The Core Of From Dust

[Cult Another World/Out Of This World creator Eric Chahi has returned to development, creating the ambitious From Dust for Ubisoft, and Brandon Sheffield sits down with him to find out the development concepts behind the impressive simulation that it offers.]

Having not worked on a game since the late '90s, notable French game designer Eric Chahi seemed to have disappeared into the art world -- becoming immersed in theater and photography rather than game development. But he has recently come back into the fold with From Dust [YouTube link], an ambitious new downloadable title.

The new Ubisoft-published title is a god game of sorts, in which the player manipulates the terrain, water, and lava in order to shape the world, guiding and protecting a tribe of humans who dwell there. It simulates the world in real-time, which creates a number of complexities -- and a lot of opportunity for discussion.

In this interview, Gamasutra delves into the underpinnings of the title, Chahi's tactics for dealing with the complexity of the simulation, and how players will interact with the world of From Dust.

In the game, I noticed there's a perpetual cycle of green and fire. The fire burns the plants, and the plants regenerate. That seems to happen in a constant loop. Do other things in the world do that? Was that planned, or is that an accident?

Eric Chahi: It depends on the spaces. We have spaces that don't start, that are resistant to fire, where fire will be slower. We have evolution of the environment, so the longer the place will stay without being stressed, the more it will be resistant because there will be more trees.

Normally, we wouldn't have an infinite cycle of fire growing by itself. It starts again because there is a source of fire somewhere.

But to answer your question, you are talking about a cycle of things. Cycles are an important theme of this game because it's a way to tie different environments. You have a tide cycle, so you can imagine that when the tide is high, part of the Earth is covered by water. When the water is going out, then plants can grow there again. The village can go there in the low tide, and maybe when you get the people to resist water, and stay there inside the high tide place.

And we have an eruption cycle -- volcanic eruptions. So you can have a cycle of eruption, plant coming, and then it's destroyed; this kind of loop. What is cool is that the player can anticipate the process. We have a desert environment where we have a kind of slow-moving ground with water moving slowly from one place to another. The player can play with this moving thing.

We have a high-level species -- I was talking to you about a kind of fern. Why does it take in water? There will be a special plant that will be a kind of tree, a big plant, that will be able to set up some fire or to help, to change, to create a lot of interaction between the environments. It will bring a new level of tool and of interaction.

I noticed when you're guiding the tribe from one place to another, it seems like the simulation is always pretty solid, but there's some potential for frustration with AI pathfinding.

EC: The player just controls the obstacles on the terrain. It would be very frustrating if the player can't anticipate what the plan is. I think the fact that the player knows where the person will go is a cool thing to anticipate. The tricky thing is when it changes a lot from one pastime to another. We are doing playtests around this. It's not a big issue, but we have to be careful of it.

Everything else seems so smooth so far; that's the only thing I see that could potentially be an issue.

EC: The cool thing is that, okay, maybe the AI will have a change of mind in moving this direction, because it's faster for him, but indeed we've put a lot of power in the player's hand regarding the terrain. He has the ability to react fast.

It will be frustrating if the player has a low grip on the mechanism. One year ago, when we didn't have this very direct interaction with the terrain, and when you were just asking for people to go there or there or there, you didn't have the power to change path of the guy. That was a really difficult issue at the time, but now it's resolved.

It sort of feels a little like a god game, plus tower defense, plus some other stuff. Did you actively take bits of different existing genres and stitch them together, or was it a natural evolution of your own ideas?

EC: We did not decide we want to take something like tower defense or take something like a god game; indeed, it is the opposite that happened. When we say there is a bit of tower defense, or a bit of god game, it is to deliver clues about the game. To give some context. Historically, at the beginning, it was not a god game.

What was it at that time?

EC: At the origin, it was a strategy game, but there was a background with the world, with a lot of erosion, some knowledge to preserve, and you are are giving power to your people. This is intact from the beginning, but the way to express it in gameplay has changed over time.

As I already said to you and as I said at GDC, the fact arose that the simulation was really malleable -- it drove us to direct interaction with the terrain. That's how the avatar, the kind of spirit of the tribe or whatever, has been input into the universe.

