The Art Direction of No Man's Sky

[Music] Okay, first of all, before I begin, could I see a show of hands? Uh, how many artists are here? Phew. Okay. Uh, and how many of you wish you were stood here in front of all these people looking right at you at this very second? Yeah, I didn't think so. So, my name's Grant Duncan. I'm the art director at Hello Games. We're a very tiny little indie studio in England. And I'm very aware that I shouldn't be here right now because I'm working on a game called No Man's Sky. And the reason I shouldn't really be here is because first of all, the game's not finished. It's always a little bit strange talking about things like art on a title that's not finished. And obviously, the fact I'm here means the game is not getting finished and so it's getting delayed and all this kind of thing. But I've been to GDC in the past few years and I'm one of

those people that always moans about there's not enough art talks, there's too many tech talks, everyone's talking about freeto play and all that incredibly boring stuff. So when they asked me to come, I couldn't really say no. So I'm going to talk about No Man's Sky. So, No Man's Sky is a science fiction game and it's all about that kind of feeling of discovery and about exploring the unknown. It's that kind of feeling that you had when you were a kid and you'd go to the beach and you just you'd be surrounded by sandjunes and you wouldn't know what was over the next hill or you'd find a rockpool and you'd go poking it with a stick and a little crab would crawl out and everything was new and everything was exciting. And it's inspired by fantastic artwork like this, science fiction covers from the 70s. And why I love this is because it's not, you know, it's not a great piece of artwork, but it's really weird. Like, what are they doing? Who are those

guys? Is that like some kind of jacuzzi? I don't really know. And the perspectives are all off. some kind of blue water, orange mountains, but it's just you just wouldn't have this these days. Nobody would buy a book with this on the cover anymore. If games had this kind of artwork, I don't know if I'd sell many copies, which is a bit worrying since I'm working on a game with this kind of artwork. So, I'm going to spend the next few minutes talking about just kind of being self-indulgent, talking about my favorite sci-fi artists. This is a guy called Chris Foss. And this is actually some of the first sci-fi books that I got into. The main reason I got into it is because they had one of these cool three-way covers, which just means I'm guaranteed to buy all three because of my OCD, and I just need to have them arranged perfectly on the shelf. Anyone who's played Home World, you'll look at these ships and you'll realize that they were clearly inspired by the same kind of artwork that's inspired me.

And what's great about Chris Foss is that he actually began his career as a erotic artwork creator. I think he worked for Penthouse. He illustrated the joy of sex and apparently his understanding of the female anatomy is what's led to him creating such incredible designs. This is uh why this is a perfect example of sci-fi art. It's like what even is this? That's water? Is it clouds? I don't know. But it's it's absolutely incredible to me. Like I can't even imagine what it must feel like to go to bed after you've just created that piece of artwork. I don't think I'd be able to sleep. This is another one. What I really love about him is he created this kind of art when everyone else was creating black starfields, gray dull monolithic spacecraft. And he'd created this and people loved it. And I can't imagine somebody just taking that kind of risk and just doing something so

different to what everyone else is doing. And I think that's why he's one of my favorite artists. This is another guy, John Harris. He's a little bit different. Like he he is quite formulaic in that he'll have it's always a sort of flat horizon line, but he just has like the the image on the left that could be Kenya. You know, those trees that could work. But then you've got this giant building. Like how big is that building? It doesn't even make any sense. The same on the right. Like those people are so tiny. It's just and the size of that planet. It wouldn't even work. But it's just sci-fi. And it's just beautiful. And what I like about these is that these vessels, they could be military vessels. They're kind of bristling with what could be guns or lasers of some sort, but it's the softness that he kind of gets with his brush strokes. And it just go I can't even talk. It's so good. Like the green he's using on the left is just everyone else is using black in space. He's done a giant green space and that blows my mind.

Another one who's slightly more unusual, but I think every artist in this room probably loves this guy and we were all gutted when he died a few years ago. But Mobius, he kind of walks that line between science fiction and fantasy. What I love about it is that it's it's a place you want to be. So much science fiction, especially in games, they're these horrible post-apocalyptic war zones and there's not really everyone's just trying to kill each other in different ways. And every game that's set in science fiction universes seems to be like you're in this amazing world and you can kill each other there and you just you're in this spaceship and you kill each other here. And it's there's something about Mobius's work that nobody's killing each other. They probably could if they wanted to, but they're not. And I find that quite refreshing. And things like this where you've got they're kind of whales, but they're also sharks. I don't know if they're underwater, but it's just filled with questions and it you want to you want to

know the history. You want to know like that guy's on like a flying duck pterodactyl. You want to know what they're doing. those those diploidocus snails on the ground. You want to you want to fly down. You want to understand what those creatures are and what they're doing. And this image is just classic sci-fi. It's a little man and some kind of giant interesting thing. And that is just something I've been doing far too much in No Man's Sky and getting away with murder. And I didn't really want to include Ralph because it's almost too predictable. But you can't you can't pretend that Ralph McCroy wasn't a huge influence. Like his work on Star Wars and like X- Wings, Tie Fighters, it's that thing with that all great artwork has, especially art directionwise, where everything fits together and it just is coherent. And even though the ships and the universe,

