Star Citizen: Reverse the Verse LIVE - It's Bender and Tracy Time

Sean Tracy · Technical Director of Content

It's a good question, actually, because it really depends if we're talking core technologies for Star Citizen or core technologies for Squadron 42. Now, we try to keep that as co-planar as absolutely possible, but they are two very different types of games, right? One is a single-player, a little more cinematic scripted kind of experience, whereas the other one is a huge MMO. I don't think there's any massive ones. I think maybe, in case the roadmap doesn't have this, the solar system transition. So going from Stanton system to another system, so the wormhole kind of idea, that's got to be addressed. So I don't know if that's on the roadmap, and if it isn't, it probably should get there at some point, but we need to get Stanton done first, and then we'll fly out of Stanton. But in terms of core tech, the thing that that wormhole stuff relies on is the object container streaming, which is very, very busy right now and definitely represented on the roadmap. I'm trying to think of other things. I mean, there's opportunities like core tech changes or things we'd want to do, but they're not necessarily blocking Star Citizen from getting made to any effect. So the major pieces of tech that we needed for this game are almost there. Really only the object container streaming is one of the biggest ones. And that changes the paradigm with how we work with objects and levels. So that's a big one.

Jared Huckaby · Host, Community

It's also a good time to remind folks that, as has been stated in Calling All Devs, in other episodes of RTV, in ATV, everywhere we can, not everything that's being worked on is on the roadmap.

Sean Tracy · Technical Director of Content

Yeah, none of my stuff is.

Jared Huckaby · Host, Community

The stuff that you see on the roadmap are the things that we intend to release in those quarters. So there are things that will be worked on continuously throughout 2018 that may not be coming until 2019. Those are not on the roadmap. And then there are things that we do think will come in 2018, but they're smaller features. They don't fit easily within one of the epics and stuff like that.

Sean Tracy · Technical Director of Content

Maybe to dig into that a bit, we've got tons of R&D projects that we work on that, again, if we put that on the roadmap, people feel pretty disappointed a lot of the times. There's a ton of times, just general game development, that happens all the time, right? You explore an idea. You have to try something before you know it doesn't work. And sometimes you say, no, we're just not going to do that.

Steve Bender · Animation Director

Because it's about making fun. If there was a manual for fun, everybody would be able to do this. It really is, in some cases, trial and error.

Jared Huckaby · Host, Community

One of the things I like to say is that this is game development, not game construction. We are figuring out the answers to these things. We are figuring out the things we want to do and then how to do them as we go along. That's what game development is. If you're doing something that's truly never been done before, if you're doing something that's truly innovative in the industry, there is no instruction manual for it.

Sean Tracy · Technical Director of Content

Yeah, and a lot of stuff in isolation has absolutely been done. I can't think of a really easy example right now, but just because one feature existed somewhere, that's never been coupled with this feature, right? And sometimes it's a combination of the two things that make it really difficult.

Steve Bender · Animation Director

Yeah, so I was meeting with both the German studio and our two UK studios, the Derby studio and the Wilmslow studio. We were going over combat AI within Squadron and obviously within Star Citizen. Making sure that we had the different user stories or plans in place and reordering those, talking about what it is that we're looking for on those cases. Also, we shot two weeks' worth of motion capture for combat AI things. And yeah, so it's been sitting down with the team working on the various FPS aspects and social aspects and things like that throughout the world.

Steve Bender · Animation Director

Yeah, so you can say, well, I want animations for people running around and shooting or something like that. But what a user story is designed is that it's about what the player or what the particular user wants. In the case of Sean's guys, when they sometimes will do tools, the end user would be, say, the animation team or something like that. But that way, it's from a player perspective. So we're saying, as a player, I want to be able to feel like I am hunting AI or something like this, right? Hunting enemies. And then what we do is we start to break that down to see how it is that we're going to actually make it feel like that.

Steve Bender · Animation Director

Cool melee animations. Well, you guys have seen some takedown things before, I think, right? So some stealth sort of takedowns and things like that. And we're looking into things such as, obviously, if you can stab someone with a knife and you can hit the melee button, you can hit them with your weapon. What if you don't have a weapon in your hand? What can you do there? So that is, again, one of those user stories where we want to be able to, without a weapon, be able to fight somebody else.

