TL;DRFavorites differ: one likes small to mid-sized ships for their gameplay flexibility, one likes big ships like the Caterpillar, and one cares mostly about how much of a ship can be blown off.
For me, kind of the smaller to mid-sized craft, because it's enough to build out where they can have a bunch of different gameplay potential to it. We can explore some different options with it in sort of the infinite timeline of things, where we can build it to be this thing now first but then potentially go to other options later, but without being such a behemoth ship where it's like, okay, I'll make this ship this year, for just the amount of work and detailing that goes into anything.
For me, it seems I get most satisfaction out of the bigger ships like the Caterpillar. When you drop down those doors and the vista is just out there and you're standing there in gravity looking at this amazing view, and then we get to blow them up too. So I have a lot of fun with the bigger ships. I'm looking forward to working on the capital ships, and we're working really hard on, I won't go into detail on it, but the gravity, inside gravity and putting little ships in bigger ships and all this stuff. And we're going to be delivering some cool stuff for 3.0.
The zone system is one of the most impressive achievements for me personally with Star Citizen. The ability to put a ship inside another ship and be able to transition seamlessly, I imagine that makes for quite a number of challenges.
Size and type doesn't matter as much as how many pieces I get to blow off of it. Honestly, that's pretty much it. When I look at a ship, the first thing I think of is this is going to be so fun to tear apart. I'm looking for all sorts of little flaps and winglets and panels. Does this animate? Oh, can I rip that off too? Where are the areas that are going to show cool damage as a player? What are you going to see? What's the damage going to look like when it's landed and you want to repair it? What's it going to look like in space? Which is the coolest ship that just looks awesome when it's blowing up and parts are exploding and there's fire going off. I know that's really vague and totally not specific.
We have to balance the amount of debris that can fall off a ship with the optimization. Each separate mesh costs a certain amount of draw calls, so we have to kind of keep it in check.
Yeah, it's the number of passes on a certain mesh. And say you have 10 draw calls on a hull and you start to break it in half, now you have 20 draw calls. You break it off again, you have 30 draw calls. It's not exactly that way, but you can kind of think of it like that.
TL;DRThe damage model should scale to capital ships very well since the setup is similar across ships, though texel density may be lower and more resources might be invested later.
Very well. The new UV2 shader damage model will work perfectly for the larger ships as it does for the smaller ships. I think the texel density will be a little less, so when we get to them, we're going to maybe invest more resources into the bigger ships, but we're not there yet, so we'll find out. But the actual setup is very similar across all ships.
Well, we handle the outside of the ship mainly, and the modelers are really doing the insides. We do some internal work for lighting and whatnot and these damage states and things. But for destruction, we're mainly concerned with the exterior at this point.
Yeah, I mainly just worry about what details are underneath the exterior surface of the ship, like geometry, girders, wires, sort of POMs, so you can't see through areas you're not supposed to. But otherwise, the interiors themselves are usually modeled and lit by the artists themselves.
TL;DRThe new Aurora is being updated for the new damage system with room for equipment, and all ships are to get a lighting upgrade, optimizing first and adding effects after 3.0.
Well, Elwin's still got it in his... but he'll be handing that off, and then it'll actually be ready for the new damage system. So everything you've seen with all the new melting, wearing, underlying stuff. He's working out some kinks in it to make sure that there's room for actual equipment that we're putting in, the physical space for it. So physically it's being modified to just sort of fit our new systems, and it'll be a total upgrade. So same ship, but it'll look much better all around.
In the near future, all ships are going to receive a lighting upgrade. Chad Zamzo has developed a thing called light groups. And if you imagine the Caterpillar, it's got a thousand lights inside of it. What these light groups do is they kind of record the place and the settings of each light and reduce it all down to one entity. And we have far more control over animation of lighting and light states. So we're going to be doing a pass. First off, for 3.0, we're going to do a pass where it's just optimizing and reducing the number of entities, which will increase performance. But post 3.0, we're going to be adding some really cool lighting effects like startup sequences and flickering of lights on the wings of ships. So I'm very excited about the lighting possibilities. Nathan Dearsley is working on a bit of that. He's been doing some prototype work.
