World Generation

[eleazzaar]

A sphere is simply impossible to do for a regular voxel world.

Yes, i agree and understand.

Here are some alternate shapes to consider...

- A double-sided world with a variable/reversed gravity gradient (digging through the world would slowly reduce your weight, you'd be weightless in the center, and digging "up" into the obverse side of the world would take you to a separate landscape.

That has a certain appeal, but you would need to have a way to move in zero G, otherwise you would be stuck.

For more coolness, make the other side of the world totally different-- i.e. as different as the nether, an opposite world or something.

These shapes can have one set of edges joined in the shape of a ring, or a Möbius strip. Alternatively, both sets of edges can be joined together in a torus or Möbius (single-sided) torus.

Hah! that's very geeky.
As amusing as a mobius strip would be, it would pretty much invalidate the compass. E & W would be consistent, but N & S would be two names for the same thing.

 

[ironchefpython]

As amusing as a mobius strip would be, it would pretty much invalidate the compass. E & W would be consistent, but N & S would be two names for the same thing.

In Ringworld, those directions were referred to as spinward and anti-spinward. They are not the same thing, although two entities travelling in those opposite directions would meet at the antipodal point on a ring or strip. But the actual directions would be consistent in orientation and separation regardless of your position in the world.

 

[eleazzaar]

As amusing as a mobius strip would be, it would pretty much invalidate the compass. E & W would be consistent, but N & S would be two names for the same thing.

In Ringworld, those directions were referred to as spinward and anti-spinward. They are not the same thing, although two entities travelling in those opposite directions would meet at the antipodal point on a ring or strip.

That part doesn't matter. To all practical purposes of a person on the ground E & W, or SW & ASW operate the same way on a globe or ring world. I'm not concerned about that.

As to N & S, you are right, i was thinking about it the wrong way. Viewed from the outside, any edge you designate as "N" will eventually will include both edges if you follow it long enough. But the player won't view it from the outside.

Viewed from the surface, an edge can be designated "N", and though it eventually will merge with with the side designated "S" from the other side, this doesn't effect the map. If you imaging walls on the N & S sides of the mobius ringworld, you can run your hand along the N wall all the way around with out ever touching the S wall.

Now that i see it more clearly, there's not much to recommend a mobius world over a ring world. In practice, they are pretty much the same, except on a mobius world half the time you can't look into the sky and orient yourself via the arch (which i consider a cool advantage of the ring world.

And from a plausibility standpoint, it's hard to imagine why someone would make a mobius ringworld

 

[ironchefpython]

Now that i see it more clearly, there's not much to recommend a mobius world over a ring world.

Well, leaving out the contents of the skybox, the topological difference between a Möbius strip and a ring are the number of sides. So you could tunnel through a Möbius world and emerge at the antipode of your starting position. Tunneling through a ring takes you to the other side of the ring.

 

[Cervator]

I'm now tempted thinking about the pre-generated "big world map" in low detail idea + a ring world that can show the big world map in the sky and the chunkified local world nearby....

 

[dei]

One question: What target-group do you have for your game? Geeks? Quantum physicians? Do you really think a Torus-World is the type of world where every player wanted to play in and is able to imagine how this could be possible?

If you derive from the standard (our spherical world) and its obvious to the player every successful game has to set a story behind it that explains why it is not as we human beeings are used to...

Keep it simple guys! Flat is ok, I don't expect many Gallileo Galilei's to play Terasology.

 

[eleazzaar]

I'm now tempted thinking about the pre-generated "big world map" in low detail idea + a ring world that can show the big world map in the sky and the chunkified local world nearby....

Don't forget that sometimes the player will be at the top of a tower and should be able to see the whole thing. Beyond the edge of the fully rendered zone blocks should be seen nearly from the side. My fear is that the transition from thes edge of the local world to the ring would look pretty bad even if you spent a lot of processing time on it. You gotta cut somewhere. Simulating AND rendering the whole world even at low detail seems a bit much. Unless somebody has a clever very efficient trick up their sleeve...


Dei:
Sure, a flat world is fine.
But I think you underestimate the ordinary players tolerance of or interest in less conventional world types. Minecraft has the nether: a parallel dimension at 8x scale, which is a pretty abstract idea-- not at all like our real world. It doesn't seem to have hurt Minecrafts popularity. Also none of these ideas require that the player understand the geometry of the world-- unless he wants to explore the whole thing.

