The World Builder - Rivers versus lakes

[RonHiler]

This might be very minimal near the edges and I'm not sure how such a large tile area will cope with this, can you give us an estimate on the size of each tile we're talking about here?

A bit of a loaded question It's not so easy to give a simple answer to that, heh.

At the equator there are 8192 points all the way around, and less as you move north or south (until there is only one point at each of the poles).

If you want a scale, figure the earth is about 40K kilometers at the equator, so each point is roughly 5 square kilometers, presuming a spherical world that is the size of our planet.

Now, that's the final world map. During the world building process, it's actually going to be 8 times smaller (1024 points around the equator), and that world will get interpolated into the bigger play world (this is done for a couple of reasons which I won't get into).

So after reading your proposal a few times over, I like it. I think it might work just fine. It has an advantage in that I don't need to have a height based sorted list of points, which is good. It should keep water reasonably flat. And I don't need to keep going over points again and again, One time and we're done.

What do you think, Strike? Seem like it might work to you?

I'm going to try to break it over the next day or so, maybe even try to do a quick and dirty simulation with code to see what happens.

[Shane Christopher]

Hrm, so you're looking at roughly 40km^2 at the equator. At that size a lake will only be one tile big anyway, if you use my idea you'll have to judge by the amount of water in the tile whether or not it should look like the water takes up the whole tile or whether it is just a river.

Keep in mind that my idea does need to be iterated but like I said it should be possible to do that over the course of the other iterations you have happening. And the calculations themselves shouldn't really take too long. You're looking at the averaging out of the water in the method I described about 1,000,000 times. Given that there's probably less then 1000 ops per tile that's around 1,000,000,000 at a rough guesstimate. About half a second or less for any decent processor.

[Strike]

First off, let me say that I have no complaints with Shane's method; it certainly seems to work whereas when I tried testing my idea, it failed.

However:

The reason being, running water will affect the topology. It's really the whole reason we are modelling water at all during the iterative process (otherwise I'd just skip it and put in lakes/rivers at the end).

Hrm, so you're looking at roughly 40km^2 at the equator. At that size a lake will only be one tile big anyway

When working on that scale...
You could *probably* not bother comparing water heights, and just compare relative terrain elevations.
Use the rainfall to determine how much water is available to move topsoil, and if the elevation difference between two adjacent tiles is "large enough" (again, depending on your scale), move the topsoil around.

Rationale: Fast water moves soil. Slow water deposits soil. Water speed is based on elevation difference.

What we haven't done yet is create deltas. If you could model that, I would consider it cool

[RonHiler]

Eh, don't get hung up on the scale guys, I provided a comparison because Shane asked for it, but in reality, I expect our world will be significantly smaller than "earth" sized.

In actual fact, I'm probably not going to provide a scale at all. Lakes will look like lakes and cover multiple points, rivers will be one or two or perhaps even 3 points wide (if we tried to work that into an earth scale they would be 15 km across, which is not too likely, heh).

[RonHiler]

Rationale: Fast water moves soil. Slow water deposits soil. Water speed is based on elevation difference.

Wow, I am SO glad you said that. That concept hadn't occured to me, believe it or not. I was working under the idea that faster water would move more topsoil, slower water would move less, but any moving water would move it.

But of course, you are absolutely right! Slower water will DEPOSIT topsoil, not move it. That small (and easily incorporated) change should vastly improve the realism of the areas around river mouths, it should give us swamps, and yes, maybe even deltas

[Shane Christopher]

The Amazon estuary is some 330 kilometres (207 mi) wide:

It'll be grand!

I hadn't particularly considered that fact either strike although I should have seeing as it's not that many years since I studied Geography and these things.

I suppose no matter what the scale it'll give decent enough results. Imagine your map as a big model on a table with water being precisely sprinkled in certain amounts on each point. The way the water reacts in that situation, while not perfect would be close enough to how it would react if that model was 10X bigger or even world sized. We're just talking about a difference in water intereaction, obviously smaller tiles would be more accurate but water will still react similiarly in each situation.

Ron how about 40,000,000 points around the equator to give us a nice 1m resolution?

[Strike]

Yay, I'm contributing again! ^_^

I don't know (and haven't researched) how fast the water "should" be moving to move the soil around, but now you just have to move soil instead of moving water. ...Although that will mostly look the same in the algorithm, I think, but you don't have to worry about bumpiness.

[Shane Christopher]

I think we should use a system where water moving at speed x can hold y amount of soil. If the water has more soil in it then y then drop the excess onto the area.

Even slow moving water can hold a small amount of soil.

[RonHiler]

Something like that. I don't necessarily want to keep track of "Moving Soil", but I think it amounts to the same thing anyway (if tile A has 10g of soil moved into tile B, and tile B has 5g of soil moved into tile C, the net result should be an addition of 5g of soil to B).