It looks amazing. Some of that you can see on this week's Citizens of the Stars that aired on Monday, some of Nathan's current work on prototyping the internal light and stuff like that.
TL;DRDamage states probably won't differ in atmosphere beyond gravity and turbulence changing the physics, though atmosphere-specific effects could be done and a more physically accurate debris system is planned.
Have we really thought of that? Simple answer is no. What happens when you break off a wing, say, is we fire these squibs and it distorts the UV2s and creates an alpha channel, so you get kind of an uneven surface, and then we fire some special effects. That's the current state of the damage tech. That's not going to change in atmosphere, other than there will be gravity and turbulence. So the physics will change.
But different weapons do have different weapon effects when they hit. A bullet will tear through the metal a bit more than, say, a laser, which will burn through the paint but not actually tear through the metal. So maybe there's some kind of damage inflicted via atmosphere that has a different visual effect. I'm just saying that could be something that's done.
We're going to be working on a debris 2.0, and what that means is we're probably going to take more into account the torque of a percussive thing, like ammo hits at a certain vector and that'll start to shear off the wing, and it'll be more physically accurate, basically. Right now we give an impulse to the debris and it doesn't really matter where we shot it from. It just kind of always randomly sort of pops off in this cone that we established through our XML.
TL;DRThe Greycat currently has only old-style damage and would need UV2s and redesign for new damage states; no pass is promised, and it can't yet be spawned in the Persistent Universe.
That doesn't mean I can't burn and melt through it, but I don't think there would be much sort of interesting geo to actually fly off it. But I mean, UV2s and then set that up, change the damage shader and get that activated. Blah, blah, blah.
We would need to steal an artist or do it ourselves, where we put the internals into a structure. So once you kind of see into the body, you can see something.
That's part of the thing we come across: a lot of the older ships, when we're doing damage conversions, they didn't have really much space between the exterior and the interior, so you couldn't really fit cool mechanical structures between the body. You couldn't even fit sometimes a plane that shows the sort of damage POM. So as these ships are being redesigned for the new damage systems, all that stuff is being taken into account, so maybe the Greycat.
TL;DRParallax occlusion mapping is already used on damaged ships, combined with real geometry where there is space, because it looks like geometry cheaply while full modeled ribbing would be too expensive.
They already do. Yep, already done, did it. Basically, there's two parts of the underlying geo. So you've got the exterior surface of the ship, and then you've got either damage POMs, which are the parallax occlusion maps, and those are a single plane that look like 3D geo due to how the textures are. And then you've got the actual geometry where you have space. So if a wing is actually thick enough to have, say, the skeletal ribbing as you go through, and then you have a POM under that, that way when you shoot the ship, you can still see through it. But yeah, we combine both techniques with that. So we've got POMs all over the ship to do your rivets and your insets and all sorts of cool stuff that we don't have to charge. And then we use it for the damage too, because why not? It looks awesome. It actually simplifies the process, because you don't have to model a full set of geo; it looks like geo at a glance. It's all extruded-looking and such. So it's great tech.
I wish we could actually model the full ribbing of a ship, build it like you'd see it in a real construction of a spaceship, but it's too expensive to draw all that when you're not seeing it.
TL;DRShip destructibility depends mainly on how interiors will eventually be split (multiple physics grids per ship are planned), plus choosing detachable pieces like intakes and wings that add gameplay without overloading performance.
Well, the biggest consideration is how the interiors will eventually be split. We can't just split at any point or create holes in the hull. We have to think about eventually when we split the ship, we're not going to disappear the interior, which is kind of a cheat right now to get around the physics issues that we currently have. What we want to do is, like a big capital ship, you break it in half: if this half has the gravity generator in power, it should still have gravity, where this half will lose gravity. But right now it's all one big physics grid for a ship. So Chris Rain is going to be working on multiple physics grids per ship.
