[1.7.10] Generating a sphere

[The_SlayerMC]

Hey, im trying to generate spheres for my dimension, but i cant quite get the sphere shape, all i can get is this: http://prntscr.com/4xam3y

 

As you can tell, its making a circle shape but its turning out to be just a 1x block thick platform

 

This is the code i wrote:

int radius = 8;
		for(float i = 0; i < radius; i += 0.5) {
			for(float j = 0; j < 2 * Math.PI * i; j += 0.5)
				w.setBlock((int)Math.floor(x + Math.sin(j) * i), y + radius, (int)Math.floor(z + Math.cos(j) * i), Blocks.bedrock);
		}

 

Whats the most efficient way to generate a sphere (a whole depth sphere)

[RANKSHANK]

you've only covered the x and y axis... you'll need a z that also has a diminishing effect on the xy radius

[TheGreyGhost]

Hi

 

The easiest way is to iterate over a cube  and for each [x,y,z], calculate the squared distance to the centre of the circle.  If it's less than the squared radius, it's in the sphere.  If more, it's outside.

 

eg

for x = centre minus radius to centre plus radius

  for y = centre minus radius to centre plus radius

    for z = centre minus radius to centre plus radius {

squareDistance = (x-centre)^2 + (y-centre)^2 + (z-centre)^2

if squareDistance <= (sphere radius)^2 then set block stone

 

}

 

Don't worry about being efficient.  Java is so fast you'll never notice the difference no matter how much you optimise it.

 

 

-TGG

[fr0st]

TheGrayGhost, even if this thread isn't mine, and I'm not relevant to the matter here, I have to thank you for that. It worked like a charm

[jabelar]

Don't worry about being efficient.  Java is so fast you'll never notice the difference no matter how much you optimise it.

 

I used to think that, but I've been doing a lot with structure generation lately and found that with very simple code (just loop through an array and place bocks indicated) it can create perceptible lag on structures of any decent size.  I actually changed my code to only do one layer per tick in order to improve this.

 

I think a sphere the size he's showing could actually show a bit of lag.  Generally though that won't be too noticeable if it happens during chunk generation because players are used to some lag from that, but if you generate the structure in response to player action (like using a tool), it can be noticeable.

 

Anyway, I was surprised that it could lag, but I think the thing is that the volume of a structure gets large very fast -- a sphere as large as shown in the OP's picture has 5000 blocks!  And I think the block placement requires fair bit of processing as well as client server sync notification.  Just giving my observation...

[fr0st]

I agree a bit, using the code he provided I tried to create a 15 radius sphere, took 2 secs to generate. But yeah, even with the most optimized code snippet ever, that amount of blocks is still going to take a bit to generate.. this is what I think of it, so yeah, you both are right

[jabelar]

I agree a bit, using the code he provided I tried to create a 15 radius sphere, took 2 secs to generate. But yeah, even with the most optimized code snippet ever, that amount of blocks is still going to take a bit to generate.. this is what I think of it, so yeah, you both are right

 

Yeah. What I ended up doing was to actually have my Y index increment over each tick, rather than loop through all Y in one tick.  That kind of spreads the lag.  I previously made it only do one dimension at a time per tick (i.e. both the Y and Z loops were incremented over ticks rather than within a tick) but at that point the total time to make the structure gets pretty long -- a 20 diameter sphere would take 400 ticks = 20 seconds.  Which was too long, but the good thing was there was no lag at all.  So it seems to be a balance.

[Eternaldoom]

Noob question, I haven't tried this before: could you generate a structure in a separate thread?

[jabelar]

Noob question, I haven't tried this before: could you generate a structure in a separate thread?

 

I think I tried that and got concurrency errors.  I'm not expert at threads, but unless you design the threads to work together you get into trouble if two threads are trying to operate on the same data -- i.e. think of one thread trying to read the array of the blocks in the world at the same time the other thread is trying to place a new block.

[The_SlayerMC]

Hi

 

The easiest way is to iterate over a cube  and for each [x,y,z], calculate the squared distance to the centre of the circle.  If it's less than the squared radius, it's in the sphere.  If more, it's outside.

 

eg

for x = centre minus radius to centre plus radius

  for y = centre minus radius to centre plus radius

    for z = centre minus radius to centre plus radius {

squareDistance = (x-centre)^2 + (y-centre)^2 + (z-centre)^2

if squareDistance <= (sphere radius)^2 then set block stone

 

}

 

Don't worry about being efficient.  Java is so fast you'll never notice the difference no matter how much you optimise it.

 

 

-TGG

 

Thanks! But can i make it so it generates random sized ones aswell?

