Developers discuss the hand-built, live-in-engine wrecked Idris R&D, new lighting and lens flares, and plans for derelict ships. They also cover the FPS cover system, Star Marine targeted for 2.6, ladders, prone, animations, bone-collider hitboxes and the new player health buffs.
It's always kind of been the plan from design to have derelict ships in the game, whether that's floating in space or crashed on a planet or underwater, these kind of interesting areas that you can put the ships that already exist into the world of Star Citizen. So really what triggered that was that, pretty much from design. It's something that we are planning to do on every ship. It's not solely the Idris. As mentioned, it's like an R&D phase. It's by no means final. It'll probably end up looking better than what you've already seen. But it's really kind of how much can we get away with and how much mood and feeling can we put into these areas before it starts to feel like it's not part of Star Citizen again, if that makes sense.
TL;DRThe scene was fully live in the engine, made possible by the new lighting system, physically correct light-driven glow and improved particle lighting that create layered pools of light and dark.
Yeah, the segments afterwards from Ali and the graphics guys was really what made a lot of this possible. It means it's not really affected the lighting per se. I've got new tools that we can drive the glow of an asset via a light, which is physically correct. So the best way to put that is if you've got a light that's 60 watts in power and you attach that to a very large light fitting, that glow is going to be very dim. If you attach that same bulb to the little LED on the front of your PC, it's going to glow really brightly. That's a lot of what made that stuff possible. We also then noticed some severe problems with the particle lighting at the time, because when you're creating environments and spaces like that you really want strong use of volumetric lighting. So that's something that was really, shall we say, pushed. I think the best way to put it was I ran around the studio flapping my arms in the air saying I can't make this happen without this working, and everybody got on board, which is kind of cool. And it means that the particles, Mike and his team and obviously Ali and his team, they all light correctly, and it means that you can create a very, shall we say, pooled lighting scenario. So one thing that the team wanted to achieve with the Idris in specific was that you're always looking through layers rather than looking at one room full of smoke or fog or vice versa, is that you've got pools of darkness against pools of lightness. And any artist will tell you, when you're going for composition from any angle, that's what you're looking for. You're looking for layers of depth, and the particles allow us to have layers of depth around corners, so you're rendering light that you can't see, which is always very powerful.
TL;DRThe shown capital ship damage was a static approach; procedural damage is planned via ship states, spawning debris is possible but tricky, and rotting "husk" wrecks are something they want to investigate.
So the stuff that you saw yesterday is focused towards a more static approach. I'm never going to say never, but the plan for the procedural damage is that of what we've already been investigating with the Retaliator, the Starfarer: we've got these states that we can go through. I think we've got five on the go now, which is obviously the on, off, auxiliary, emergency and evacuate. So all the lighting and atmospherics that you saw yesterday, that's perfectly achievable. We can do that. The spawning of the debris and the meshes, it's possible, but you'd have to be clever, because you wouldn't want to be walking down a corridor and then this debris spawns in front of you. I know a lot of people might think that's okay, but it's not okay on my front. So there are tricks that you can do there, like a lot of games with ragdolls, for example. You'll look at a ragdoll and the ragdoll's on the ground, and then you turn the camera and look back and the ragdoll's gone. So we can do tracing to understand where the player is and spawn things in and out accordingly. But the focus for this was, it's taking it, I would say, 60% towards its complete death. So the one that you saw yesterday was focused on: it's just happened, it's just gone down. You'll notice there were two distinct styles going on. The first minute of the video was very flowing and moody and dead. And then the second half was, okay, this side of the ship hasn't taken the impact. It's still functioning. It's still trying to put fires out. The emergency lighting is still going on. The sirens are going woo woo woo. So there's two archetypes going on there. But beyond that, there is again high-level plans, what we're calling husks, which, if you think about the opening of Episode VII, when she's exploring that Star Destroyer, that's something that we certainly want to investigate as well. So we'll go to a point where a ship has potentially been crashed for years, so it's rotting in its state, whether that's in a swamp or whether it's in the desert. That's something that we can do, again, with mostly shaders. And the beautiful thing about doing all this stuff is that it actually becomes cheaper and cheaper and cheaper in the engine when you do it, because you're not rendering so many lights and you're not rendering so many particles.
