Comments on: How To Make a “Falling Sand” Style Water Simulation

By: White Shadow

White Shadow — Thu, 14 Jul 2011 08:59:59 +0000

Ah, now I see it. One possible way to fix the problem would be to only allow a sand particle to move left/right if both the corresponding diagonal and horizontal neighbours are empty:

if ((x > 0) && (sand[x - 1][y + 1] == 0) && (sand[x - 1][y] == 0)) {
	left = true;
}

if ((x < (WIDTH - 1)) && (sand[x + 1][y + 1] == 0) && (sand[x + 1][y] == 0)) {
	right = true;
}

(I haven't tried this myself.)

By: shpen

shpen — Wed, 13 Jul 2011 18:47:38 +0000

I’m actually going from top to bottom, but bottom to top still causes the same problem.

Maybe you can take a look at my source and help me figure out what’s wrong:
http://dl.dropbox.com/u/77961/GameTemplate.java

Also, here’s a jar which will show you what the problem looks like:
http://dl.dropbox.com/u/77961/SandTest.jar

Thanks!

By: White Shadow

White Shadow — Wed, 13 Jul 2011 06:49:05 +0000

Hmm, I don’t see how that could happen. If you’re iterating left to right, bottom up, the entire lower line should fall down “at once”, from the perspective of the simulation.

By: shpen

shpen — Wed, 13 Jul 2011 01:38:06 +0000

I’m having problems with creating an even falling pattern. Because I am iterating through each sand particle one by one, from left to right, the sand tends to fall to the left, since a slot on the left is opened before one on the right. Is there any way to fix this and cause an even falling pattern?

By: White Shadow

White Shadow — Tue, 21 Dec 2010 23:31:00 +0000

True, but I suspect performance would suffer if you copied the entire world (twice!) every frame.

By: Grant

Grant — Tue, 21 Dec 2010 21:19:38 +0000

Whoops! Instead of “…but modified the copy array…”, it should read “…but modifies the copy array…”

By: Grant

Grant — Tue, 21 Dec 2010 21:18:41 +0000

There is a very easy fix for the “teleporting” problem. Before the main loop (updating every tile), copy the entire world into a “copy” array. Then, whenever a particle is updated, it gets its information about the world from the original array, but modified the copy array. At the end of the main loop, you copy every value from the “copy” array into the origininal array. The reason it works is: every particle gets its information from a non-updated array! No particle can influence another particle DURING the updating process!

By: Henk

Henk — Fri, 27 Aug 2010 21:22:00 +0000

Neat tutorial, I followed it on a bored friday evening :). I implemented the teleporting thing using a simple ‘updated’ boolean 2d array that sets fields to true if they’ve been updated during that round, but that’s mainly because I’m not much of a bit twiddler.

Haven’t got the water going horizontal right yet though. It’d probably become either very complex, or need to be rewritten if I ever add other materials.

By: White Shadow

White Shadow — Wed, 30 Sep 2009 13:14:58 +0000

You are correct, updating bottom-to-top is a good approach. I didn’t mention it in the post, but that’s exactly what my demo app does.

However, that alone wouldn’t be enough to prevent “teleportation” in a more complex simulation. For example, if you added a rule like “if this fluid particle has another fluid particle above it, move this one sideways” (i.e. a very basic pressure factor), you’d get particles teleporting left or right depending on the direction of your X loop. So tracking which particles have been updated still makes sense.

By: Ross Shannon

Ross Shannon — Wed, 30 Sep 2009 13:00:31 +0000

Good tutorial, thanks. I don’t understand the problem with particles teleporting downwards though — it seems to be that the only reasonable way to program this is to do the simulation updates bottom-to-top. Otherwise, if you update top-to-bottom, when you update the game state array, any particles that have another particle above them will be overwritten by the top particle falling into their position?