Star Citizen: Reverse the Verse LIVE - Props! Props! Props!

Well, it's a pretty big deal for us because it basically means that it's very easy for us now to get things like material effects, lighting and so on dynamic on our props. So say, for instance, we can create a lamp entity, and when you shoot that lamp entity, it will spawn a particle effect, the light will go out, it will roll across the floor. And it basically means we can get all of that done inside DataForge, which a lot of people have probably heard of. It's an in-house tool we have for creating these entities. It basically just allows us to make those quite quickly and quite easily.

I would imagine it makes syncing with animations easier.

Yeah. So I think, to expand on what Cory was saying, it's our in-house tool chain to bring all the different elements of the game together. The ultimate goal, at least from my point of view, or the props team's point of view, is that when we make an asset, we know that it can be dropped into the world by anyone that wants to use that asset, and it will have the functionality built into it. So we're not having to constantly repeat the same setup, or we're not having to drop in one explosive barrel in one area of the game, and someone else uses it in their area, and they're like, oh yeah, it does this much damage, and then in another area it does this much damage, and that sort of stuff. It's a way of us containing all of that information just in the actual single asset. So it is a lot of working with other teams. It's a lot of working with the guys that write the tool and the guys that support the engine and just bringing all that information together. And I can't remember whether or not we talked about it last time or not.

I think we were just on the cusp of Item 2.0 becoming a thing.

But we're a lot more confident with the actual tool chain now. So we have the ability to set this stuff up. And then one of the things that is really nice that it allows us to do is we set up one asset and then we can child all the other similar assets that have the same sort of gameplay elements underneath it. It's the same system that the guys use for doing a lot of the other interactive terminals and airlocks and all that sort of stuff in engine. But it means that we can iterate the visuals and still have consistent gameplay.

You can inherit all of the features of a certain entity and then you can make changes to that. You can make children for that because it's an inheritance system, and then you can make changes to those for the art, but you can still retain all that functionality as well. So if you make changes to the parent, those changes will then propagate down as well. So it allows us to iterate on, as Ben said, stuff like values can be changed.

That's an important thing. I know it may seem like a dry subject, but building these processes that allow things to be inherited, things like that and the procedural generation stuff, those are the only ways we're going to get to making everything that we have to make for a game of this size. You have to build smarter, not harder, as they say. So a lot of what Star Citizen has been doing the last couple of years has been building these tools that will allow us to build the content as we go forward. And it's great finally being in content phase.

Yeah, and I think that's absolutely it. Because we are a content team. Everyone enjoys making pretty assets and making the things that they really want to show off. And then you get to the bit, and if that setup is really difficult and time-consuming and just repetitive, then it's no fun. Whereas if you have a system where you know that you can get it working once, it might be a case that we'll make an asset, we get a template set up and we send it over to design. They can balance it from a gameplay point of view. The lighting guys can light it and get their default lighting set up and all that sort of stuff. And then we can come back and we can just take that. And like you say, we're making a universe. It's massive. If you want to make that amount of content, you need smart tools and clever ways of working.

It's also really exciting just being able to shoot a prop and seeing particle effects and hearing sounds and it all working now. Seeing that coming together is really nice. I think in the latest ATV, that was something we showed off.

I think metrics is a really hot topic, at least from the internal side of things, for the fact that they touch so many different elements of the game. Like you say, we've been in development a while. We've got a whole plethora of assets, and we also have a lot of performance capture and animation stuff that we need to retain as much of that on-set acting as we can and get that into game. And obviously because our props can be used across lots of different environments, lots of different areas of the game, it's really important that everything works together. We're not going back and going, oh, okay, this metric isn't quite right anymore and this, that and the other. But what we are doing is, because we're in a position where we have this new functionality and we can feel confident making metric templates... Before, we just had screenshots on a wiki page that laid all this stuff out, and it wasn't particularly user-friendly. It wasn't always the clearest as to what the images were actually trying to tell you. So we're making an effort now to actually take that information and create proper in-engine templates that have all the functionality that we want to have, all the performance capture plugged in, and having that all tied up into one thing that is in-engine, it's in 3ds Max, our content creation tool, and it's also documented. So it means that if another department that might not normally make beds needs to come in and make beds, we can go, oh, here you go. Just drop this into your scene. And all they need to worry about is iterating on the art. And they know that all the technical side of things, all the helpers, all the naming conventions, all the dimensions is all just going to work. And it's one of those things that you do once, you're done, and it works. Again, it goes down to scalability, and we want to be able to get through the content quicker and not have to worry about the more technical side of things. Now we are in a position to do that.