they're different and they're not like our universe, you don't question it, you just accept it. And this is some of the work that's not as well known, but he just has that same sort of fantasy sci-fi softness. And I think this is actually was one of Ralph McCrory's favorite pieces, which this has. It could be Earth, but you've got twin sons. You got a mysterious alien landscape. So, to me, science fiction, it's not about war. It's about that feeling of the frontier, that feeling of adventure. It's not post-apocalyptic. It's exciting. It's full of hope. I also thought Star Trek, even though I've never been a massive fan, it had that, you know, humanity had done okay. We'd managed to get to a place where we just had wonderful technology and there was wars going on. There were enemies, but

as a species we were doing really well and it was a place you wanted to be. And another inspiration from No Man's Sky is this book, which I read when I was a tiny little kid, and it's it's great. It's that classic adventure sort of thing where you've got, if you look at the cover, what are we seeing? We've got a pterodactyl in the sky. Got adventurers taking photographs. We've got Iguanadon in the center there. Stegosaurus. But if you look up on the tree, some kind of wild Tarzan man. In the bottom left, we've got angry natives. And if you look in the very middle, you can see this like a giant alligator fighting a some kind of prehistoric fish. And I just I love that. And it's just books like this and movies that came around at the same time. Journey to the Center of the Earth, adventurers, explorers going off into the unknown, movies like The Mysterious Island or Simbad or The Odyssey, where it's even though they're pioneers going to islands, there's a definite crossover

with sci-fi, whereas rather than going to an island, you're going to a whole new planet. And obviously these movies are quite cheesy. They'd have giant crabs and things like that. But it it definitely captured that feeling of especially watching them as a kid being like, "You've arrived at this island. Anything is possible." I remember when we first announced No Man's Sky, it was incredible how many developers and artists came up to us afterwards and they were all saying like, "Ah, that's the game I've always wanted to be making." Or, "I started working on a game a bit like this." And it's that thing that we're not pretending like No Man's Sky is this idea that nobody else has ever had because everyone has had this idea. It it's just ever since you were children probably if you've been thinking about working on artwork that you'd build a world and then you populate it with your creatures, with your characters. And No Man's Sky is the ultimate art project for that. That I can create a planet and

I can do whatever I want. And that is absolutely terrifying. So this is a piece of key art we created quite early on. Hopefully you can see a lot of the influences of what I just talked about like Ralph McQuary, Chris Foss and all the lost world and the journey to the center of the earth, mysterious floating islands, the cresant planet on the horizon. And we create this kind of artwork almost as a as something we could as artists we can all look at and agree that this is the vibe we're going for. So, we wanted that feeling of a world you want to explore, a world that you don't know about, and a world that has history. It's very easy to kind of forget about history on a planet that is a jungle, but it's actually really important. Like when we first started creating environments, you just add trees, rocks, bushes, and things. But unless you can see where

they came from, your brain kind of rejects it. So you have to have little hints that you know, younger trees, signs that like ivy and hanging branches, just all these little hints that it hasn't just popped up, erosion has been involved. various sequences have occurred. So with that in mind, the very first thing we did was that we started with terrain. No Man's Sky is one of these projects that I can't I'd love to take full credit for, but it is incredibly coder orientated in that I have had to work very closely with programmers and since we're a room full of artists, you can imagine what that was like. Oh my god. But I spent the first year because No Man's Sky began as like a little prototype within Hello Games. So there's four of us working on it, me and three coders, and we were we stuck ourselves in a little room together and noodled

around. And it started with a terrain engine, which Sean, our main code dude, he was um I shouldn't say dude any um he was in his bedroom basically and in his spare time he started working as terrain engine. And one of the things we really wanted to do was really focus on the real world because it's so tempting when you're working on a sci-fi game just to go straight in and make crazy alien landscapes and giant tentacle trees and all this kind of thing which we all love drawing. But what we tried to do was look at the real world and understand things like landscape and why things occur. And this is why I wish I'd paid more attention to geography because I vaguely remembered all these terms like glaciers and limestone caves and coastal erosion and all these kind of things which I now know I'm an absolute expert in now. So what we tried to do was we're not simulating these things

because doing that in real time would be pretty crazy. What we try to do is get a similar result. So we have limestone caves, we have lava caves, we have wind erosion, coastal erosion, glacial erosion, and volcanic islands. At least we have things that appear to be those kind of things. I'm going to show you a little video of our terrain in action. So what you're seeing on the right here, you have our loading system. So if you watch this is all this terrain is being generated as you're traveling along the ground. So it's not stored anywhere. And on the right you can see three levels of lot. The top one is a far distance and the bottom one is a near distance. So you'll see things like the little rocks, little shrubs. They're all in the low l. We're in a debug cam here flying along. Obviously, you wouldn't ever see this in game, but what we can

do is we can fly incredibly fast. And when we do that, you'll see because this is faster than you'll ever fly in game, but you'll actually see the train can't keep up. And that's because it's trying to calculate too much at once. So, if you slow down, you'll see that that's like the visible pop, and that's just the LOD trying to catch up. And this is all created using noise. So it's different algorithms that Sean and the code team have created to give this illusion of like a real place. What you're seeing here is like setting it to something really basic which is just a sign curve as just an example of actually what's going on behind the scenes. And this is fairly standard stuff in the world of procedural generation like Minecraft is a similar thing. But what we have is because we can have overhangs and caves and complex cave systems and it means we can have a lot more believable terrain things like this. This shows um like volcanic not volcanic sorry um glacial.