Sean Tracy · Technical Director of Content

Yeah, and we rely heavily on DirectInput for this. It's an entire library that is in Windows. And generally, we're probably not going to replace that. It's a lot of work, ultimately, just to give consistent IDs or whatever. And quite honestly, Windows has got to fix this. This is something that's going to come from the drivers necessarily. So we're not about to change the entire input system for that. Is there something that we could put in place so that if it does detect the same kind of things, it's not going to remap or change the IDs? Yeah, probably. It's just not really the highest priority thing right now, to be fair.

Steve Bender · Animation Director

Yes. At the moment, that is not on the short-term goals, but absolutely, especially in the case of Star Citizen. We talk about how we want to make a living, breathing universe, and a lot of our players will say that they're really interested in role-playing and living within that universe. So connections between players is going to be very important for us. But right now what we're concentrating on are the core pieces of the game to build this out properly, one-to-one. And then from there we will extend out.

Sean Tracy · Technical Director of Content

Yes, actually. So we've got a couple of different hairstyles that are coming down the pipe. We haven't talked exhaustively about it for this year, but the DNA gene splicing stuff, which is going to be what part two of the character creator ATVs is. So we've come up with some solutions within there that are going to allow us to do this stuff, which is super nice. Especially to do with beards, which are usually one of the harder things to do when you're customizing and changing a face behind the scenes. But one of the trickier parts to it is we've got to deal with gravity. So gravity right now is just local to the player himself. So all the character attachments, they are affected by gravity. It's just not actually the gravity in the world. It is just kind of a fake value that sits on the character only. So we've got to receive that from the planet. We've already got the setup problem kind of fixed up for the simulation. So that's one side of it. But yeah, that's the first thing we've got to do. And then on the wind front, the wind got unified to a new system. So the CryEngine had wind before, but again, it was a Z-up kind of assumption, right, with a level plane, and wind was just an XY sort of idea, and there were little boxes that would fly through the world. So now that you're around a planet, we've got to convolve that entire system around a planet, and Marco Cabello has done that within our engine. So we've got to get the hair listening to those values, basically. There's a couple other things with that, is that short hair still has to be affected somehow by wind, because otherwise it just looks super static. So it's whether or not we actually want to get world wind for that or whether you just kind of fake it a little bit, where you've got wavy stuff going on. There's no real reason to spend a ton of performance on just simulating somebody that's got hair as short as mine or so. But the longer hair, yeah, we've got to do something with, definitely. The one thing to keep in mind, though, is that when you throw these helmets on, we still have the sort of helmet hair, which is a little more like the back of my hair here. We put the Snoopy cap on and we actually remove that hair so that it's not simulating inside the helmet. It would look a little silly. But yeah, now whether we go Fu Manchu kind of beards, that's going to be totally up to Josh and whether we want to support those. But yeah, they're going to simulate just like, as crazy as it is, this silly monocle chain and things like this. That's all going to simulate and be affected by what's going on in the world.

Steve Bender · Animation Director

Yeah, so that's a really good question. One of the things that is really important overall in Star Citizen is player choice. And you're able to customize your ships. You're able to customize your loadout, your armor, what you're wearing and everything like this, what your face looks like, what hair you have. But typically in a game, you're not able to customize the specific animations that you have. Animation-wise, one of the things we wanted to do is to make sure that you could customize that, so that as we put more walk cycles into the game and we create this appearance that not everyone that walks around walks in the same way, we could then give you the ability to access some of those so that you can start to customize your character. Say there's a more funky walk and that's what your character's like, you could choose that, and various other things as well. So different emotes and dances and things like this.

Sean Tracy · Technical Director of Content

It's even as simple as the idles, right? Like, you want a different idle for fighter jock stance.

Steve Bender · Animation Director

Absolutely. So it's the whole thing, right? The whole package where you could begin to start customizing every aspect of it. We're already doing that with the clothes and with the character faces and the colors of everything and things like this. But if I want to stand there and my guy is kind of, that's who I'm playing, this kind of broken-down character, I should eventually be able to customize those aspects.

Sean Tracy · Technical Director of Content

Yeah, that's exactly something we need. And one of the things that the physics guys are looking at now is the SDS and the physics for the ships. And that ship specifically is the main reason that we have to look at this. So, yeah, that's going to absolutely have to be supported. So, can we do it right now? Nope. But we're going to have to be able to with the Hull.

Jared Huckaby · Host, Community

I think that being able to separate and change physics like that is a big part of our non-death states for ships, too. Because right now, the reason you shoot a ship and then it explodes and everybody dies is because we can't actually separate the...