A lot of what we've done so far has been gameplay stuff, and that will always play an element. So we'll get the design team coming to us and they'll be, okay, we want to have this new game mode and part of it is you've got to go and find this thing and destroy it, and it's the first person to destroy it, or that sort of gameplay element. Or it might be that we want these particular cover assets to be able to take cover behind them, but then as you're taking cover and taking fire, they start to come apart, and that sort of stuff. We are making a high-fidelity game, I guess is the right way of saying it. So generally our assets fall into a few different categories. We have our set piece props, that would be tables, chairs, lockers, the larger stuff that you wouldn't necessarily expect to be physicalized. If you bump into it, it's bigger than you, so you're not going to really affect it that much. So that stuff tends to be static. We then have our destructible gameplay ones, which are things that hurt if you shoot them. Environmental hazards. So like gas canisters and the like. And then we have more of our dressing assets, which are generally the stuff that you would be able to pick up. It's still fairly early days with being able to say, oh yeah, every single thing that you can pick up in one hand, you can shoot and be destroyed, because we don't have the data on how much that's going to impact our performance and all that side of things. But that's the goal, really. The goal is that if it looks like you can shoot it, you should be able to shoot it. That's where we want to be.

Anything that's not bolted down really should be physicalized, for the most part anyway. Apart from the Big Benny's machine, because that seems to be physicalized anyway. That's been physicalized since day one, right?

It is. The Big Benny's machine was accidentally physicalized in the early builds. It was not intended. There were several Big Benny's machines in that level, but only one was physicalized, and that began the Big Benny's run. It's a happy mistake.

Being in zero G obviously has an impact. But yeah, things that are bolted down stay bolted down. But generally, we want to have a good variety of things that you can shoot.

And just things like a melon. I feel like it's a bit of a crime in a game if you see something like a melon and you really want to shoot it, but you shoot it and nothing happens. You get a wood particle effect or something. It gives you that sort of texture in the universe when you know these things are actually made of the substances. Well, they act like they're made of the substances they appear to be made out of, right?

Yes. You've got the VFX guys, and they care about what happens when you shoot things. You've got the audio team, and they really want to support these systems with beautiful sounds that are different depending on what it is you're shooting. The ethos is that you want stuff to act how you expect it to. You want it to be that immersive experience.

So, sub-items. These are the parts of the ship items that would deteriorate over time. You're going to have your different ship items, and you might be able to swap different grades of ship items that have different performance abilities. And then these sub-items again allow you to further tweak their... This is very design-led. I don't know much about it specifically. We make them pretty, they make them work. But these are the things that you are going to be able to swap in and out. Depending on the size and the quality of the ship item, the sub-items, you might have two, you might have three, you might have more. And because, again, we do things from single-seater fighters all the way up to big capital ships, these are a consistent thing that will allow you to tweak your ship's performance, be it your shields or your power systems.

They behave as stat modifiers, basically, I think, but in a sort of cool dynamic way.

And then that gameplay of being in a ship with your friends and something blowing, and one of you has to run down and, mid-combat, rip it out, find the others, pop them in. I think that's certainly going to be fun. That's what I'm excited for.

I think it's going to be a big part of ship combat, actually dealing with failures in your ship items and sub-items and the like.

One of the things that I've really been enjoying about playing Sea of Thieves lately is when you get into ship-to-ship combat, somebody has to go down below and start patching holes in the ship and start bailing water out. This will be our much more complex and driven version of that.

Yeah, and the combinations of sub-items, ship items and ships is huge. With the ship items alone, you can swap these things in and out, and how you combine that stuff is going to give you a lot of different functionality, I guess.

You're not going to be able to fry an egg and watch the egg white go from translucent to white. The reason we're doing that is because of the software we're using to create the food. Because that software is procedural, it allows us to get these values and expose them and change them, so we can very easily iterate on and alter these assets we're creating, to get rapid iteration on them and get them to look right. So it's a byproduct of the pipeline we're using to create that stuff that gives it the procedural stuff that you saw. In the game itself, we're basically going to be baking all that information down into a texture for the most part. Baking, that's a good pun. I didn't actually mean to do that though.