So you can sort of see the sort of trademark spiky mountains. But we can also do by breaking these rules that we've learned we can create more interesting terrain. And so we can have floating islands or mysterious crystals and it's quite fun to play about with. And the way we do like things like textures will build a huge library of these. This is um in Zbrush we'll have tiling materials and we can spit out height maps from that. And I was like really proud of this system. I was talking to Andrew from Naughty Dog the other day and he's like, "Oh yeah, we did that in ch Uncharted one." And I was like, "Okay, so maybe it wasn't that clever." But what it means is we can have cliff faces and we can have sand on the ground and when the angle's correct, you'll get sand pooling in the cracks and the mountains. And it adds this real believability to the world. And we have hundreds and hundreds of these materials and we can blend between them and we can tag them to say how dense they are and

how likely they are to erode, which means we can create things like plateaus and overhangs. And that's all based on how dense a material is. And depending on how dense it is, it'll kind of pick it'll have different specular values and have different uh it's a good way to explain. Uh essentially the more it'll have finer detail if it's more dense. That's I'll do. So, we had this system for creating all these different looking terrains with all these different materials, but that wasn't really enough because that didn't get across that lost world, Journey to Center of the Earth vibe, which they always had those kind of you'd go to an island and it would always have something interesting, interesting foliage, interesting jungles. And especially once the guys started to talk to me about procedural generation and I was the only artist working on the game at the time and we were obviously it was early days so they were experimenting and they started to use 64bit seed values which meant that they

could do 18 quintilian unique planets which was very scary to me at the time. Uh so there's four artists now so we it's not as scary but um obviously as an artist had to uh figure out a way of how you divide up 18 quintilian planets between four of us. How and that was quite a big task. So it turns out you have to be very methodical. You have to divide things up. You have to become a little bit of a robot. And the reason for that is without constraints you just end up with a real mess. So we kind of added our own constraints by dividing these things into different categories so that we could actually tackle small problems at once because if we tried to think about everything all at once we probably would give up. So one of the things at the very beginning we we thought about doing was that we were going to divide up the kind of foliage system into biomes which is the

way games like Minecraft and loads of different games do it. It's the equivalent of like playing Sonic where you have like a fire level, you have an ice level, you have a, you know, a jungle level. But so you'd fly down to a planet and this one would be like a sort of alpine sort of coniferous forest kind of planet. And so we're gathering lots of reference from the real world. Again, the logic being that if we could mimi mimic the real world and get the rules to be in that sort of a believable universe, then at the very end or towards the end, we could do really fun things, break those rules and get all the alien variation from that. So, we had like woodland, which be deciduous trees and that kind of thing, which is more like Britain, which is what I'm used to, although it's normally a lot grayer than that. And there's your jungle level. This weird little bird guy. And this would be like a Sahara desert type planet or more like your American Nevada deserts with your cacti and your harsh spiky terrain. Swamp

planets, deabar, they're always good. But what we found by having by creating those sort of biomes, those separate biomes that you'd human beings are really really good at spotting patterns. So you'd fly down to a planet and it would be a swamp planet and you'd go, "Oh, it's a swamp planet." And and instantly that breaks this illusion and you just realize what's going on behind the scenes. And the thing with video game artwork is so much of what we do is an illusion. Like you know, especially linear games where you can look out a window and you can see a mountain in the distance, but we all know that, you know, you can't get to that mountain. As artists work in the industry, everything we do is, you know, where it's all pretend and we don't want anybody to see behind the scenes. And having those separate biomes, it absolutely revealed that. And our game is so it's incredibly important that you don't

get that sense of illusion. So, what we ended up doing was creating what I call a super biome, which is a biome which covers all those biomes in one fell swoop. And it's what's a good way of explaining if you imagine a uh if you've ever used like a terrain generation tool where you'll have diff lots of different sliders that control different things like spikiness of mountains, trees, and we have that, but we have hundreds of different sliders. And we have min and max values for all those different sliders. But I don't have to go into every planet and manually move those sliders. What happens is when a planet is terraformed, those sliders are moved to different positions. Now, originally those positions were random. But what we found really interesting is if we start to link those sliders to things within the universe like distance from a sun, whether the planet has a moon or not, then the results start to become

believable and you start to spot patterns. But it's patterns that you accept because you start to realize like, oh, this planet's really close to the sun, so the foliage is either going to be non-existent or it's going to be it's just not going to be very green. And that's um where the believability comes into it. And another thing from looking at the real world is I realized that our planet is a very strange place like these in the Philippines. They're called the Chocolate Hills. Uh there's a really cool folklore about this which is uh apparently two giants had a really big fight and they were throwing boulders at each other uh but then they made up and they walked away but they forgot to take their boulders with them and that's where the chocolate hills came from. Anchorwat if you ever been to Cambodia and seen these it's it is like something HR geiger would make with these weird tentacle like roots destroying the old ruins.