Sean Tracy · Technical Director of Content

Well, we just can't animate it. We just can't move it around. That's the trick. So with some of the transforming ships, we totally worked around this fact, right? Like the Reliant specifically, there was a bunch of things that they had to do. He's talking about this all the time, so he's really after the physics guys for this. But like with a lot of things, it's not going to get a huge amount of priority until that ship is coming closer and closer. And I think it kind of is.

Sean Tracy · Technical Director of Content

Object container streaming, by far, because that's the thing that is changing. Our paradigm has already shifted away from levels and prefabs and these kind of things to the object container approach, but it's not complete until we've got full asynchronous spawning across all those object containers. We're able to profile them. And then even better is once we're doing that, we can rework or update the tools. So that's the biggest one for me anyways, because it's stopping us from scaling huge, and it's also making it not confusing but unclear how to write certain pipelines or tools for the design guys to work within it, because we don't fully have it. Now, they do work with object containers, yes. Then it'll be a matter of streaming them. But once it's a matter of streaming them, it's a matter of how big are they? Can we profile them? Can we export them in the build process? Can we share them day to day? There's a lot of questions that need to be answered within there. So for me, it's object container streaming and asynchronous spawning.

Steve Bender · Animation Director

That's a really good question. So I don't know if this actually answers the question, but I'm looking forward to especially the core AI combat things getting worked out, because even if you're not necessarily interested in going to new places, meeting new and interesting people and shooting them in the face, there's still a lot of the aspects of the gameplay that go into this, which once we have that set and ready to go, we can extend out, and it really becomes a far deeper experience for us animation-wise.

Steve Bender · Animation Director

Wow, that's a really good question, because we've done a lot of different mocap shoots and a lot of different content for the game. Probably the one that got the biggest sort of result on the set, where people really got tied into it, were the takedowns. And there was some really brutal stuff going on there. There's a really good one that we're holding on to at the moment. I specifically want to be working on that particular asset. Although, at the end of every mocap shoot, so not every day, but say you do a week or two weeks or something like this, we do just a little five minutes at the end of the day. We try and cut off five minutes early, give everybody a bit of a breather there. And we just ask the actor typically to just give me something. And we had one of our mocap actors who is a special operations director, and he was talking about the Orange Fanta commercial or something like that. 'Wanna Fanta, don't you wanna?' And there's this guy in this orange suit who just runs up to somebody and goes like that and runs away. And this guy did this on set, and I lost my mind. I'm glad we did it at the end of the day with five minutes left, because we never could have gotten control of that set ever again.

Steve Bender · Animation Director

Yeah, so I've seen people asking about this sometimes. We actually shot three different speeds based on the role-playing intent or gameplay intent that you're going for. One was sort of, I'm just walking and I'm climbing in. I called it Grandpa Goes to the Store. Then there was combat, which is, say, scramble fighters and you're getting in. And emergency, and that would be, say, the entire Vanduul horde is chasing you and you need to get into the ship right now. The same thing with exits. The first one that we've implemented is the combat set. And just like anything else when it comes to animation, when we get those things in and we go through playtesting, we may need to speed up those assets and stuff like this. But actually, we built each of the ships on the mocap set and had the actor try and get into it as quickly as he could. So the answer to that really is yes, and there will be faster ones coming. There's a user story, actually, for getting in at an emergency sprint sort of speed and also getting out.

Sean Tracy · Technical Director of Content

I actually don't know. If we're talking volumetric VDB gas cloud stuff, I think we just showed a whole bunch of it in ATV yesterday. So, I mean, they pretty much see where it's at. If they're talking about planetary clouds, that's a whole different conversation. Now, you might think that you could just use the same thing, but the problem with clouds, real ones anyways, like on a planet, is they change. They're always moving, they're shifting, they're flowing, these kind of things. The cloud and the volumetric nebula gas stuff is all as good as it looks, but it is static. It is an export of one frame of a voxel simulation from Houdini. So that's not something that we could do unless we had clouds on planets that just don't move, which is just not where we want to go with this. And the Frankfurt engine team, the guys in the engine room, will be working on this. I just don't know when. I think it's going to become very important with our next couple landing zones, though. That's the thing. I just don't know where it sits on the actual roadmap itself. We've got kind of, I don't want to say half-assed ways, but yeah, kind of half-assed ways of showing just the clouds. Intermediate approaches, right, exactly. Kind of particle techniques or just some real simple 2D stuff. But I know Carsten Wenzel, who's one of the best rendering engineers I know, and he's tech director on our project, did a lot of the rendering work for the atmospheric scattering, and he wants to continue that onto the clouds. So I know he's got some great ideas for this. It's just a matter of time. And since he's heading up the performance team, and I hope everybody's been experiencing the huge, huge wins that we've been making on the performance front, once that's in a good state, that's probably where I guess he'll move on to. But yeah, I can't say for sure. I mean, he's my senior, so I don't really know.