That pun was definitely intended in the ATV. I know Ricky said it wasn't. It was.

Yeah, so we use Substance Designer. Go ahead.

So basically, it's not a system in-game. It's basically an art tool that we're using. A byproduct of that is being able to have that extra control over it.

I think people look at it and sometimes question the fidelity we're going to. What Substance Designer gives us, and the fact that it is node-based and more procedural, is that rather than having to create high polys of stuff, or photosource stuff that probably won't stand up to quality, or do the whole photogrammetry thing, which has its use cases...

Yeah, we do use it in some pipelines.

But for us making this, Substance Designer is a really quick way, especially in David's hands, the guy who made it. He can iterate on this stuff really, really quickly and get the quantity of assets we want really quickly. And also it means that if we do need to make changes, we're not having to go back to stage one again like with all the traditional methods. We're just like, oh, can we tweak that yolk a bit, or can we adjust things, or if the writers want the curry to be more greasy, or it needs to be more lamb, because lamb curry is a big, big deal in Star Citizen. So having that ability to make changes and get the look we're going for without having to spend loads and loads of time creating the source assets is really advantageous to us. And the other thing is, again, we're making a universe and we don't have real-life reference of alien cuisine. So having tools that aren't tied to real-life source material is, again, really important. And that's why we take that approach with Substance Designer.

It's basically non-destructive. So once you've created something, that's not all you're left with. You also have all the stuff you used to create it, and you can go back and tweak those things and just get a different result. You can basically hit a randomize button and get 20 different curries if you want. Well, obviously within certain parameters that a curry could change. I'm not promising 20 curries in the game.

We saw some comments, some people think that creating procedural tools is somehow additional work, and it's actually the exact opposite. You build the tool to reduce the amount of work that you have to do in creating all these.

That's absolutely it. The Substance package has really been gaining momentum over the last couple of years. It's almost industry standard across a lot of development studios. It is about creating things smart, and it's not always about taking the fastest approach. If you can do stuff really fast and it might not be to quality... If you can do stuff fast and it improves the quality, then it's a no-brainer.

Yeah, it's the best tool for us, and it's not a waste of time to do it that way. It basically just means we get more reuse out of the time we invest in it.

Like I said at the start, we're a content team that serves the other teams. So we work with the designers and with the different artists in the studio, and we'll get request lists from them. So they give us, oh yeah, this is a particular gameplay feature we want. We want something that does this. And then it comes to us. We then work with the concept team. We've got really strong concept guys here. And we've gone through a few different phases of how we approach doing this. What works really well for us is that when environments are being concepted, the concept guys also consider what functionality is in there, what gameplay is going to be in there, and what props they're going to need. So if we're doing a hangar, and we'll talk about that probably later, what makes sense in this environment, and what will have gameplay? And as we expand to different areas of the game, we just continually add to that list. When I first took over the props team, or created the props team, probably three years ago, I guess it was, I already inherited a few thousand props, and trying to get that stuff organized was quite a challenge. We have a pretty strict naming convention and folder structure and that sort of stuff that does help. But you're right, the sheer number of assets we make is pretty huge.

Yeah, as well, specific sets of assets can go a very long way. So that's another thing we do: if there are assets that are very specific that aren't needed first, they'll likely go to the bottom of our backlog. We're not going to make things that are going to be used sparsely and are not necessarily needed for gameplay right now.

It is quite a difficult balance sometimes, because the stuff that you need to make a level work is one thing, but then there's the stuff that you need to actually make it feel exciting. The surface outposts were a really fun thing to do because a lot of that was storytelling, and that extra layer of immersion and history is a really nice thing to be able to do. We also work quite closely with the writer teams, because they will look at a particular environment and they've got the knowledge of the lore to be able to say, oh yeah, this particular back alley is where this character lives and his history is this. It'd be really cool if we could have these additional props that sell his story, or this guy's really into taking things apart and he needs this real cool workshop. Even to the point where, if you look at Squadron and at some of the AI characters' bunks, it might only be one or two lines, but those one or two lines, like I say, they sell that story. So they might have a particular medal from when they won a race around a moon or whatever it is. So we get things fired at us from all different areas, and we just prioritize them and get through it.