Yeah, and then you obviously think about performance-wise, but you also think, that's probably going to blow off. You look at a piece and it structurally doesn't look like it's going to take a huge pounding. At some point, that piece is going to tear off. So we try and pick and choose pieces that'll add to gameplay too. Like, that's the first thing, intakes, right? When I'm breaking things off for, say, Matt Sherman. So we'll break off intakes, break off wings, we'll see how that affects how it plays, and then I'll be like, well, what's left? Is there anything cool? There's only a few pieces breaking off; anything aesthetically appealing, easily detachable, that's not going to overload everyone's computers. So that's kind of my process.
One of the features I'm asking for is, especially on the big capital ships where you're going to have multiple layers of rooms and they may not line up perfectly on a bulkhead, we want to have the ability to swap the interior for, say, a split room version. So imagine the Titanic breaking apart and you see a bit of a room with some mess coming out and people falling out and screams and smoke and fire and all this stuff. So that's kind of where I'd like to head. I'd like to see that in the atmosphere and watch people falling out and let the gravity take them.
TL;DRThe team is most excited about Item 2.0 because it unlocks many blocked features and lets ships be built the way they always wanted; one developer also likes lights, and turret spotlights are planned.
Probably just the MacGuffin umbrella of Item 2.0, because it's like, oh well, we can't do that until we get this 2.0 stuff online, we can't do that until we get 2.0. So just all the knock-on effects that it's going to bring in terms of really just fleshing out gameplay and opening up a lot of the possibilities that are just like, all right, these are our items on the checklist, but once we have that, then we can really start checking a lot of this stuff off.
To this point, we've been building ships using modified default CryEngine, or I should say Lumberyard stuff now. But now with Item 2.0, it's really how we've always wanted to build the ships, where the ship itself can communicate and it's networked and it's got proper optimization. So I'm most excited about finally getting there, and it's really happening right now.
I think they talked up all the features. Personally, I like lights. I think they look cool, and I can't wait until we can connect them all and have all the different states for all sorts of cool stuff. Because I think it's really the visuals, what's going on, how the world feeds back to you as the player, that really pulls me in anyway.
I've talked to you a lot of times about lights and about having external lights. Like searchlights, so you can go through an asteroid field and have a light when you're doing search and rescue, looking for a ship.
TL;DRItem 2.0 should let unpiloted ships stabilize themselves instead of spinning, bring better turret gameplay, and improve ship optimization and appearance because cockpit joysticks will stop affecting the ship's LOD switching.
Well, just one of the basic ones, because I also know from a lot of times just flipping through different community streamers while they're playing the game, they're like, oh, I wonder when this bug will get fixed on this ship, of getting out of your ship and having it just spin for no reason. And that's one of those sort of simple ones that we actually will be able to start really addressing and getting locked back down once we have Item 2.0. Because, like you were sort of saying, the current setup now: if you aren't piloting your ship, and not just, oh, I'm sitting in the co-pilot seat; it's like no, are you in the pilot seat of your ship? Okay, cool. Everything's powered on. Everything has some level of awareness. You get hit, the ship will counter-correct and stop its own movement. You get out of the ship, it has none of that communication. It has none of that awareness of self. So once we have Item 2.0, it's like, all right, cool. Now you're allowed to auto-stabilize again, because your ship's IFCS, even when you're not in the pilot seat, is going to say, hey, why am I spinning? I shouldn't be spinning. Thrusters, stop my spinning. And then you may come back and you've tilted from taking a hit or taking an impact, but it shouldn't just be spinning on its own and then pinball you out into nowhere if you approach the ship wrong.
The thing that excites me the most that's happening right now is the upgrade to the turrets. We've had a lot of problems with the gameplay of the turrets, and I'm excited. I just saw Mark Abent's new implementation. It's amazing. So I'm mostly excited about that. I've been helping him prototype a couple things. But beyond that, optimization is going to be... and the ships are going to actually look better when we do Item 2.0. And it may not be clear, but we optimize the ships by adding LODs to the ships. In a normal engine, LODs switch at a constant rate of distance. But in our engine, we have a switching rate based on the smallest average poly size of the LOD. So if you have a ship hull and it's got a little tiny poly in it, that LOD is going to switch to the next lowest LOD fast. So we spend a lot of time looking for these small polys. And then due to the old tech that we've been using, say to hook up the joysticks to a player and run IK, traditionally the joysticks have been part of the exterior of the ship. You haven't known that, but that's actually been a thing. So when we move the joysticks to cockpit 2.0, that means the little LODs that are on these joysticks won't be influencing the LOD switching of the entire ship, which means we get a better looking ship closer up and as it switches. So it's going to be great. I'm very excited.