Of course the ultimate considerations IMHO should be:
1) what does it add to gameplay
2) is I practically possible

 

[overdhose]

whoa

let's keep things friendly, it was merely an idea, and I wanted toss it out there because many people here seem to have tinkered a lot more on issues like this then i ever have, so I wanted to get some feedback and decide for myself what might possibly something interesting along the way, be it in terasology itself or a branch of it, or even a personal experiment out of curiosity.

It was not my intention the criticize / replace the current way terasology does things, when I see the direction this thread is heading atm, I feel kinda bad to have started it in the first place.

de gustibus et coloribus non disputandem est (if I still remember that correctly), so let's not start disputandem please. I like some of the insight and pitfalls mentioned so far, let's just keep this an idea for the moment and not an actual viable option, even the intial work of porting code and examples was quite taxing to me, there are better / other things to spend time on atm.

 

[Cervator]

I just see lots of good discussion, no real negative arguing or anything. But then I'm also generally a hopeless optimist

It is a meaty topic and most the stuff here won't make it to outright implementation. None the less I feel richer with the discussion having taken place, so thanks for kicking it off

 

[ironchefpython]

Do you really think a Torus-World is the type of world where every player wanted to play in and is able to imagine how this could be possible?
...
Keep it simple guys! Flat is ok, I don't expect many Gallileo Galilei's to play Terasology.

I don't know if you remember simple games like Pac-Man, or Defender, but those were played on cylindrical worlds, and Asteroids was played on a torus. I doubt an average player would have difficulty understanding or adapting to those topologies.

 

[eleazzaar]

I'm not the least bit angry or offended, and i haven't got the feeling that the other participants are either. I think discussing the implications of these geometrical oddities is fun.

Dei's concern is a legitimate thing to raise, though probably based on a misunderstanding of how confusing/complicated any of these things would be to the player.

I don't expect all these things to be implemented. I believe implementing any of the non-rejected ideas would be pretty simple, with the exception of rendering the ringworld in the skybox.

 

[Cervator]

*quietly continues to sit over in the corner drooling over the sheer idea of a ringworld rendered in the Skysphere*

That does bring one semi-related thought to mind - adjusting the "big world map" with changes made at the local scope (just for an in-game world map, doesn't need to be fancy). Not likely to be typical (or even hard), but there'll be players insistent that that durn island over yonder is interfering with their otherwise perfect view.

Yeah, don't mind me, I'll be going back to drooling now

 

[ironchefpython]

*quietly continues to sit over in the corner drooling over the sheer idea of a ringworld rendered in the Skysphere*

Unless you can travel faster than running speed, you can render the "arch" of the Ringworld as a static graphic, considering it would take 23,000 years to walk completely around the Ringworld. If you want to have perfect accuracy, you can re-render the perspective of the base of the "arch" every 100,000 miles the player travels, or once every three years of continuous movement.

 

[eleazzaar]

*quietly continues to sit over in the corner drooling over the sheer idea of a ringworld rendered in the Skysphere*

Unless you can travel faster than running speed, you can render the "arch" of the Ringworld as a static graphic, considering it would take 23,000 years to walk completely around the Ringworld. If you want to have perfect accuracy, you can re-render the perspective of the base of the "arch" every 100,000 miles the player travels, or once every three years of continuous movement.

It doesn't have to be, and probably shouldn't be that kind of centered-on-the-sun, and 1 AU radius ring world -- A.K.A the classic Niven Ring. That's not a limited map in any practical sense (for file size or player exploration).

What's needed is a "little" ring on the scale of a planet or more practically, a small moon. Like the "Halos" in Halo. Thus the world would be small enough, to maybe keep the whole thing active, even where the player isn't.

 

[dei]

Another perhaps dumb question:

Why not simply 'teleport' the player from one end of the map to the other (like the snake game on nokia portable phones). This way directions on the compass are always the same and the player can walk around the (semi-)world.
Of course the edges have to be matched by the level generator so there is not an wierd cliff you fall off when surrounding the world.

