Well,
great to here, that you have bought GC1. How about showing, that you can make better images with rivers? Register to the GeoControl forum (www.cajomi.de/Forum), and you can test GC2, no joke!!
I do not want to speak about all problems, and mostly not about the solutions. I need 6 weeks only for that algorithm, and there is still a lot to improve. This is the maximum time I ever needed for a single algorithm. A good erosion needs about 2-5 days for the base (that is where the river and lake algorithm is - the base) and one week for getting the best out of it (with a lot of testing), and 1 days to full implementation (including slider values, saving and so on).
One great problem in opposite to erosions is, that water has itself a height. So the water is not only floating to the lowest point, but also to the second lowest point. Now simply imagine, how often a flow can be splitted into two flows, and you get a astronomical number for flows.
Another difference: If the water reaches a lowest point it creates a lake and then flows out of this lake. Let us assume, you simply took the lowest point of the lake border. But now the flow stops soon again, because the direction was a dead end, and you create again a lake, which now grows into the first lake. If you look from above with human brain, you would see at once, where the water should flow, but a algorithm has some simple rules to follow, and now this process can take really much time, just because there a really many deadends. Of course this dead ends may split in between, create a third lake or fourth, before running into the final deadend.
The results of this two problems are hanging machines, fast runnings out of memory or, when you begin to avoid the hangings, soon stopping flows.
The better the terrain, the more fine details, the more chances for such problems. A real fluid simulation for water on terrains, to find the river, are at time not possible. What may work, is a real fluid simulation in an before generated river bed, like GeoControl can produce.