I don't know how you pronounce that. Guiuelin Guian in China which are uh again a lot of these are natural formations and it it does make sense from a geography point of view but when you see them it's just it is like an alien landscape and I love these acid pools. I've used two images because it absolutely blows my mind that that's a real place. And the conclusion from these images is something we kind of realize quite early on that we don't actually need to go full full alien. We could actually just mimic the real world and we get absolutely incredible spectacle just from doing that. At the giant's causeway in Ireland, this has got a good folktale as well. Apparently there was an Irish giant and a Scottish giant and I think it was the Irish guy slagged off the Scottish guy. So what happened is the Irish guy started to build this causeway because

he was going to go and beat him up and the Scottish guy started to do the same thing and they've actually got two examples of this causeway either sides of the channel. I don't know who won but I imagine it was the Scottish guy. So, um, what we wanted to do was create a few key concepts that would capture the vibe of this lost world, this No Man's Sky. And like, if anyone's seen the the video we released years ago now, the E3 video where you walk down to the watering hole and the antelope runaway, this was the piece of key art that kind of influenced that. These are real quick sketches, but all trying to kind of get across the vibe that we wanted to have each time you fly down to a new planet. And what this did, it gave us targets. It meant we could tweak things like slider values to try to try to make

these kind of events more likely. And you'll see they all well a lot of them have this kind of these landmarks. And we always wanted to have that sense that you fly down there's always something on the horizon that draws you to it rather than just being a dull terrain. There's always something you see that you want to go to. And uh obviously that's fine having all these nice little pictures but actually getting that with a procedural system has been probably the most challenging thing. Like I should mention this project has been by far the hardest thing any of us have ever done. And if you're thinking about doing procedural game, I probably wouldn't. It's really difficult. I was trying to um I did a talk on Monday and I was uh surrounded by people much better than I am. um like uh Neil from Boware and I was asking him if you might consider you know taking this the the strength of Boware and maybe creating some procedural tools for the rest of us to use so we don't have

to create them all ourselves because it's just too much too much work for us. So procedural white boxing, I'm getting distracted, is the idea being that you have props that get placed in a terrain just like anytime you decorate an environment. But because it's procedural, you don't necessarily know what those objects are going to be because they get created procedurally. So what we do is we like we know that procedural props will be within a certain scale or will have a certain silhouette and that's how we kind of control it. So when we procedurally white box like this, it means like even though these aren't props, these are just rough rough versions of it. It means when we're balancing like min max values, we kind of get a vague idea what it might look like. So if I was loading this environment and replacing all these props with real procedural props, they'd kind of roughly fill those shapes. It means we can kind of ensure it's more likely to look

artistically pleasing. And talking about procedural props, we use our blueprint system to create these. So, it's not talking to I've had I've been talking to a few artists over this over the past week and they're all really interested in procedural artwork, but also very skeptical of it, understandably so, because a lot of the way it's presented on the internet and things is just this magic button that you click and then artwork pops out. But that's and I imagine some people do that, but that's not how we do it. we it's still incredibly artisticdriven and it's it's your procedural artwork's only as good as the base template that we create. I'm going to talk about that. So this is a tree trunk and this is the blueprint system. It adds a type of branch.

It then adds a type of leaf. It then selects a bark type and a leaf type and it is then decorated to match the environment and it is then skewed to be different. And then we can see it for a different version where we take the same trunk, a different set of branches selected, different leaves, different texture and different shape. And essentially every single prop in the game is created like this. And what you're seeing here is in the same way I talk about the train having all these different sliders. We have the same for trees. So all that's happened here is it's gone. I'm going to add some branches. How many branches will I add? Select it based on a slider. What kind of leaf am I going to have? Select that based on a slider. So these are just two examples. And that's just one type of trunk. We have hundreds of kinds of trunks and thousands of different leaves

and hundreds of different bark types. And the way it feels is that we're actually, if you've ever played like World of Warcraft, where you load up the game and you spend the first 10 hours fiddling around with the character creation, just scaling bones, moving decals. We have that, but for every single prop in the game. So, it's all based on this set of parameters that we custom make for each thing. And what's cool with that as well is that we can have to say here about the age of a prop, but that's I hate the word prop. It can be age, the age of anything. So we can have trees and you can say, you know, if you want a tree to be within a certain age, you know that the trunk will be thinner. It'll have less branches. And the same with a creature, you know that the head stays roughly the same size, the body will be smaller, the legs will be shorter. And it means you can very easily have like a a whole flock of sheep and you can have adolescence, you can have babies, you

can have grown adults. And it's I don't have to custom make those. All I have to do is say when you spawn a group of sheep, you should spawn, you know, 10 adults, 15 adolescence, and it knows because you tell it the min and max values of those that it'll just create them. And that adds to the believability of the world. This is a blueprint system on the trees. So what you see on the left here is just all the variants and on the right you essentially see what is currently picked. This is just an in-game tool we have for viewing variants. So I click on view variants and it creates a load more on the left and you can see the currently selected one on the right. This works for plants as well. Every plant in the game you see all the variation on the left. I can cycle through them. And the way that's possible is when I was talking about being turning yourself into a little robot, that's kind of what I had to do. So, we just have a system for creating as many kind of trunks as

