- [Sébastien Bénard] Spartan boots is, it's a weapon in this game. It's not exactly a main weapon, it's more like a side weapon. And very useful if you want to kick an enemy off a platform, for example, or something like that. But most people actually don't like it. I think I have changed this weapon like maybe, hmm, 10 times or something. I thought it was funny, I thought it was a good weapon, but actually most players, they don't like it. It used to be a bad weapon. But the newcomers, they don't know this legacy of this weapon, and they actually think, okay, it's a fine weapon. 

I am Sébastien Bénard, I am game developer, I am lead game designer, or lead whatever, at Motion Twin. Because we are all leads, so that doesn't matter. And this is why we decided to cheat, to make the player believe he is actually in control. [dramatic orchestral music] [lively string music] 

"Dead Cells," so it used to be something really different before. Because it used to be a sequel, some sort of sequel of games that we made on the web. "Dead Cells" used to be a web game and a mobile game. It used to be a free-to-play game, it used to be a tower defense, and co-operative multiplayer game like every buzz world and [mumbles] game. The most important was the multiplayer. Actually, it was really not working right. And we made a single-player version with a basic tutorial and a small story just to explain the game. And actually someone, a fan of us, told us, maybe the single-player version that you did is actually nice, and it works. And maybe you should make a single-player game. And at first it used to be a joke. We were like, okay, no, you're kidding. And we thought about this, and we decided to remove completely the multiplayer aspect of the game. 

At first we said okay, let's make a traditional 2D game, 2D platformer, and let's make everything by hand. But the thing is, we have just a few people. We have developer, one graphic artist. How could we still make a huge world, and a huge place to explore but with so few people. Going to early access was like a difficult thing for us. Because we knew it, you know it was the time of the indie apocalypse and everything, and so everyone told us yes, that's not a good idea to go on early access. But we were running out of cash at this time. Many people told us okay, you will see, Steam early access is not that easy, and the community could be harsh. After one week or something, we had so, so many feedback from player. Because when they posted negative feedback it was like three page long, always. It was a huge post explaining why this thing sucks in the new game. And most of the time was, at first it was, oh, shit. But still, we have a problem here that we should address because yeah, this guy just want the game to be good. He don't want to bash the game or whatever, he just want to play the game that he will like. 

Those people, they want you to make a good game. They have all the interest into making this game as good as possible, just like us. The speed running community was really a very important part of the game right from the beginning. We knew that they would find bugs and glitches to exploit. For example, in Castlevania you could use this back dash move which allows you to move a little bit faster than just walking. So you, that's the kind of small glitch that you could have. When we had our very first glitch that allowed the player to jump over above a single boss and to get to the exit directly, we asked ourselves is it something that we want to have. Is it something which is game-breaking, is it something that we could live with? So that's something that we don't want. 

We don't want the experience to be changed for most players. We want the experience to be changed for just the speed running community. And sometimes you just have to say, okay, this thing is not really such an issue. It's funny, they are, like, moving super fast. But yeah, why not? And it's not breaking the game in any way. 

So we went to alpha, and the first version was really, like, uh, boring as fuck. [soft piano music] 

Level generation is some very tricky part of the game. Because, you know, it was a time of "No Man's Sky" and everyone was, like, traumatized by huge levels with no actual meaning. The real problem with random generation is all about getting good results. That's really complicated to explain, in some way, to a computer, what is a good result. Because that's something that actually takes lots of intuition, lots of feeling. That's not something that the computer can actually understand. So making sure that you translate these rules into some kind of algorithm, that's the difficult part. Making the generator, it actually took one week or something. And we, after maybe one month, we didn't have always on point good results. And on, during the same amount of time actually, someone on the team tried to make rooms by hand. And in about two weeks, he had made like one thousand rooms. He had already one thousand good results. Because all these rooms were made by hand. And that was really a very important moment for us. Because it was like, okay, instead of working on this level generator algorithm for one month, let's just make rooms for one month. And it will add much more quality to the game and when we build a level, we know there is one entrance, one exit, and in between you have like, maybe three combat rooms and a shop, or something like that. And for each part, we just pick an entrance inside the dictionary of entrance rooms. And we just pick one, we put it in the world, and then we have combat. 