And for anyone who might not know what LODs are, they're levels of detail. It's the difference between seeing every single poly of the ship when it's right up close and then watching smaller and progressively less detailed versions of it.
And you're trying to find a balance between range and visual loss. We're trying to make it so you don't see any pop; you're not supposed to see the model change. That's the perfect world scenario of an LOD. But after a while, when you're really just supposed to hit a certain poly count versus certain pixel space, you're like...
Yeah, exactly. That's usually an LOD. Could be streaming, could be a model just appearing, whatever kind of issue, but definitely we try and make those as clean as possible.
Now the artists are doing all their own LOD work, and we're just kind of helping implement. We used to do a lot of LODs, but it's not our problem anymore. We were LODing everything in the game for a while, and now they want to do it more by hand. We were automating the process using a software package that we had bought, and it gave good results, but you can see some weird shapes sometimes, and we wanted, at least in the first few LODs, to have a cleaner result. So they're hand done.
TL;DRBreaches are currently limited to airlocks and turrets because hull holes would also require matching interior meshes; swapping in a holed interior wall might become possible, but it would need careful planning and resources.
If they need a breach where gas will escape... right now, we're limiting the breaches. We're limiting breaches to airlocks and to turrets, where you can blow a turret off, and so you'll be able to go in that way, and you'll be able to have some door gameplay where you blow open the door, or the airlock, I mean. But breaches anywhere in the hull are really difficult, because it's not only the outer layers that you have to deal with, it's the interior meshes in there. So in order to do that, like I said with the Titanic example where we spawn some geometry on its death, if I get that feature, we might be able to switch out a wall in the interior to have a hole in it that matches the exterior. But there's a lot to think about.
If it was the only thing we were working on, that could be something. But considering everything has to play well with everything else, there's always a resource trade-off.
When I'm doing the setup, I could definitely shoot through one side of the ship to the other. My goal is to actually stop that from happening, so you don't see anything actually going wrong. So I'll tell the modelers, oh, we need more damage POMs here, we need more geo here, because we should be able to shoot through there. But something has to stop you from seeing into the interior, because they also have different lighting conditions. The environment probe of the outside world doesn't necessarily match the one on the inside. And there's all sorts of object containers and things going on that don't blend well together right now for actual straight-up breaching.
And we've mostly been working on small to medium ships to this point. As soon as we get into the larger ships, that could become something that we want to do. But we'll have to plan it out very specifically. It can't just happen procedurally or organically.
TL;DRShip balance never stops in an online game, and since everything is still alpha, the team expects to keep trying wide-ranging changes each patch and retest as new things are added.
Yeah, the best way I've always thought to describe balance in any game is it is a constantly moving pendulum, and the best you can hope for is to shorten the swing arc, but you can never expect to stop the pendulum from moving. And it always needs that back and forth. And also in that same analogy, something being in an alpha, that is when you should expect and should want the swing arc to be as wide as possible, because that just means you're covering as much ground as possible, you're trying out as many different ideas as possible, you're iterating to really dial in on what is that appropriate range. So when you have the live product, what is your acceptable short swing arc of balance between, okay, this may be a bit more powerful now, but this can get bumped up later, but then that'll change something back. And you always have the back and forth, but it's just constraining that when it's appropriate to constrain it.