we can. I mean, that's essentially what it is. It's just we do as many as we can and as just try to change the silhouette as much as possible to get as much variation as possible. So, we'll have trunks, branches, leaves, different materials, and various extras that we can add on like fungus, moss, and each one of those small extras adds to the variation. This is an example of some of the alphas we create for leaves. And these are really cool because you just create as many as you want and it's amazing the impact it has. And what you're seeing here is an example of the blueprint system. Like this is what I'm showing here is exactly the same planet, but all I've tweaked is the seed value of the trees. So you're stood in exactly the same place and I've just clicked view variance and it's moved. It's just changed the the type of tree. And here you can see it again. And what this is showing is that even if we

weren't to change a terrain at all, if we weren't to touch mineral colors or anything like that, you can still get a very different feeling planet just from changing tree shapes, which to us was really amazing because one of the number one the number one thing people seem to worry about is how you make every planet feel different. And the moment we started to fiddle around with those kind of values, it made us all feel a lot calmer. This is a video that's showing a uh a few of the worlds. Hopefully, it'll work. [Music] yeah. [Music]

Heat. Heat. [Music] Yeah. So, this is a message I got from one of our coders at Christmas time. It's always It's always nice.

Uh yeah, this is the Milky Way, which is our galaxy. It's a a spiral arm and we're somewhere in there. That's quite scary thought. So, what I'm going to talk about is our galactic map or our our in-game galaxy. And it's a really strange thing to talk about because this game more than any other game I've ever worked on, it's been quite an educational sort of process because like I was saying about geography and things like that, I never as an artist, I didn't really pay a lot of attention to school. I was very interested in what I was interested in. So making when having to do galactic artwork has meant I've had to Google things like spiral arms and it's really really scary and I don't know if any of you have read uh Hitchhiker's Guide to the Galaxy but it mentions something which is called the total perspective vortex

and the story goes something like this. So there was a scientist and he had a loving wife, but his wife used to nag him a great deal. And she would always just pull him up on these really small, insignificant, at least in his eyes, little things. So he thought like if there was only some way I could show her how unimportant these things she's moaning about are. And being a scientist, he thought, I know what I'll do. I'll invent a machine that she can put on and what that machine can do is in a split second it will show you the entire universe and it will at that same split second show you in comparison to the entire universe. And I imagine it's something a little bit like an Oculus Rift. You'd put it on and he turned it on and it destroyed her

brain. So that was our main kind of inspiration for our galactic map as well as cool things like the pillars of creation which I think is an awesome title. This is the Eagle Nebula which is also really cool. Space is filled of really cool things. And this is a little video. So I should mention like one of the things I've tried to stay away from in No Man's Sky is the color black. And that was one of those bold things I said at the very beginning of the project. Chris Foss, he uses colors, you know, none of this black. But when you're trying to do a starfield and you're staying away from black, it becomes this real weird challenge because it stops feeling like you're flying through space. It's really easy for it to feel like you're flying underwater. So finding images like this of the Pillars of Creation was really useful to me because it's it's still obviously space, but it's colorful. So, this was a massive uh a really useful image when I found it.

Oh, wrong way. Um, so what you're going to see here is every single one of these little dots is a solar system you can fly down to. And that's until you play the game, it's not obvious. Um, you just have to keep that in mind. Heat. Heat. [Music]

[Music] problem with the end. Uh, spaceships. So, you can't really make a science fiction game without thinking about spaceships at some stage. So, I thought I'd talk about it straight after that. I asked Aaron, who is uh one of the artists I work with. He is an incredible artist. He he's one of those guys that lives and breathes hard surface modeling. And so, I asked him if he could he's he's the main guy. He is responsible for all our spaceships basically in the game. I can't take credit for those. I did like the one in the prototype video, but he has taken those and he has done amazing things. So, I asked him to put together an image that showed what he was thinking when he's been designing spaceships and he thought he would use comic sands in his image. So, uh I was I was going to take it out, but screw him. It's his problem. So what we're seeing here is we have three

types of ships play that you can uh get hold of in the game. So we have trading ships, fighter ships, and explorer ships. And essentially you can pick any of those ships, whatever you want, but certain ships are more useful. If you want to be a trader, then you're going to get a ship that has a bigger cargo hold and is more armored. And fighters generally are smaller and nippier. and explorers are weird and scientific. So, this image just kind of shows some of the inspiration that Aaron took from various places. So, let's get away from that comic sands. One of the things he was saying is it's absolutely impossible not to tread on some toes when you're designing spaceships. Like, we get so much stick online of just being like, "Oh, they've just copied an X-wing or they just copied a Viper." And it's like, you should try to create a spaceship and not tread on some toes. It's absolutely impossible. But the way we've tried to do it is to kind of look at a lot of the World War II planes because they have that sort of chunky style which is