Steve Bender · Animation Director

Well, you could certainly load an animation on one layer and then create a second layer and load a different animated result onto that. But what you end up getting, actually, is a bit of a mess. I don't think you would get something that looks like a believable alien. And it would just look completely broken. And the problem there is that you're then unable to control the outcome of the hundreds of animations that come off of that. So could you have a happy accident and go, wow, I hadn't thought about that, and then start to plan out how you would create that later on? Maybe, but I would assume it would end up being more of a technical mess. If you're talking about something, let's say you have an alien where the bottom half is mechanical and the top half is organic, and there's an actual disconnect between those two, you could certainly do something like this. But what you're doing is you're combining the top half of something with the bottom half of something else. You're not actually meshing the two of them together and just sort of going, I got red, green, blue, purple, plop it down, and then you end up getting brown.

Sean Tracy · Technical Director of Content

He kind of answered his own question while he was asking it. So to give the full idea of the workflow, we call it zone culling. And it's not really culling, and it's not really zones, but that's what we call it. What it is, is we have some tools in Maya for the characters specifically. So when we make a character asset, we've got some tools that will allow us to project the vertex colors that we need for all the given zones, and I think there's something like 32, 38, it changes. So we've got a lot of the different zones that need to cull out depending on the attachment that is attached. So whether you're putting a jacket on top of a shirt, on top of a body, we've got to cull all the lower layers. And the reason for this is a little bit performance, and actually much more clipping. Because when you have dissimilar topology, for example, shoulder is a good example, or elbow is an even better example: if you've got an edge here, an edge here and an edge here for a shirt, and then you put a jacket on and it's only got an edge here and an edge here, you're still going to get deformation coming through. And there is no way you can fix this. The only way you can really fix it without actually culling some pieces out would be to keep the exact same topology between all the assets. And that would have required that a long time ago, four or five or six years ago, whenever they started making some of these character assets, everybody just followed the same topology, and that's super restrictive for artists. So it's really hard to work with and it's really hard to maintain that all the way through. So what we came up with instead was an ability to mark these areas up with vertex colors in Maya. Now, that doesn't go all the way to the engine. That just tells the tool within Maya, separate this mesh chunk, this mesh chunk, this mesh chunk. And then within our DataForge records, we tell a whole bunch of different areas to hide depending on the attachment or the removal of an attachment. And there's also partial layers and things like this. But the tech artists deal with this generally. But it's a lot faster than it sounds. We've got a really nice tool that Gage Hallman has been working on. Actually, he's working on it today, even, to make it faster for the character artists to do it. But what we really need, and Josh and you character artists: stop changing the zones. Stop changing the zones, because then we've got to update all the assets every time we do it. But yeah, the zones should be pretty good at this point, I think. So that's how we do it. And I know a lot of other games do it the same way. I just don't know how they necessarily put their meshes together. For example, I've seen bugs in GTA where I'm like, oh, hey, they're doing the same thing. That's cool. I just don't know how they actually marked up their assets or whatever.

Steve Bender · Animation Director

Yeah, well, that's a good question. So from the code aspect, we actually do have the ability, when we're playing an animation, to kick out of it. An example of this would be if you, say, threw a grenade at somebody, and they were about to throw a grenade at you, and they explode, they go to ragdoll. Whether design wants you to be able to have that ability or not, it really becomes a design question. But the way that the animations are authored, they start from the bottom of the ladder and they go up into the seat area. You could allow the players the ability to interrupt that by, say, pressing the space bar and they jump off, but that would be something that design would need to request from code to be able to happen.

Sean Tracy · Technical Director of Content

I've heard ideas, just ideas that have come from the community. I think we've talked about it before. It might be cool. Todd might not agree, Chris might not agree, Steve might not agree, but I think this would be cool. When you're getting in the ship, usually if I'm in a panic and I want this thing to go a lot faster: press, press, press, press. The faster you press, maybe it steps through different speeds of the animation or whatever. It's just basically button mashing as you're getting into the ship. If you want that thing to go faster, go faster, right?