You'll see in the game a better than that fidelity. That's the accumulation of about seven weeks' work. Obviously, we've got the base Idris. It made sense that we use that as the starting point to investigate this stuff. I think, as I mentioned in the video, it was really just trying to be efficient and clever with the assets that we've already got and see how far we can go. And I think we've come out here pretty much about 80% of what we wanted to achieve. I think it looks quite powerful, it looks quite strong, and it's, like I say, something that we'll be doing on probably every ship. But it wasn't R&D for R&D's sake. There's always a bigger picture behind what we're trying to achieve, if that makes sense.
TL;DRThe damage shown was handmade as an exploration of the engine, with hand-placed geometry plus cloth, lighting and volumetrics to create a jungle feel, and caustics planned but not yet working.
It's handmade, but that's the exploration. A lot of the mesh work is hand done. Obviously, at the very beginning, when we started — I worked on a couple of Crysis games, and I think I touched on in the video that we wanted to do something different with water, and I've been dying to use the engine to really its full extent. So a lot of cloth, a lot of lighting, a lot of volumetrics. You'll notice that there's cloth and stuff moving, broken glass. It's something that the engine does really well. We never had the opportunity to use it. So it's really a case of, okay, let's go for it and see what we can achieve. So to get that, obviously, there's a lot of hand-placed geometry. And to get what I wanted, that jungle feel, so the dense, moist air, and more organic shapes being created from the very hard surface shapes that are already in the Idris. So if you study the video, you'll notice all corners have debris brushed up against them, and that's mimicking, shall we say, piles of leaves up against a tree trunk. You've got a lot of vines and stuff hanging from the ceiling; that is mimicking that. Dappled light: you're in a jungle, there's a lot of dappled light coming through. So there's no real — I think there's one area where there's big lighting going on, but the rest of it is all very dappled little bits and bobs. Leaks, puddles. We haven't got the caustics working yet, but we will be able to reflect caustics. Caustics are something that's generated from light reflecting through water. So if you go to the swimming pool, you get all those lovely patterns. The plan is to have that in there as well. So a real jungle feel to it, if possible.
TL;DRThe focus was a terrain-surface wreck, and wrecks in space could use particles and sound for effects like decompression sucking particles toward a target, which the particle system can already do.
As mentioned, I think in space, any kind of terrain surface. This focus was on a terrain surface one. So the way that everything behaves in the ship is perfectly how you'd expect it. With the space wreckage, you've got the opportunity to do some very creative things with, again, the particles and the sound. So, for example, when you open a door in zero G and there's a decompression, I'd expect all that moisture in the air and all that thing that we try and achieve to get sucked out. And that's doable. There's already, in the particle system itself, we can tell particles to get sucked towards what we call a target. So you trigger a vent and everything gets sucked towards that target. So there's little things like that, little cracks in windows. You may have the whole — what was the bad Alien film? You're the Alien man.
Alien Resurrection, where he gets sucked through that little hole at the end. We can do that kind of thing with particles, if that makes sense. They're a real physical thing in the world. They may not seem like it, but they are. You can expect to see stuff like that, I guess.
TL;DRHe has no single favourite effect but is fascinated by how light interacts with surfaces and atmosphere, and likes the engine because everything is real time with nothing baked or faked.
I don't know, that was the question. I guess when you have so many tools, a lot of artists will tend to favor a specific bloom or fire or whatever, and sometimes they abuse it a bit and everything has this. Is there a particular effect, like the particle lighting or whatnot, that you...
For me personally, I'm always intrigued by how light reacts to the surfaces and how it reacts to the atmosphere in the air. There's not one thing specific that I like over another. I like to study light a lot. I like to understand how the sun comes through our atmosphere, scattering. I will sit there and read books about that, and then how you can use that to your advantage. When you're developing a piece of 3D artwork or an environment or something like that, if you've got an understanding of how light works, you can totally play it to your advantage. We're really lucky with this engine because everything is literally real time. Nothing's baked, nothing's faked. It's just what you see is what you play. I like using this engine for lighting quite a lot.
TL;DRNathan says CryEngine's rendering has always been among the best and will improve, citing the new automatic lens flares and saying the project is gelling visually, though he admits bias.
