original (pre-Wilbur) terrain.
Another approach is to just use real terrain in your initial hills mask, then run that through Wilbur to gen your rivers mask. Not tried that one yet but I think I’ll be trying that in future- it may or may not work that well.
As you can see I’ve still go a few things I’d like to try. 
Re .shp. Yes, they have z info but I don’t think each vertex can have a different z value. They all have to be the same. I don’t know. That’s one for Stephen! .shp is the industry standard GIS (Geographic Information Systems) vector format. It’s used in applications like Google Earth, WorldWind.
Me too with the artistic approach, what I know is just stuff I’ve picked up over the years. I think all of the big challenges will have to be tackled by programmers 
monks
I painted it in the first place…