[Eternaldoom]

Yeah, radius = Random.nextInt(some number)+min size

[The_SlayerMC]

I tried this just for a normal one:

 

int radius = 8;
		int centre = radius / 2, squareDistance = 0;
		for(x = centre - radius; x < centre + radius; x++)
			for(y = centre - radius; y < centre + radius; y++)
				for(z = centre + radius; z < centre + radius; z++) {
					squareDistance = (x-centre)^2 + (y-centre)^2 + (z-centre)^2;
					/*if(squareDistance <= (radius)^2)*/ w.setBlock(x, y, z, Blocks.bedrock);

				}

 

But nothing is seeming to generate at all

[TheGreyGhost]

Don't worry about being efficient.  Java is so fast you'll never notice the difference no matter how much you optimise it.

 

I used to think that, but I've been doing a lot with structure generation lately and found that with very simple code (just loop through an array and place bocks indicated) it can create perceptible lag on structures of any decent size.  I actually changed my code to only do one layer per tick in order to improve this.

 

I think a sphere the size he's showing could actually show a bit of lag.  Generally though that won't be too noticeable if it happens during chunk generation because players are used to some lag from that, but if you generate the structure in response to player action (like using a tool), it can be noticeable.

 

Anyway, I was surprised that it could lag, but I think the thing is that the volume of a structure gets large very fast -- a sphere as large as shown in the OP's picture has 5000 blocks!  And I think the block placement requires fair bit of processing as well as client server sync notification.  Just giving my observation...

Hi Jabelar

I agree with you, it will take a long time to place that many blocks.  I'm just about to release a mod that copies up to 256x256x256 blocks and it can take up to a minute or more.

 

What I meant was - I've seen other folks try to optimise out some of the "squareDistance" checks.  That's a total waste of mental effort.

 

-TGG

 

 

[TheGreyGhost]

I tried this just for a normal one:

 

int radius = 8;
		int centre = radius / 2, squareDistance = 0;
		for(x = centre - radius; x < centre + radius; x++)
			for(y = centre - radius; y < centre + radius; y++)
				for(z = centre + radius; z < centre + radius; z++) {
					squareDistance = (x-centre)^2 + (y-centre)^2 + (z-centre)^2;
					/*if(squareDistance <= (radius)^2)*/ w.setBlock(x, y, z, Blocks.bedrock);

				}

 

But nothing is seeming to generate at all

 

Ah, I think I might have confused you.

 

When I wrote ^2 I meant 'squared', but that's not what the ^ operator does in Java.

 

Try this instead

squareDistance = (x-centre)*(x-centre) + (y-centre)*(y-centre) + (z-centre)*(z-centre);

 

You've got a typo here which is probably causing your problem

 

for(z = centre + radius

 

-TGG

 

[The_SlayerMC]

So now i've got this:

 

int radius = 8;
		int centre = radius / 2, squareDistance = 0;
		for(x = centre - radius; x < centre + radius; x++)
			for(y = centre - radius; y < centre + radius; y++)
				for(z = centre - radius; z < centre + radius; z++) {
					squareDistance = (x-centre)*(x-centre) + (y-centre) * (y-centre) + (z-centre) * (z-centre);
					if(squareDistance <= radius) w.setBlock(x, y, z, Blocks.bedrock);

				}

 

And it's still not making any sort of generation, even if i remove the 'if' statment

[The_SlayerMC]

Wow, i derped really hard ahhaha....

 

So now im using this:

 

int radius = 8;
		int centre = radius / 2, squareDistance = 0;
		int x1, y1, z1;
		for(x1 = centre - radius; x1 < centre + radius; x1++)
			for(y1 = centre - radius; y1 < centre + radius; y1++)
				for(z1 = centre - radius; z1 < centre + radius; z1++) {
					squareDistance = (x-centre) * (x-centre) + (y-centre) * (y-centre) + (z-centre) * (z-centre);
					/*if(squareDistance <= (radius) * (radius))*/ w.setBlock(x + x1, y + y1, z + z1, Blocks.bedrock);
				}

 

And its generating a big square

[Eternaldoom]

You commented out the if statement.

[The_SlayerMC]

With it there, it generates nothing. Without it, it generates a square..

[TheGreyGhost]

Hi

 

I've sent you a PM

 

-TGG

[The_SlayerMC]

Replied

[Villak]

If anyone needs a working one for 1.18:

int radius = 9;
int centre = radius / 2, squareDistance;
int x1, y1, z1;

for(x1 = centre - radius; x1 < centre + radius; x1++){
	for(y1 = centre - radius; y1 < centre + radius; y1++){
		for(z1 = centre - radius; z1 < centre + radius; z1++) {
			squareDistance = (x1 -centre) * (x1 -centre) + (y1-centre) * (y1-centre) + (z1-centre) * (z1-centre);
			if(squareDistance <= (radius) * (radius))
				context.level().setBlock(new BlockPos(x + x1, y + y1, z + z1), Blocks.BEDROCK.defaultBlockState(), 11);
		}
	}
}