Yeah, the rendering side of things has always been pretty good. It's always been up there, as far as I'm concerned probably the best. But I would say that; I'm a little bit biased. And it's only going to get better. The lighting, the new flares stuff is just such a pleasure to use. There was so much manual effort that went into setting up lens flares on lights. And again, I wouldn't use them in the J.J. Abrams style; I would try and use them to mimic atmosphere around that light fitting. I don't have to do that anymore. It's all dealt with automatically, and it looks consistent. So when you look at the Idris, like standard, or you look at a Starfarer or a Retaliator, now we're starting to get under the skin of the rendering a little bit more. The whole project is starting to gel together a lot more visually. That's my personal opinion, but I hope other people will see that when all that goes into the build.
TL;DRThe hologlobe reflection combines screen space reflections with a light at the globe's centre given a bulb radius matching the globe, set at very low intensity, giving a near-perfect reflection.
It's a combination of two things there. So we have screen space reflections, SSR as you've already called it. That's okay, but it's got a nasty tendency that the falloff, which is called Fresnel, on the surface, it's not a linear gradient. So it cuts off really harshly on very shiny surfaces, and of course water is very shiny. So what I've done there was — we've got the ability to, you put a light in the center of that globe, we're able to tell the light to have what we call a bulb radius. So the radius of that globe, I think, is like 1.5 meters, so I'm able to put in a diameter on that of three meters, and then you get a perfect reflection. So you set the intensity very, very, very low, and then you get what you saw: it's an absolute perfect reflection of what is essentially a gigantic light bulb. The key with that stuff is that we want to get that then dynamically changing to the hologram tech when that comes in.
Yeah, so I'm a gameplay programmer and I'm on the FPS team in Wilmslow. So most of the stuff I do is player traversal. So I do a lot of stuff to do with ladders and cover and vaulting and jumping and all that type of thing.
TL;DRGordon says it was the first showing of the long-in-development cover system alongside FPS work on movement and weapon feel, and praises DataForge as a validated central database replacing hand-edited XML.
Yeah, so that was the first time, I think, the cover system's been seen ever. It's been something that's been in development for a long, long while. I've been working on it probably since back in November, give or take. So we've always been working on that, iterating on it. This is the first time it's really been shown off. So there's the cover system itself, and there's the general FPS stuff we've been working on, because it obviously had a lot of work. We've done loads of stuff on player movement and weapon feel and just all the core features of the FPS. And obviously you saw DataForge as well. It probably doesn't get shown off quite as much. I know Ash, the guy that wrote it, said cheers to me this morning, because that's the first time it's ever been given a shout-out.
We showed them DataForge twice. This was the second time we showed DataForge. We did a segment with Mark Abbott a couple of months ago where we showed it off. It's not one of the sexier features in game development, but it's such an important tool that we waited so long for. And Ashley Canning, right? Ashley Canning is a legend as far as I'm concerned.
As far as game development, one of the big aspects is games are basically a big database of values. It's lots and lots of numbers that you need for lots of different things. So having DataForge is just — it's a centralized repository where we can put everything, and it stores it well and it's all validated, and it checks against all, like if you've got a bunch of audio clips, it checks all the assets exist every time you try and export. Obviously, there's the tool itself with the live edit features. It's so good to use and so streamlined, as opposed to being lots of XML files everywhere that you've got to edit by hand. It's just glorious.
TL;DRWeapon raising already works in zero G, but the cover system is turned off there and will be gravity-only at first, though peeking around cover in zero G could be investigated later.
Yes, so like I said in the video, the cover system is based on the weapon raising stuff. That's the system that Ivo, one of the animation programmers in Germany, did. So that's already in, I think. It's already in the live universe. So that already works in zero G. So if you float up to something, your weapon's going to raise out of the way. The cover system itself doesn't work in zero G. It's just turned off in zero G at the moment. I could see there could probably be scope for it getting added in. At the moment, it doesn't hugely make a lot of sense. If you're floating around, you can be any orientation, you could be upside down, you could be face up to a wall, whatever. It doesn't really make sense to lean so much, but I could certainly see in the future we could have some kind of peek around cover, peek around walls, while in zero G. It would certainly be something that would be interesting to investigate. But certainly for the first iteration anyway, it's gravity only.