Then the default paths of the events -- the tsunami, the volcanoes -- that was in the original concept, to have events, to anticipate events, to have uncertainty regarding that there may be events, and of course to have cycling events. That was in the original concept, but at the beginning I didn't think about tower defense or the timeline.

It feels to me like the gameplay arose from the simulation you had created, and you were thinking, "Okay, we have this simulation. How does it become something the player interacts with?" It seems like it arose from there; is that correct?

EC: Yes, it is correct, but the simulation drove us to say, "Okay, how can we create more interesting gameplay to encompass the simulation?" Because we had erosion, but before the player could interact directly it was like something in the background. It was regarding the fantastic playable simulation, it was a step back.

Let's go to a metaphor. It's like the simulation was... you see the wall behind this window? Imagine that, maybe, there is a beautiful painting, and you can only see part of it. That was before the directed interaction. Then you break this wall, it makes a big window on this place for the picture.

So, then, the simulation that you created, the tools that you created for this -- it's kind of crazy to make that for one game. After this game is released, do you own the technology, or does it go back to Ubisoft so they may use it in Assassin's Creed or something like that?

EC: This technology is the property of Ubisoft because we created From Dust, but... They could use it for something else, but we are required to use it for From Dust.

What about multiplayer?

EC: Multiplayer definitely would work; cooperative or competitive gameplay. But it's a challenging part of the design -- not about the ideas, but more how to do the interaction. For example, on the same screen, or how to have the have the multiplayer ready for two people who are playing at a distance. It's a big thing.

Right now, we have a lot of work to make the solo; it took [a long time] to make all the foundation sorted, so regarding the project we can't do that. If we do that, again, it will be released more six months later. We must stay focused on the very precise elements. It will be for later, definitely; it is a must-have.

I would recommend online-only so each player only has their own screen so you don't have to deal with same-screen gameplay. I don't think you can make it work.

EC: I think I have solution. (Laughs)

Oh, you do?

EC: Yes, yes. It's good to be close to someone, and, when sitting together. When people are playing, the other is saying, "Okay, please do that." We can keep interaction, but then they are not really playing together.

If people have their own avatar, everybody can take matter and act together, maybe on the same screen. They can go for each other and the game world.

Yeah, that would be interesting for co-op, but for competitive it would be trouble.

EC: You suggested split-screen.

That would be really hard with the kind of crazy simulation you have, maybe. But maybe not. So with a game that has so much unpredictability, how did you ensure that the game displays things properly?

EC: Experiments. We can't do special cases in the simulation. It is very simple rules, and these rules -- there is a kind of convergence. So when there is a pattern, maybe there is a very loud pattern. Okay; it happens. But what we wanted to avoid is repeating patterns. These kinds of patterns we can identify just by creating a different extreme or different situation.

It's like rendering in a way. You create the situation, you let the simulation run, and you just observe how it evolves. If it looks wrong or weird or patterned, we just change the parameters and check if it's is okay. It evolved in this way; okay, this parameter doesn't influence this kind of pattern. It's really an empirical process.

When you were thinking about this, you were trying to think, if you died tomorrow, what game would you want to be remembered for? Do you think that this game is turning into that game? Is this game turning into what you really envision for that?

EC: Okay. You're saying, if I die tomorrow, would I be happy with that game and would it be a really...?

...fitting legacy for you. But you said it first!

EC: (Laughs) Well! Uh, not yet, because the game is not finished, of course!

But do you think it will become this? I'm not going to let you out with an easy answer. (Laughs)

EC: I think yes. Because the project evolved from my first idea, and I did not reach fully what I wanted because it is very difficult to bring it exactly. But what we have today is partly what I wanted and partly varies more into incredible things I did not expect. So overall, yes.

But if it's not fully the game for your legacy, I guess that means you can't die tomorrow and have to make another game!

EC: Yes! Okay. It is just -- which game would I want to do if I died tomorrow?

I know; I'm just giving you a hard time.

EC: (Laughs) So after this game --

-- you have to think which next game you might want to --

EC: -- I would want to if I die tomorrow. (Laughs)