You have this huge dictionary of combat rooms with every possible shape: long ones, large ones, or small ones. And each room it's just a basic map of collisions of walls. You have exit and entrances, and platforms. This part is actually what we call brute force generation. We just pick something, it doesn't work, we cannot place it, we discard one, we take another one. It's really all about thinking, is it faster to make all the content by hand, or to make, like, 1000 element, different element, by hand, or is it faster to actually write a generator and to make sure that it always works. So it's really some things that we learn along the way to, how to decide what should be generated and what should be actually hand-crafted. [dramatic orchestral music] 

If you've played a very old game, like the first "Prince of Persia," maybe most player that did play this game would remember that when you want to jump from a platform to another platform, you have to anticipate, like, a lot. You have to press your jump button maybe two or three meters before the end of the platform, because the character animation is really slow, and you will have to anticipate and to start the jump. And in modern days, games, you don't do that any more. You want to be able to press the jump button just in time. Just when you are about to fall from the platform. But if you try to do that, actually, you fail. Most of the time, you just fail. Because you press the jump button a little bit too late.

- [Man] I hate that part here, okay... Yes. No!

- You have this sense of fairness which is really important in permadeath-based game. Because when you die, you should always feel, okay, this is my fault. I did something wrong, it was obvious, I should have avoided this enemy, or maybe I should have checked before jumping down. Because we wanted to make sure that at no point you could blame the control for your own death. Because you, actually when you die you lose maybe 10 minutes of play time. And that's really a difficult moment. And we don't want you to feel, like, bad for that. [dramatic orchestral music] 

Let's talk about the jump, for example. You, that's a very basic thing that happen in the game. You have two platforms, you want to jump from point A to point B, so it's a very basic situation. And there are some things that, those happen like a lot in this game. When you start the jump, one time out of two you are probably out of the platform when you press your jump button. Like one frame too late, and you are starting to fall. And in such a case we actually allow you, when you are off the platform to consider you are still on the platform for maybe two, three frames after. So we allow you to jump afterward. For example, when you arrive on the next platform, um, most of the time, also, you are maybe five pixels too far from this platform. And in such a case, we have tons of scenarios that we just put you on the target platform. You should have missed it. But we don't do that, we don't allow player to miss this platform because it's not fun just to, you know, just to arrive on the platform to fall just in front of it. Nothing, no one wants that to happen in a game. It's really just not fun.

 That's something that did happen in old games. So now when you make a platformer, yeah, you should take care of the player. We don't want this to be in the way of the player. We just want this to be part of the gameplay. But not a challenge. Not a specific difficulty or something like that. But when you get an enemy in the way, you have to figure out, okay, this guy is strong from the front and can I get back behind, and can go behind him, maybe he is about to shoot, so we have to crouch, or to jump, or do whatever. So in such a step, we want you to think about that. We don't want to think about the fact that your jump will probably fail or you won't be able to get past this guy or anything. So yeah, it's really all about figuring what the game is all about and putting energy where you actually matters. [slow synthesizer music] 

Some things that we noticed that didn't, we didn't really expect at all, we saw that in the reviews on Steam afterward, most reviews, I mean, like really, maybe 80% of the reviews, something like that, they do talk about the fact that the game is easy to play and very fast and easy to control and everything. They don't explain exactly why, but they feel like okay, it's really dynamic and everything and then they talk about the roguevania thing of the game. You know, they talk about the actual gameplay. And that's some things that we actually noticed, the player really, very interesting and very invested into the game just because the controls are right for us. And we know that now the moment you give the controller to the player and let him play the game, this single, maybe five or 10 seconds is, like, really important. Just to make sure that there is nothing that will get on the way. It should not feel wrong right away. That's something, and this part really takes a lot of time, to make sure that nothing bad happens in this moment. So taking care of the controls, that something that we definitely do, yes. For any future project, we'll do. [peaceful electronic music]