Eric Chahi on From Dust, Peter Molyneux and what's next

The simulation. It demands a lot of computation, so we had to do a lot of optimisation, optimisation that uses the power of today's computer. Richard Lemarchand [Uncharted 3 lead game designer] was talking about the PS3 and the Cell. Typically in From Dust you are using this kind of architecture to have it running at the right frame rate. We have a big grid in the game, and each cell computes all the vegetation, and so there is a hundred thousand to around 200,000 cells to compute for each frame.

This has been coded not in C, but in a micro language close to the VS Assembly language. It takes advantage of processor architecture where you have a cache memory, so the computation is done so that it stays in the cache memory and can compute things very fast. This is for the behaviour of the simulation.

Aside this we have the visualisation of the simulation. Doing visualisation of something highly dynamic, like all the terrain - you can have lava, rocks, and all this mixing - the next frame it can appear, disappear. So all the shaders have been designed to represent this dynamic world, and especially all the fluid movement. We have special texture scrolling so it follows the direction of the flow. But we can't have a texture scrolling for a long time on some vertex, or else it becomes totally distorted and it's stretched too much. So we have to reset it, but reset in a way you can't see. It's really complicated to manage.

There was complexity regarding the algorithm, and complexity regarding the visual art, so everything stands out for the player. The player must understand when it is rock, when it is sand, when it is vegetation, when it is water, when there is not a lot of water somewhere. We wanted this to be beautiful.

"We did not code a river algorithm or a volcano algorithm. It's just a lava emitter and then it flows with the rule of the simulation."

I'm going to go back to the simulation side, because the world evolves. The landscape changes, where you can have a river being created, you can have a volcano become bigger and bigger and bigger. All this is emergent. We did not code a river algorithm or a volcano algorithm. It's just a lava emitter and then it flows with the rule of the simulation. It took a long time to tweak it so it gave a beautiful result but is suitable for gameplay.

There is always something we would like to do, but we didn't have the time or was too complicated to make alive.

Yes, there is something regarding the people. The game is a game about erosion, about the passage of time. We have it on the landscape and regarding the story, with the recovery of memories and knowledge. But at the beginning of development we wanted to have this on the people so you can see a child growing, becoming an adult, becoming old and going to dust. The passage of time would be viewable on every level of the game. But that was too complicated in terms of gameplay and game mechanics, so we preferred to leave it. It was complicated also to have a correct animation, to have them morph from a kid to an adult.

One side we could have developed was the animal side, and have more wildlife. But aside from this, we crafted a game so it was working well. I'm very happy with the result, especially regarding all the simulation. I love when the game rules merge with the meaning of something and with the gameplay. Here we have the simulation used fully for the gameplay and it expresses something about our world, about the relationship with the forces of nature. The team did incredible work on this.

No. I was hoping the game could have a good reception, but you never know.

Yes. I read many of them, especially to see whether they understood the part of the game we wanted to communicate. Sometimes yes and sometimes no, of course. But I had the pleasure to see it was often understood.

There were some expectations regarding the god game side. Some people were disappointed by this because they were expecting a full Populous or something like that. But some people understood the game.

It's true there is some weakness with pathfinding sometimes. But when you're playing the game, you can see all of the weaknesses or you can see what is really original. I'm not criticising anybody. I'm just saying it's really personal and I'm happy.

I can't tell you the number because Ubisoft don't want me to. But it sold very well. That's good because we can push them to create more original things.

Yeah! It would be cool if more games like this came out. I would love to see other games with this kind of simulation. Maybe Peter Molyneux will create a new game. He played From Dust. He went to Montpelier one year ago. We showed him From Dust. It was not final but he loved it. I think, maybe it stimulated him. I don't know.

He gave us a true compliment. He said it is the first god game he's played.

I was really surprised. I was thinking, maybe he's over-complimenting. But well, I understand why he said that - because of the plasticity of the terrain, and he worked on Populous, where you can interact with your terrain. In Black and White there were some volcanoes, but it was for a limited time and a limited area. He was very kind. As a fan of Populous and his work, I was very touched.

From Dust was hard work. You have to fight to make this kind of game alive. The team did fantastic work. There was a lot of perseverance. Before it existed some people didn't believe. Some people believed.

It took time but it was not so hard. It had a good reception since the beginning, when I showed them the concept first in 2006. But between 2006 and January 2008 was spent just meeting the right people, because it's not one person to convince. It's several people. Then we had to find the right timeframe so a team is available at Ubisoft to do the game.

Then, of course, when the top management is okay, well, there are many people in Ubisoft, and you have to work with them. There is a stage in game creation when everybody believes. When you're creating something you are exploring. Sometimes you find a dead end. Some people can start to think, ah yes, but will it work?

Helpfully, the top management, Yves Guillemot and others, always believed in the game. They were always supporting us even when it was more fragile. So thanks to them to support the game until the end.

It was very important to talk about the game to the public one-and-a-half years ago. In 2010 we gave some trust to Ubisoft. Sometimes it can save a project, to expose it to the public. It is the reception from the public that can give trust to a publisher.

It helped, yes.

I've finished my work on the iPad and iPhone version of Another World. After From Dust finished I worked on this. Now I'm free for something new.

Not precisely. I want to continue to create systemic worlds and complex systems so that while you're playing you interact with it. A system with emergence. But precisely, I don't know yet. I know I want to create and to work alone, at least at the beginning. Then, when everything is prototyped, I will see if I do the whole game creation alone or not. But at least at the beginning I want to do it alone, because sometimes describing something on paper is not the best way to convince, especially if it is different and original. You have to implement it. And it can be wrong. Implementing can give you new ideas. That's why I want to work like this.

I would love to do another From Dust, but right now there is no plan. And also, I have a strong desire to create something very original. Not ambitious in terms of content, but really focused on bringing something new and fresh.

I have a couple of directions, but I need to clear my mind and start to code.

I've only had one break during my career. It was between Heart of Darkness and From Dust. But before I never took a break. After Another World I started Heart of Darkness. Okay, the development was long, but there was no break. That's why it was so hard. Before Another World I'd just finished Future Wars. A month later I started development of Another World. Since the end of the eighties - I started in 1987 as a full-time game developer - I've never stopped. So it's curious. This big break gives the perception that I'm taking breaks in between each game, but no.

Yeah. My nature is to be independent. The perception could be, well, what is he doing now, since I'm not inside a company. But I'm working at home.

I would like the project to be relatively short - something between one year and two years, maximum. One year would be great, but knowing me, well, I'm going to plan one year and it will end up as two. I have a ratio - it's 1.5, 1.6. And after that ratio there is another ratio to polish. So a factor of two is realistic.