Steve Bender · Animation Director

Well, like I said earlier, we did shoot, and we do have on the user stories, a couple of different speeds implemented. So for instance, the sprint speed, so that if you hold down the sprint button and are executing that, it gets you into the ship faster and stuff like that. So it really does come down to exploring those gameplay aspects.

Sean Tracy · Technical Director of Content

Yeah, because you had an idea once, and I think it even depended on the rate at which you were approaching your ship, right? Because you liked it where, if you were in an all-out run towards your ship, you would do sort of a slide across the hood of the car type thing.

Steve Bender · Animation Director

Yeah, so we've been talking about that. When you go up to your ship, you have to hit the use key and it opens up everything and stuff like this. Well, what if I'm on a run, I'm on a sprint, I got the Vanduul horde chasing me, 'Chuck, get the plane going, Chuck!' And as I get near it, it begins to open and I hit that in full stride in the sprint. But that's not something that design was currently after. They wanted to make sure that that was specific, that you were opening that and things like this. So I'm not sure which aspect they're talking about, whether it's just partially getting up the ladder and then hitting the jump button so they can leave the animation, or exactly what that is. I'm not entirely certain.

Sean Tracy · Technical Director of Content

I like how they said that rather than 'steal', right? Because it's definitely not stealing. Yeah, absolutely. And if you're not doing that in any industry that you're in... I very realistically compare ourselves against, it's not even competition, but similar things in the same space. The thing I hate doing personally, and I think a lot of people hate doing this, is re-solving a solved problem. This is something that Dennis Crowe says all the time, and I like the way he says it, which is: if a problem has been solved in a very good way, why go through the trouble of re-solving that? So sometimes I say, well, that's kind of a solved problem. Why don't we do it this way, right? Okay, we want to implement whatever it is, a chat system in the game. Okay, cool. There's like 30 of those, probably a lot more than 30. You've got Skype, you've got, not MSN anymore, I'm kind of showing my age. MSN, AIM, America Online. But basically the point being, they've got Slack, Facebook, whatever. So do it the way that they've done it. It doesn't need to come up with a brand new solution. Don't reinvent the wheel every single time. Sometimes it's a solved problem. So yeah, definitely I let myself be inspired. A great one right now, and I still can't understand why we're not doing this, so hopefully you guys post on the forums about this, because I'm going to make a big deal about this right now. You know how cool it is to get into the ships? It's really, really hard to find the damn doors on the ships. If I've not gotten a briefing on a given ship, I have, as one of the tech directors of Star Citizen, been in there for 10, 15 minutes trying to find that door. And to go back to the question: playing Zelda, if anybody's played Zelda: Breath of the Wild, great game, a master class in open world design, super cool. But one of the things that they do really well in these Divine Beasts, and I'm not trying to give any spoilers or anything, is in the Divine Beasts they have this 3D map view that is phenomenal. In Zelda, it's like a 3D puzzle within these Divine Beasts. And in that, you have to know where you are and you have to know where certain switches are within this 3D space. You can rotate the thing around, you even control it, but it's got this great holographic, kind of transparent view with really, really bright orange indicators for different things. And it's not dissimilar to what our targeting stuff already does, but that would be a super nice way to give that map to the players so that I know where my door is. So, point being, I absolutely am looking at other games. And if other people in the industry, or other people in their own industries, don't do that, then they're selling themselves short.

Steve Bender · Animation Director

Competitive analysis. I do that a whole lot, actually, because when we're talking about animation for something or we're talking about a feature and things like this, obviously we have our own ideas and our own visions for this. But as Sean said, it's about seeing if anybody else has solved this problem before and how they did it. And it allows us to say as well, okay, this game did this like this, but we don't want to do it that way. Or, here's an interesting way that they solved this, so maybe we go in that direction, but we give it our own spin or something like that. And you work in this industry long enough, you have people that come up and go, oh, you worked on that thing? Oh, yeah, we saw that, and we were like, how did they do that? We want that, right? So there is a lot of give and take that goes on industry-wide, and people being inspired by the work of others. So absolutely.

Steve Bender · Animation Director

Yeah. We've been looking at, for instance, signals and things like that. Emotes are one of those things within an MMO that's a living system, and you keep adding and you keep adding and you keep adding. It's a content sort of thing. The idea with emotes, we have them for dancing, we have them for saluting and waving and laughing and various things like this. And the more life that we want to get into the universe, the more animations we're going to be making for those and giving to the players.