something at Hello Games we've always been real fans of. These are some of the early concepts of the fighters. And it's an interesting thing with procedural systems because we could if we wanted make spaceships abs, you know, completely procedural and unpredictable. But when we were doing that, we found it made like as an artist sometimes people talk about things like gameplay and gameplay is something that's been created to make make art worse. So we have to listen to these guys and then we make our art worse and it makes the game better and people are happy. So we have fighters and we have drop ships and we have explorer ships. And the problem we had is when we when they're entirely procedural, you couldn't really tell the difference between them. So although they are still procedural, we just have a set of rules. So we with fighters, even though they're all different ships, you can kind of tell like a similar vein. They'll have like two wings and they'll have an engine and they'll have a cockpit and they'll have a nose. And all that is is like a set of

constraints. That means your the end result is recognizable and readable for the player. You can see some more examples here. And this is like the in-game blueprint view. It's the same system as uh as all the trees and things. You can see all the variation on the left, the main one on the right. I click on view variants. We can also have procedural decals. These are the drop ships. And you'll see they have a different set of rules. So they don't have the same engines. They don't have noses, but it means you get like very readable shapes. Like even on the left, even though they're all very different ships, you know they're drop ships. You know they're traders. And what it means is when you're flying through the galaxy and you see a ship flying along, you can very quickly understand what kind of ship it is. And it's that was hugely important from a gameplay point of view.

The same with freighters. We don't need to have as much variation within them, but they're all procedural as well. And because we have factions within the game, they can all be linked together really easily. We can have certain shapes that are more likely to be used in certain factions. We can have certain decals, symbols. Same with space stations. Space stations can be procedural and they can be linked to the factions, which is really cool fun to play around with. Uh, also we have creatures in the game. Just going to drink some water. I don't think we actually have camels. So that's a false advertising there. Uh so the same at the beginning we started to as I was in full robot mode. So I thought I'll try to categorize creatures to what kind of environments they might inhabit. So snow creatures. But as I was going through this process, the same way, the same revelation, which everyone else would probably just immediately know, but was news to me,

that some really crazy creatures on our planet. And like these creatures could easily be in a sci-fi movie and you would just accept it. like if you was going to reveal the protagonists in the game, this is who you actually play as in No Man's Sky. So, another thing I discovered, even though I'd been a character artist for many years, I'd made a load of creatures and rigs and whatnot, uh, in looking at skeletons, because essentially what we when we were breaking things up and dividing things up, we looked at the real world, looked at those photos from earlier and said it would we would we want to have creatures that are a bit like a deer or a bit like a rhino. And in doing that, like immediately, obviously, when we're thinking like that, we're limiting ourselves. But luckily, I'm really glad I did. I have this really big uh anatomy encyclopedia

uh at work and I was flicking through it and I was literally about to start building a rig when I think I just randomly flicked between dog and cat and uh I discovered that dogs are basically cats. See, isn't that interesting? Dogs are also horses. Horses are pigs, just a bit more bent over. Pigs are sheep. Lions are red pandas. And red pandas are triceratops. So that might seem like nonsense, but it I'm sure I had a point. So the point is that if you would have a system where you could move bones around and lengthen bones and tweak bones without having to redo a rig, then you could make all those creatures on a single rig, which for you know Boware, they wouldn't bother doing that. They just throw a thousand artists at them and create all

the creatures. But uh we don't have that luxury. So we had to come up with a system for being able to create all these creatures and have them share rigs. And that involved, I can't take credit for this either, we have a very clever guy called Charlie who has created this beautiful IK system which means we can pull joints around. We can completely push a rig to its absolute limits and the IK system compensates. So, you know, I can do terrible things to a rig and make it look absolutely hideous, but the creature will still somehow find a way to traverse the terrain. So, that meant we knew we could have these shared rigs. So when we're designing creatures, suddenly again I talk about how we added constraints to ourselves, but we kept doing this where we'd add a constraint and then suddenly the coders would find a way of avoiding that constraint and we're like ah I was quite looking forward to just creating a small set of things but suddenly it all blew up and we could be like oh it turns out that a wartthog can have the same rig as a duck build platypus and or a frog and which

is great because it means we get all this variation but it's also just like as an artist I'm a complete troll freak and having that kind of power. Uh yeah, I just wanted to get back in my little box and kind of curl up. So these were um yeah, I'm just getting ahead of myself. These are like some of the concepts which we kind of created to as like in our blueprint system when we have built like a base template you as I was saying that that base template needs to be really really good otherwise it's like it is the foundation on which all the generation is built. So if you have a bad template then all the creatures you create will be bad. So we have to be really careful at the very beginning to plan exactly the kind of variation we want to get from a rig. This is the uh the whale/ shark/water eel alien crocodile rig. And this is a little blue dude. This is our blueprint system for creatures. So you have the base template and it gets skewed.