I suppose for me, it's that you have to anticipate what sort of features you will expect to come online in the future. For instance, before we had all the destruction stuff or the entity stuff, we still had to think ahead for that functionality. And it's also just figuring out what sort of situations that prop's going to be used in, because we make these things and it's hard to say where they're going to be used or what they're going to be used for sometimes. So you have to think about that on an individual basis.

Yeah, I think we do have an element of realism in our game. So if we're making a cargo loader and we get a beautiful concept from the concept team, we're like, yes, ace, how's it going to work? It might not work straight away, but we have to anticipate someone saying, that's cool, I want to be able to drive that. Okay, cool, right. And as long as we're making those decisions at the start and we go, okay, we might not necessarily be making it working off the bat, but we're going to model it and approach it in a way that it could work, we know that eventually when we get to that point, we've done the hard work, rather than having to go back and go, oh, those proportions don't work, we can't extend and actually pick things up, and that sort of stuff. Although we are making a sci-fi fantasy game, it's futuristic, it is set in the future, we've still got to consider how technology will work at that time and what that functionality will be. So I think a little bit of time spent processing that now is really important. I guess the other challenge is that you can't just take a concept and make it. It is very much: does this have any performance capture attached to it? Is it going to have any performance capture attached to it? Or is somebody on the design team going to see it and go, oh, I like that, I could use that for XXX?

And you have to also be responsive to the sort of things people want to do with them. So you may have to take something back and then alter it to work with what they've been inspired to use it for, which can sometimes be quite interesting.

Yeah, it's a nice thing when people are like, yeah, that's really cool, we want to use it for that. That sentence alone is like, yeah, you know what? We're doing a good job here. This is cool. And it's the same for the community as well, when we get positive feedback, they're like, oh yeah, we love this and we love that. It's nice to see our assets used not just to make things look nice, but actually to affect the gameplay and bring more exciting things to the community.

Personally or as a team? Because as a team, we have a comms tower that currently holds the record for the tallest prop. Although we have a crane that was made recently that I think has topped it as the tallest prop. Smallest prop, we have things like pills.

We're not making individual grains of rice here, just putting that out there.

But yeah, we do have very small props like a pill. It's not modelled, it's a texture, but we have props that are incredibly small.

Yeah, like pens, chopsticks, noodles.

That's the thing about our team. Again, something that makes it quite challenging is the huge variety in scale and in what we're making. It can be quite crazy jumping from, say, making a shipping container to, okay, now I have to make a set of planters, or I have to make a glove, or I have to make some cloths or bedding or whatever. You're going across lots of different skill sets and working on various different things.

James did actually start blocking out that cargo mover, which is probably going to take the crown of the biggest one soon. It's not ready yet.

There are a few things that we put in that are fun to us. So we might pay homage to things. I don't know sometimes whether or not they're too subtle, maybe. But making video games, that's one of the luxuries you get, I guess: you get to put your spin on things sometimes. Obviously you've got to be careful; everything needs to work in fiction. You don't want to break that immersion. But there are definitely things where we think, oh yeah, we're making a particular set of assets and it'd be cool if it was this, because that pays homage back to a particular movie or a particular game or any of that sort of stuff. I don't want to get into specifics.

You mentioned you don't want things that break the immersion in the game, and I couldn't help but feel like you were talking about a locker. Were you talking about the Locker from Another Universe?

Oh, no, I wasn't.

Back when Behaviour was working for us, they were making a lot of the hangar flair, and they actually created a sub-flair called the Locker from Another Universe. It's three lockers, and you go through and there are things like Mark Skelton's gold jacket. And one of the things in there is a bottle prop called Disco Lando Fever Syrup. And it was in there before I got hired in.

Yes and no. Because of the quality of the game we're making, a lot of these props are interactive. You can pick them up, you can look at them, or you will be able to eventually pick them up and interact with them. We need to make sure that they stand up to quality. One of the things that we are really conscious of is our draw calls and our material usage. We obviously are very conscious about our vertex count and our poly count budget, but we're more conscious of our materials and texture usage. Our layer blend materials are slightly different to the way the ship teams and the environment team work. It shares a bit more with the way the character and the weapons teams work. But it gives us the ability to get really high-fidelity surfacing on assets but still maintain nice cheap draw calls, which is the really important thing for us.