I think that's one of the things that can sometimes get lost for our project versus every other project out there. Because everybody else is playing every other game much farther down in its development, if not in its release, then in beta or in early access or whatnot. We are still so early in our development. Since we're lucky enough to have a live client, so we can have far more testers than just our internal QA team, that affords us the opportunity to try crazier things with each patch: to try bumping the speeds way up this patch, to try pulling all the speeds back, taking out crews, putting in crews. It allows us that freedom to try all these different things. What we're doing today is not necessarily 100% representative of what there will be when that day comes, when we're like, we're launched, we're live. But it's that thing that we want to test now, at this period.
So that's one of the great advantages of this project, that everybody who plays it is a tester. It's important to never lose sight that there's Evocati, and there's PTU, and then there's Live, but our Live environment is still an alpha environment. All Star Citizen environments are test environments, every single one. So we're going to try all kinds of stuff.
TL;DRArmor details are still being worked out; deflected shots that then damage other ships are unlikely because of the physics cost, though a visual rebound with no further damage might be possible.
Some of that's still being worked out, just in terms of how different pierceability things happen. Now, in terms of having it deflect and then go hit another ship and do damage, while it would be cool, that's a little bit more unlikely, just because it's like, all right, now let's do all these extra bizarre real-time physics calculations based on angle of impact and angle of deflection. I would imagine Matt and Patrick would definitely be able to say better if maybe we could get some visual facsimile in terms of just how a shot could rebound off if it doesn't meet a pierceability value. It still would probably lose any damage after that and then just hit a very short decay cycle before it just stops being a live round in the universe.
TL;DRThe team says they already add every visual they can until performance suffers, so better optimization mainly means more ships at full detail in big battles; staggered startup lights and X-ray sub-component targeting are mentioned as plans.
I think almost the idea of optimizing enough where we could have another cool ship in the same scene at that same level of fidelity and detailing. Getting us to the point where we can do these bigger battles, where, hey, if we can get to the cap ship versus cap ship without having to pare down the detailing to keep both of them existing in that same screen space at the same time, if you're a fighter flying between that same fight...
For lighting specifically, I mentioned we were going to have these more interesting startup animations, so the back lights will come on and then it'll kind of raise up on the front and the cockpit lights will come on. So it'll be kind of this orchestrated way instead of everything just kind of coming on like Windows. It's a more staggered thing. But one of the cool things, and I don't know what made me think of this, it's totally not related, is we're going to look into sub-component targeting, and I'm hoping you will get this kind of X-ray view where you'll see the interior component. It's still being designed, but we're planning for something along those lines so you can pick apart.
Even if we didn't have it, that would be a benefit to people who know the ships. You could take it upon yourself to learn the interior layout of as many ships as possible, so you know where the components are physically on that Constellation or on that Javelin, so your own personal knowledge can benefit you through how well you can disable a ship.
Doing that legwork should definitely yield you an advantage, because if you put the time into it, there should definitely be a reward for those kinds of things.
And we've talked about different scanners, and we'll probably upgrade scanners, and the best scanner in the world will be able to look really into the ship and pick it apart, and you'll find the resources that you can harvest. But the basic scanner is just going to be 'ship', can't see anything.
What aspects in particular? Well, I like the challenge mainly. You come in everyday, and it's a different type of puzzle to solve. You constantly learn new things, and there's new tasks. I need to be specific. Lately I've been learning how to put ships together and then tear them apart again and that's cool to me. You go in, you see the model, you see how it's put together; you yank out the pieces, put them in the xml and then you go in the engine and you shoot it. Then it blows apart, and there's explosions everywhere.
Yeah, that's great. Well for me the most rewarding part for me is just when the whole thing comes together and you finally get to climb into your ship, turn everything on and take off and fly. Ultimately, I mean obviously I enjoy doing all the artwork for it and the struggles you have to go through to make that happen, but when you actually get to play the asset and everything works the way it's supposed to, that's the most gratifying moment for me; cause ultimately I want an awesome game.
Yeah very easily. The last couple of days I've been very specifically been working on the beginning development stages of the Caterpillar. That means doing some whitebox work, that means taking some concept work and bashing it in to make sure it all conforms with the gameplay. It also means starting to do some work and research on how materials work. So specifically we're putting a lot of effort into developing how the paint wears on the metal in the Drake ships because they have a very different style than Aegis for example. So that's what I've been up to.