It then selects type of ear tail. It selects the texture. This is again like the tree just one one possible version out of thousands or however many hundreds of thousands we want. Then get skewed. This is the bone scaling where you can pick any silhouettes. That's real fun. And this was like once we got this system working, it was absolutely amazing because suddenly we were populating these worlds with even when we only had a single rig and we they were being spawned on every planet. Suddenly you were flying down to a planet and I wouldn't know what I was going to see and it was unpredictable and I'd never had that before because as an artist you normally know exactly what you're going to see. And these are like some of the Zbrush because we have like a system where you can um create normal maps that we can blend between a bit like how some uh like AAA games they'll have like wrinkles and things that they can blend on and off. We have a similar

thing for like scales or muscle groups. I'm going to show you a little video of the blueprint system on the creatures. So what you see here is again on the left you've got the variation. We've got the parameters set to be within a small amount. So, you'll see when I click view variance, it'll be similar looking creatures. So, you know, the scale is roughly the same size. It's not a million miles away from what's before, but we can change the values, which I'll do in a second. And so, yeah, that's a slightly fatter version, but if I change the scale, put it right down, then you'll see a very different looking creature, hopefully. There we go. So that's exactly the same rig, but hopefully you'll also notice it it chooses a different animation depending on what type of creature it gets. So you've got a cow, it knows that because it's a heavier creature, it's going to use a slower walk. And if I can tweak the scaling there, so you can see I've made him with really skinny little back legs and really heavy four legs. But hopefully what it's demonstrating, and you can get robots. You got to have

robots in a sci-fi game. What it's demonstrating is that you're getting a an awful lot for very little. It's a shark dude. A little alligator. And the next guy I really like. This is like our unicorn guy. Unicorn whale. And his attack animation. He has like a little happy happy face. So at this stage we have all those creatures. You're clicking view variants. You're getting very different looking creatures with the trees. You're clicking in view variants. You're getting lots of different looking trees and plants and the terrain. We're talking about those sliders and each one has a min and max. But so imagine like you've got a planet, you've clicked on random basically. It's just picked, you know, that antelope creature. It's picked the uh unicorn whale and it's picked pink trees and it's picked like blue grass.

It's all going to well I know for a fact it looks absolutely disgusting when you have that. And that's what I'm referring to when I say procedural soup because if you don't try to control that, you'll end up with a real ugly mess. And I think that is what as artists we get terrified with procedural stuff because generally you don't have any control. So you're just click you're just randomly hoping that the result you get is a pleasing one. So the way we kind of get around this is uh again I wish I could do all myself but I work with some really talented guys and girls and uh it wasn't a pleasant experience but there's a lot of back and forth. Um we work together teamwork it's always rosy and there's never any arguments and we all agree and we all do what's best for the game. But essentially what happens is we'd implement a system we'd load up the game. I would be offended with my eyes would bleed and I would then paint over and try to stop whatever was

happening. And then what would happen is the co the whoever I was working normally Sean like I've known Sean for like I've worked with him for seven years now and we have a very healthy relationship. So we don't mess around who I just say Sean let's look let's change this and what he's really good at is looking at what I've done and then he makes me a better artist. He says, "Why have you done this in a paintover? Why have you done that?" And there some artists are really good at knowing what they're doing. Whereas I've just always been a little bit more gut driven. I just do things because I think they look nice. But being a robot man like Sean is, he thinks, why why are you doing this? And I think, I don't know, because it looks nice. And what he was really good at doing was saying, well, oh, so you I noticed you've put like bushes around the base of that tree or you've put little rocks and things like that. And I'm like, "Yeah, of course. That's what you do when you're doing environments in games." And so what all he does is, "Well, why don't we just tweak the environment system so that every time you place a tree, then it does that?" And so it's that's like a

really simple example of just like one of the little rules we have. It's it's all these tiny little rules and systems that basically make it more likely for the procedure result to be like something I would have painted. Whe whether that's a good thing or not is for you to decide. We also have rules when it comes to things like shapes and we talk about believability like say like a a sort of swamp environment that's been when the planet's been terraformed it'll have it'll have generated this climate and ecosystem and again based on all these nice little universe rules like how how far away from the sun or just the minerals in the soil and that's all just like nice little controllable factors but what happens is we know that certain we can tag certain types of leaf and tree and branch as being matching to an a climate. So essentially what that means is on a swamp planet you will get swamp leaves and swamp branches. That's sounds like it's

limiting but it works over a huge number of things and it means that that can also get filtered through to creatures. So in a sort of damp environment, you're going to get these more coldblooded sort of waxy damp looking creatures and it and also some of the color from the environment will leak through to the creatures so they'll be natural camouflage. And in the same way we can have the opposite of that where we can add in props that are complimentary so they'll stand out. And that's that's when the real fun begins. And what I love talking to Sean about is how objects get placed on a terrain. And I know for a fact this is something that anyone who's ever worked with procedural generation with a terrain will know exactly what I'm talking about that when you place objects using noise, humans just know there's something not quite right. And so what we have is like a clumping system and it means that things get placed um in a more pleasing way which means

that like they understand shape and they will add like a large prop and they will place medium props and small props and it will kind of gradiate down and every environment artist does these kind of things but setting the rules it meant we were guaranteed for that kind of thing to happen. Another thing I had to do was uh this isn't I teach color theory to a robot. I just love that idea. Essentially, it's a bit like having a junior artist when you're working in procedural technology, but it's a junior artist who is like the worst. He's complete idiot and like you cannot you can't teach him anything and he just doesn't listen. So you have to be absolutely rigid and just it's a bit I remember I was talking to another guy and he said that um he found that when he did tutorials as an artist that it made him realize why he made the decisions he did. And in doing this I've had a similar experience that there's all sorts of decisions I make that I