One point I think I didn't got yet (sorry if I'm not as agile in 3d-geometrics as other devs):
Why it isn't possible to place the magnetic north-pole for example in the middle of the generated world (completely regardless of the coordinate system)? In my understanding compasses would be wierd on exactly this spot (could be overwritten with a big red dot or a cute polar bear if you like). This way you always find back to the center of the world and departing from that point you would have one clear north direction (and with it a south in the opposite direction). Of course if you go over the world's edge and get 'teleported' to the other edge there would be a dark-spot too (which could be overwritten with a cute penguin ). (some simple vector-geometric functions should be able to calculate the compass directions)

OK, we would have an increasing distance (going from north) to surround the world on every longitude which isn't very realistic but apart from some eskimos or physicists playing the game noone would feel unconfortable with it or even notice it. (the same could be duplicated on the 'backside' of the flat map, to increase the distance nearing the equator, resolving the problem with the surrounding distance) Is this the only thing that annoys you? Or is there some visual distortion that can't be calculated in an efficient manner (if we dont cut the corners and have some kind of 'squarry' earth on the equator)?

Sorry if my question is a little bit retarded, I like to be enlighted...
(I know that north-south directions would leave the orthogonal block system, but who cares?)

 

[ironchefpython]

It doesn't have to be, and probably shouldn't be that kind of centered-on-the-sun, and 1 AU radius ringworld. That's not a limited map in any practical sense (for file size or player exploration).

What's needed is a "little" ring on the scale of a planet or more practically, a small moon. Like the "Halos" in Halo. Thus the world would be small enough, to maybe keep the whole thing active, even where the player isn't.

A Banks orbital would take about 285 years to circumnavigate at walking speed, so you could adjust the skybox every 22 hours of gametime and still maintain pixel-perfect accuracy.

Still, I adore the idea of a massively multiplayer voxel-world using the scale of a Ringworld. You could have 61 trillion players each with 25 square km of land to build in. (Yes, I realize that represents enough excess capacity to handle the entire population of earth 10,000 times over. But if we meet lots of aliens, I want them to be able to log in and play too!)

 

[ironchefpython]

Why not simply 'teleport' the player from one end of the map to the other

That's basically what happens on a torus, except that from the viewpoint of the player, they keep walking through the world without interruption.

Why it isn't possible to place the magnetic north-pole for example in the middle of the generated world

Yes, you can define any arbitrary point on a isometric grid mapped to a torus, and identify an antipodal point, and at any time you can calculate the direction of the unit vector of the shortest distance to each point (which will always be 180 degrees opposed), and display that direction on a compass. You could even give that direction a name (I would suggest spawnward and anti-spawnward). However, these don't have to exist exclusive of orthogonal North/East/South/West cardinal directions.

 

[Immortius]

Interesting article:

The difficulties of an infinite video game world

 

[B!0HAX]

Really nice article, makes me think of many interesting cases...
One I would like to share :
·Infinite world vs. finite worlds connected to an "infinite*" nº of worlds. (*Stargate could be an example)

Players could learn how worlds are structured if they are finite, infinite is difficult to handle...
If they wanted infinite exploration they could have a way to explore an infinite number of worlds.
But then it all depends in the aims of the foundatinal structure of the gameplay experience intended.

 

[eleazzaar]

Really nice article, makes me think of many interesting cases...
One I would like to share :
·Infinite world vs. finite worlds connected to an "infinite*" nº of worlds. (*Stargate could be an example)

Players could learn how worlds are structured if they are finite, infinite is difficult to handle...
If they wanted infinite exploration they could have a way to explore an infinite number of worlds.
But then it all depends in the aims of the foundatinal structure of the gameplay experience intended.

I agree with one of the commenters in the article-- Theoretically infinity sounds cool, but what makes it interesting in practice, is being able to go out and discover new and interesting things. An infinite amount of very similar terrain isn't very interesting.
A large number of finite (small enough than non-compulsive people could thoroughly explore them) world-- each with different world-gen parameters, and maybe different animals and plants, would perhaps better provide the player with new and interesting things to see. Each worldlet could have thus have a distinct personality-- to a degree that would be difficult adjacent biomes on a single large world to exhibit.

Speaking of stargates, image if each of these worldlets had a random number of gates/portals on it, say between 1 and 4, and each portal/gate was linked to a single gate/portal in another worldlet. That would give you something to explore for.

Possibly-- (i don't know if it is technically feasible) similar portals could be built to link different people's worlds with their friends worlds, in an ad-hoc kind of multiplayer.