Ok. For me in particular transitioning from optimisation to helping create these, they're called ship set up levels and it's where you bring all the pieces together. In terms of optimisation i've been finishing up just the level of detail pass on all of the flyable ships in the game to get a little bit more performance out of them.Did a segment recently on that, goes into more detail. But, yeah! Putting all the pieces together, trying to make the work flow and the pipeline a bit easier for other artists. Just go in, check out the ship, shoot it, look at it in different lighting conditions, look at all the lights etc, etc.
To me in particular, well, to me a successful ship is one that has made it through the entire pipeline in an efficient manner. What that means, in particular, is going through all of the different phases of development , from concept to actual modeling, texturing, lighting, the whole destruction system, and just making it into the engine, and performing optimally. To me, that's a successful ship. That means that it went through, you know, who knows how many hands and then ended up, you know, boom, in-game going, ready to go. You don't have to go back and fix things that you've broken. That's a successful ship to me.
Right, I mean I can agree with that, to some degree. I know you are really focused on the development part of it, but I think to me, a successful ship really is a ship that when I first show it to someone, they're like holy -, I wanna fly that thing right now. That's basically a successful ship. And I hope we can develop a lot of those.
It's kind of large question. There's a lot of variables that come into it but really ... really every ship is a bit different. Every ship is its own puzzle you need to solve. And your general goal is to get them down to roughly the same level. So what I mean by that is at least just in terms of level of detail, when the model slowly disappears we'll focus on that. That's really important. That saves a lot of performance.When you are up close to someone, you see the ship: it has all of its internal stuff, all of its external stuff. All that full detail stuff, right? But when you are far away, you may not realise by you are not rendering all that stuff. Like things get culled: the interior might disappear, the model might degrade in quality a bit. Not noticeable. That's where it comes down to. Things like that are where we come in and we look at the statistics and we say "How is this performing? How many ships can we have in an area? How much detail can we get at various ranges?" And then we just optimise from there.I mean for a specific number? Really there's not specific number of assets or ... it's just like, you know, whatever people's computers can really handle based on what we use to work with.
Yeah. I mean the beauty of having this game run on a PC is that the more power you have, the more of the beautiful visuals you can have rendering at the same time.
So part of it is hardware related, but obviously we're trying to optimise as much as possible to make the game run as smooth as possible for as many people as possible.
Interesting, so I spent a lot of time working on the Merlin and I made that thing so that you'd be able to open up every single compartment to access, every single component, but for technical reasons I had to rip all that stuff out at the last minute and cap the holes to get the ship to run because it was pulling way too many draw calls.So that's one example of many that we've had in the game, but the process of developing a game often tends to be trying to go as far as you can with a specific type of idea and developing it as far as you can and realizing, "Oh man, this isn't going to work, so how do we solve this problem with less resources?". So that's just kind of a theme for game development in general and that's what happened with the Merlin. It made me sad because it looked really awesome when you pushed the play button and everything popped open.
Yeah it was very cool. Maybe we can get that back into the game at some point in the future, but at the moment that was one example of what we had to tear out.
Well actually not really much at all. The concept for the ship and the main base ship and everything that makes it up is based on the Manufacturer Style Guides. LODs are based on the actual geometry for the ship itself and the best way to optimise that down to a level where we can include it with all the other ships and maintain high performance.So LODs aren't really based on style guides. You might, you might try, you obviously want to keep the silhouette of the ship and keep it looking like, you don't want it to just look blocky. So if this ship has these nice curves along the edges you don't want the LODs to just block them out or chop them up, you want them to maintain that shape.
Right. I mean what that does mean though for LODs there are some ship styles that are more difficult to deal with, because of that. So a very blocky ship, generally speaking a very blocky ship, would be easier to LOD because you wouldn't notice the changes in the surfaces as much as you would as a ship that's a little bit more rounded, so yeah. Style Guides don't affect it, but the Style Guide does, indirectly affect, how hard it is to make the LOD for it.