One thing that's changed the way I think about stuff while working here is that the size of a prop, to me, is the size of a prop in screen space, right? So where it is on the screen. You may have this coffee mug, which when I hold it here is the same size relative to the screen as a shipping container 200 meters away. So you have to consider that you're making something to fit that amount of screen space. You're not making something that size.

And it's about spending the budget wisely, I think, is the best way to put it. If we're looking at assets and they've got lots of stuff they don't need because you're never going to see it, then that's not a good asset. But if an asset is slightly more expensive, but it really improves the silhouette and makes sure that it stands up, like Cory says, when you see it up nice and close, that's where we're like, yeah, do you know what? Okay, the poly count's a bit more, but by increasing the poly count, we're not needing as much texture space. We're not needing as many unique materials. That makes sense. So it's about balancing where the asset is going to be used. If we're making stuff that we know for a fact is never going to be interacted with, you're never going to get close to it, then we'll make it as cheap as we possibly can. But most of our stuff, or the smaller-scale assets, is interactive. So it's about just being smart with it, really.

And we're also very keen on things like doing LODs, good LODs that halve the poly count each time. So even if you have this coffee cup and it may be, say, 300 triangles or so when it's in the mid-range distance and maybe a bit more when it's up close, you're not necessarily drawing all those polys all the time, because it may just be on a desk.

And for anybody that hasn't watched every single broadcast we've ever done, LODs are our levels of detail. Close up, it's at full resolution, full polys, full tris. And as you go farther away, it drops back, usually through five stages, until you get to what's just barely more than a pixel.

There's a lot. For me, I don't get to do much of the creation side of things anymore, but I really like the dressing stuff, the smaller little things that make people smile, I think. It's the stuff that you see and you think, yeah, you know what, that looks really cool. You can see they've put in the extra effort to make things that are interesting. So the dressing scale of stuff is what I enjoy.

Yeah, I had a think. I definitely think some of the most interesting stuff I've made is probably the ship items, just because we got such lovely concepts for those and it was so fun figuring them out and so much fun making them. And we really got to push the detail with them as well, because obviously they're going to be an important part of gameplay in the future, so we want to really make them sing. I think that's probably my favorite.

Yeah, actually, the toilet cubicle. I really liked making it. It doesn't open currently, but when it does... I spent a lot of time on the interior, and it was almost like making a very, very small level. I just had a lot of fun with it. And also the door mechanism. In the outpost, if you go to the toilet, just take time to appreciate the door mechanism, because I put a lot of thought into that.

It's very restricted by how much space you've got. And again, functionality: if it's going to open, it needs to open right. No magic doors into walls for us.

How much time did you spend on the toilet cubicle, Cory?

Well, actually the outside was made by someone else. So I spent maybe a week on the interior, maybe a little bit less. I can't remember, it was so long ago. But it was a lot of fun. I don't think anyone's seen the interior yet. That's why I'm really excited for that time. Another thing, I just remembered, I really quite enjoyed making the bedding. For our bedding, I don't know if it's been mentioned on any of your other episodes, we use something called Marvelous Designer for our cloth, which is basically a cloth simulation program. They use it in the fashion industry to concept dresses and the like. But we use it to create a lot of our cloth assets. And when I was making the bedding, I actually had to get a little character in there and set him down and then tuck him in using the cloth simulation program. So that was quite fun.

I think that's another example of working smart, using the tools that are available, rather than having to go in and hand-sculpt every single fold in the material you want, and then someone coming along going, those folds look a bit ropey, and then having to spend loads of time iterating on it. Having a system where you can simulate that stuff gives you your high poly, your source asset. And then if we go, actually, do you know what, that character's not quite working right, or this doesn't quite work...

Yeah, it's very easy to iterate on different versions. I could very easily and quickly make a neat and tidy tucked-in blanket, and then I could make one that's hastily thrown on or hanging off the bed or whatever, and it's very quick to iterate on that. But it's just quite a lot of fun playing with simulation programs in general.