TL;DRStar Marine has had a lot of work since being pushed back, with new movement, weapon feel and animation refinements, and a recent team playtest felt like a cohesive, proper game.
So yeah, Star Marine has not really been talked about in like a year, I think. I think we were quite close to releasing it and then it got pushed back a little bit. We've done a lot, a lot of work on it since. Like I said, there's lots of new movement stuff, there's lots of weapon-feel stuff, lots of animation refinements, lots of stuff to just make the game feel a lot better and play a lot better. We've not long done a playtest over here, actually. It wasn't quite studio-wide; it was a team-wide playtest where a bunch of us jumped into the map and had a, I think it was 4v4, I'm not quite sure. And it's come on so far and it feels like a proper game now. It actually feels cohesive and something you'd actually want to play. It needed work, it needed refinement. But yeah, there's so much stuff being done in it. It's a smallish team that's working on it, but it is very dedicated to getting it feeling good, getting it playing well. Like I said in the video, obviously it's Todd in Germany that's overseeing it, and John over here. And they're really going to proper lengths to get it feeling right. There's loads of new features as well. I mean, simple stuff like all the reload animations have been redone quite recently. I think the last couple of weeks they went in. And it makes them feel much more — I mean, the whole thing about Star Citizen is it's as realistic as you can get, and they feel weighty now. They feel like a proper reload. And lots of other stuff, like selecting and deselecting weapons feels better. We've got crouch improvements, prone improvements and jumping improvements. And the jump was something that for a long while felt a bit pants. It always felt a little bit off, and so we got the opportunity to go over it, and all the animations got redone from the mocap. Dan over here spent a long time working those animations with Uzi as well, properly working them out, and I did the back code to glue them all in, and it feels a lot more natural.
Not yet, no. I know it's been mentioned, along with — at the moment, if you've got multiple people on the ladder, I think they climb through each other. I know there's been a lot of talk of either climbing over people or pulling people off ladders as well, but it's certainly something that's on our radar. I don't know how high the priority of it is, but again, it's something that would be quite cool to investigate, and quite cool to be able to pull your gun out and start sniping from halfway up a ladder. I know as well ladders were mentioned for zero G, because at the moment also I think they don't work in zero G. That will be something that will need to be resolved, and there will be some R&D going into that, probably.
Again, not at the moment, simply because of the pose: when you're prone, you're so low down, and for the weapon to raise up it would need to come out like this. Yeah, I mean, you can already do the rolls in prone, so I could see you being able to do something like rolling onto your side and peeking out. But again, it's not a feature we're currently working on, but it's something that'd be quite cool to R&D. Prone itself also needs a little bit of work. It's probably one of the glitchier features of the game just now. I know it's on the schedule to get looked at again. I know there's been some bugs fixed for it, so certainly that will be getting some love at some point as well.
I hadn't actually thought of that. There were non-cover leans at one point. They looked a little bit naff. You pressed the key and you just went... I know there's plans for sidling, which is like if you've got a narrow corridor, you put your hands up against it and sidle along. So I could see something similar for that: you go up against the wall and just look around or peek around.
Good question. I'm not sure; that would need to go to design. I'm not sure if they would be blocked or if they'd try and crawl over each other. Yeah, that would be a design question. That would normally be something that would be fielded by Todd or probably John.
TL;DRYes, Star Marine is getting many new and updated animations (movement, reloads, selecting weapons), and this work will also make its way into the wider game, not just Star Marine.
Yeah. I mean, on Star Marine, the two biggest teams are the code team and the animation team, because there's so much animation, there's so much mocap that's been done. There's a lot of new stuff coming in. There's already the prone rolls that are in place. I don't think the animations are new; I think they've been there for a while, but I think they have been tidied up a little bit. And there is a lot of other stuff. A lot of the movement animations are being updated. And like I said, all the reloads and all the selects and deselects. And yeah, just lots and lots of different animations. A lot of this stuff, obviously, it'll go into Star Marine, but it'll make its way into Crusader as well, obviously.
Star Marine is a game within a game. It's almost like a testbed for our players, so they can just get in there if they've got 15 minutes to play and they just want to go around shooting stuff, practicing their FPS skills. Everything for Star Marine is going into Star Citizen proper. So that sometimes gets lost. It's important to remind folks.
