How can I generate random lakes and rivers in my game?

Question

I have a 2D block building game and am trying to make randomly generated lakes and rivers. I have looked into the Perlin noise algorithm, but, I couldn't get it to generate random and nice results.

I have been trying to use the python noise library, but, it didn't create maps very randomly.

Is there some seed function I am missing on that library to make it more random? What variable do I change if I want it to go more random? If possible, give me a less technical answer, with less math and more python terms.

The map is a 2D tiled map. Here is some examples of the non-randomness of the other algorithm. The following code was outputted 3 times in a row. I was randomizing the octaves and frequency with something like this: freq = 16.0 * random.randint(1, 500000) * 0.000001 + 0.5

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1
1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1
1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1
1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1
1 1 1 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1
1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1
1 1 1 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1

This is the code that produced the output above:

"""Writes a 256x256 grayscale simplex noise texture file in pgm format
(see http://netpbm.sourceforge.net/doc/pgm.html)
"""
# $Id: 2dtexture.py 21 2008-05-21 07:52:29Z casey.duncan $

import sys
from noise import pnoise2
import random
octaves = random.randint(1, 500000) * 0.000001 + 0.5
freq = 16.0 * octaves
for y in range(30):
    for x in range(40):
        n = int(pnoise2(x/freq, y / freq, 1)*10+3)
        if n>=1:
            n=1
        else:
            n=0
        print n,
    print

Answer 1

Well, as it seems, you are not seeding the random number generator. In python, it can be easily done with just a random.seed().

And I can see too you're generating a number between 1 and 500000 and making it be between 0 and 1. It's a functional method, but it is capped to just 500000 possibilities. You're better with just using random.random() it already generates a number between 0 and 1, but with much more possibilities! If you still need a number between 0.5 and 1.0, as your code suggests, you could just do: (random.random() * 0.5) + 0.5

Your final code should look as follow:

import sys
from noise import pnoise2
import random
random.seed()
octaves = random.random()
# octaves = (random.random() * 0.5) + 0.5
freq = 16.0 * octaves
for y in range(30):
    for x in range(40):
        n = int(pnoise2(x/freq, y / freq, 1)*10+3)
        if n>=1:
            n=1
        else:
            n=0
        print n,
    print

That's all!

Answer 2

This isn't answering your specific programming question, but consider that creating lakes and rivers isn't about randomly placing blobs of water and strips of water between them. It's about terrain height - about depressions (basins) that turn into lakes, and water that flows from higher to lower spots.

If you want a great example of creating lakes and rivers that make sense, you might check out this blog post -> http://simblob.blogspot.com/2010/09/polygon-map-generation-part-1.html It's a good reference for this kind of thing if your goal is reasonably realistic hydrography.

Answer 3

It's very simple: if you're getting the same map three times in a row (or more), it's because you didn't randomize the seed.

What does this mean?

Computers are inherently deterministic (non-random), so they simulate randomness. It's actually repeatably random (that's why we call it "pseudo random number generator").

How does this work?

When you create a random number, you have the option of giving it a "seed." The important thing is if you always use the same seed, you will always get the same sequence of random numbers, in the same order. Always. This can be good or bad.

In your case, it looks like you're not seeding the random generator, and by default, you're getting the same seed -- it probably uses some component of your date/time. Hence, I recommend you randomize it.

As Gusatavo mentioned in his answer, you need to call random.seed(). The docs state that "if X [the default parameter] is omitted or None, current system time is used." This should be sufficient.