just did automatically. But in having to implement these rules, it meant that that I was understanding why I was doing what I was doing. So we have like materials have different pallets. So you'll have like I can which I can define and so there's no random colors. You can have hue variation, but I can still say if I want like to say dirt is always one color. I can absolutely do that. But obviously in doing that dirt would then be the same color in the entire universe and that would be really boring. But it means that I can control things like I can say what color fur is and I can say I want when you have a creature that has scales I want the markings to be a complimentary color. I want the underside of certain creatures to always be lighter. And it means you don't get this horrible garish mess. And I can do the same with trees. And we have a tagging system. So everything is linked together. So the color of the grass, the color of minerals and soil

will all filter through to the to like the trees and the props. And it just means it all feels coherent. And this is our view of our little color droids that we send down to every planet because when you've got this many planets, I can't personally fly down to every single one and decide whether I like the color of the grass or not. So we have uh these little dudes that fly down. They spit out animated gifts for me. I can then and it puts it online. So I can then go and look at it at my leisure and I can click on the image and get a seed value and then fly down. And what it means is I can balance colors like a god from above and tweak them and then send out my little minions and all the all this stuff. All it means it's a way for me to control the output and it means I can link things together and the results are something like I would create myself. And hopefully you can see here like we have a tagging system for props. So you

can see that the trees in style and color and shape they kind of match the props the uh the shapes you see on the ground. And even the creatures you'll see their markings and the the colors. They're all linked to the environment. So everything just feels like it belongs. And you can see the sort of clumping system with the complimentary colors where you've got the blue against the orange and you've got the green sky and objects placed around cave mouths. Like I haven't placed any of these props, but hopefully it doesn't look too procedural. And we get some really cool bugs. Um I'm trying to think the best example I can give is like I was fiddling around with this. Um I wanted to create these little hippos that spawned in caves just as like a little experiment. When I say hippos, it just means like a mammal type creature that's within a certain scale. And I know that and you I could set it to have like a bulbous sort of nose. And so I'd done this on my PC and I thought, cool, that's quite fun. I'll just check it in and just, you know, we'll decide on it later. Uh, and then one of the guys, he'd obviously got latest and he'

loaded up the game and he'd gone into an underwater cave and obviously I forgot that at that stage in time caves were to say whether they were above water or underwater. So he'd swam around the corner and been to his surprise found like 200 hippos just splashing about in the water. And that's the butterfly effect. I can I can tweak tiny little things and it has this crazy crazy effect which is really scary. And I'm going to show a final video which hopefully shows all those things off. [Music] Heat. [Music] Heat. [Applause]

[Music] Heat up here. Oh, [Music] [Applause] [Music] crap. [Music]

Yes. [Applause] So, I've been told to remind people to fill out the little uh judgment cards, but you can forget if you're going to give me a bad mark. So uh but if anyone would like questions then I think there's a little microphone. I don't don't know how much time we have but yes in uh creating all these procedural systems uh you talked about creating the art cohesiveness rules. Yeah. How did you incorporate things uh like gameplay concepts like uh uh threat uh threat level or difficulty or strength or things like that? How did you incorporate those kind of concepts into that same system? That's good. So we have things like um so the question was how do we incorporate gameplay elements like threat into a procedural system and through our tagging system we can like

when we create like generated props things like horns and spikes that on things like creatures then we know like scale values. So if a horn for example gets above a certain size, we can say a creature becomes more dangerous. And what it means is when you see a creature, you can kind of predict how dangerous it's going to be just by the way it looks. So like the more ferocious the larger and the more spiky a creature is, the more likely to be dangerous it is. And but we can also play with that and have like a little fluffy white bunny and have him be really dangerous. But you just don't know. Hi. Um, I had a couple questions about your production pipeline, like for your art assets, because it's kind of really unique. Um, so when you guys were doing R&D for your proceduralism and all the different things that you guys were trying out, so you guys have multiple rigs, you have multiple geometries. Yes, we can. And so do you guys how much do you guys rely on like blend shapes and

morph targets? Like you guys blend in different normal maps. Can you talk about that a little bit? Uh, yep. So we have the thing with um procedural systems is you almost want to have as as much flexibility as possible. So we can we do have entirely generated uh systems where the artist has no control and things like UVs are automatically generated and um textures are automatically applied. But what we do is if a if the result isn't good enough that's when we'll start adding in handmade assets. And like you were saying, we have blend shapes. And because we have bone scaling and bone positioning, it means we don't actually have to have too much different geometry because it just kind of works by pulling it around. And it's and we'll with normal maps, we create as many different ones as we want for like a creature. And often that's just things like muscular or bulbous. And we can blend between those or scaly. And we found you don't actually need to create that many to get the variation. But the more we do, the the better it becomes. Is that cool?

Thanks. Hi. Oh, sorry. What? Oh, time's up. Sorry. Um, if you want to ask me more questions, I'll be in room. This is important. 2002. We can chat later if you want. Cool. Thank you very much, everyone. Yeah. Yeah, I will. Okay.