TL;DRHit detection uses colliders tied to the character's bones rather than a cylinder, so shots hit exactly the exposed body parts, and leaving and returning to cover is meant to be consistent and predictable.
Yeah. So Star Citizen as a whole is different from Battlefield in that we don't have a cylinder collider. It's all bone colliders. The player himself, it's tied to his model. So if you're getting shot in the arm, you'll actually get hit in the arm. If you're getting shot in the leg, you'll actually get shot in the leg. It's not like a volume around it. So as you see the player, that's what they look like. If he's sticking out and there's a little hole for his arm, you can shoot through the hole in his arm if you're accurate enough. Or if he's sticking his head just a little bit out, you can just clip the side of it. So it's very accurate. And it's one of the big things about the cover system. It's something that Todd was quite insistent on: as you step out of cover, when you move out, you know exactly where you are, and when you come back in, you'll come into exactly the same place as when you left, so it feels natural and you've got that knowledge of exactly where you're going to be and exactly how much of yourself you're exposing. Without that, it wouldn't be usable. It would be too random.
TL;DRThe main challenge is the sheer number of cover animations and transitions that must be set up and kept right through repeated iteration, which is eased somewhat by switching to procedural animation.
Biggest challenge. There's been a lot. I mean, the core thing that has to keep going through your mind is it's all about the gameplay. It's all about feeling good. It's all about playing good. So there's got to be constant refinement to make it feel right. In terms of challenges, cover has been a little bit buggy at times. There have been occasions where it used to be quite easy to get stuck in the cover pose, just because of the way it was set up. And from a technical standpoint as well, there's a lot of animations. Like I said, there's three leans: there's the peek, the blind fire and the step out. Three different types of them, and for each of them you've got four distances. So you've got to be a certain distance from the edge of the wall, and then it subdivides it down. So if you're half a metre away, it'll play a half-metre step out. If you're 12 centimetres away, it'll play a 12-centimetre step out and the resultant come back in. So there's dozens and dozens of animations. I think we've got 60 or 70-something transitions set up for the cover system in Mannequin. So it's a lot of assets and a lot of transitions to set up and get correct. And I don't want to say it's a headache, but there's been a lot of refinement on the system, a lot of iterations. So every time that it iterates and it changes slightly and I've got to go in and add another 20 animations, it's a little bit... Yeah, I mean, it would be good if the animation system was as good as DataForge, if it was as easy and straightforward as that. I mean, that was one of the brilliant things about the procedural stuff. Originally, it was all overlay animations, so there was an animation of them bringing the gun up, so it would play an upper body and it would make the gun rise. And we switched it to being fully procedural, so it's just a number of values. It's just: the gun moves up two centimeters. It's such a massive time saver, because you can literally go in and go, here's the rifle parameters, and now we need one for when he's looking to the left, so move it half a centimeter that way or however many degrees that way, and then we need a version for the shotgun. The shotgun's a bit bulkier, so copy all those values and just move it a little bit. And it saves so much time, and it's so much easier to iterate on and get the right balance between the designers who want it playing good and animators who want it looking good. Obviously, they both want the other thing as well. You get the nice combination of both aspects of it.
TL;DRNot straight away; a new health system with debuffs exists, and damage that impairs cover use is expected eventually, but it is something being looked at rather than part of the initial release.
Not straight away. There's a new player health system. I think it's already gone live or it's just about to go live. It might be 2.5, I'm not sure. And that was done by Nikki over in Germany. And it builds on the health system we've already had and expands on it, but it does add the — he calls them buffs or debuffs. So you've got like vision blur, or you've got — again, I don't know how many of these are going live straight up, maybe a bit later — but you've got like tinnitus, or you've got — again, this one I know is not live yet — but if your arm gets damaged, you'll only be able to use a pistol, and so on. Certainly in the future, I can't imagine we'll ever get away without having some kind of thing that stops you using cover, impairs your ability to use cover, because certainly the amount you step out from cover — if your left leg is damaged for whatever reason, you're not going to step the whole way out; or if one of your arms is working and one of your arms isn't working, you're not going to be able to aim the same way. So that's obviously going to come in there.
Yeah, certainly. I mean, it's part of the thing of an FPS. You've got to get the core of it working properly, and you've got to keep adding features and keep building up, which is what we've been doing.