Star Citizen: July Subscriber's Town Hall w/ VFX Team

Michael has an excellent answer to this one. You talked about the scaling thing.

Yes, actually. So in this game, it's an RTS game, we were not concerned about the scale because you had a fixed camera distance, but you cannot do that in Star Citizen. So let's just say in RTS games you can have sparks which are as big as... But in Star Citizen you can get up close and far away. You need to get scale right, and that sounds simple, but actually it's making the huge difference in how you approach your VFX. You need to take care of the close and far read of the VFX, basically. That probably applies to all of us in this game.

When I'm making the impact for the laser cannons, for example, it's not enough that it looks good from the ship in the cockpit. It has to look good if you shoot me where I stand and you shoot one meter ahead of me, sort of. And that's tricky.

Very often I'm doing an explosion and it looks really good like 10 meters away, but then I'm up close and I'm destroying just a vehicle and I see nothing. Like, what happened? It's the worst explosion ever.

And besides from that, it's the fidelity that we are striving for. I mean, the quality is much higher than any game I've been working on. That's a challenge, but it's a nice challenge.

This kind of question should be asked of the graphics guys, as those kind of complex systems, such as clouds or, let's just say, auroras on the planet, those things are handled by the graphics team. They are using completely different technology, which should be very efficient, very generic, and we are not handling those super generic systems.

Yeah, but I think also, I mean, we've been looking at those videos. We think it's one of the coolest things to see, like the swirling after helicopters.

That would be super cool to have.

But I think we would probably... It could never be that an actual cloud gets turned into smaller wisps. It has to be some kind of smoke and mirrors situation where we might fade away the cloud and then spawn particles that make it look like that's what's happening.

That's definitely a very complex and great challenge.

Yeah, so basically we want to do that, but the question is... We'll see what the future brings.

Well, it's not mainly other games. We look at big movies that have spent millions in their VFX budget.

Mainly that's our source of visuals.

And we tend to look for movies that have a scientific style to them, more than just doing VFX for the show of it, sort of.

In movies they can get away with a lot of things just because they are making effects right in one shot, but that doesn't really work in our game. So we are very picky about what we are looking at. But also, I don't know about you, but I find a lot of inspiration in some science fiction books. Especially there is a series called The Lost Fleet by Jack Campbell. It's a great series about warfare in space, and he has really great descriptions of how this could look, and he's really paying attention to scale and time. So when you're actually dogfighting, they take like two seconds because vehicles are so fast and you just don't know what's happening. I just find it really inspiring, even though I'm not seeing that. It's really cool to read that and just think of those things.

Wow. So you get the inspiration from books. Amazing. I'll write it down. But yeah, if we look at Guardians of the Galaxy, for example, I can see something that I really like, but often I have to try to make it more realistic to suit our game.

So first of all, when I'm doing interior damage for vehicles, we are first looking at what manufacturer that is. So let's just say Drake is a really low-tech vehicle. So based on that, we know what assets we should use, like smoke, sparks, fire, those type of things. We wouldn't use that for, I don't know, some kind of Xi'an vehicle, because it's a high-tech manufacturer. So the assets I would use would be different. But for a Drake Caterpillar specifically, I found it really cool because the interior is kind of dark and grungy. So it was nice to put tons of fire in the engine room, let's just say. The whole room is burning. I can imagine that players would be kind of scared with what's happening. Like, what's up with my ship? It's kind of like a living torch now. The first thing, we decide what manufacturer that is, then we look at each room and we're trying to dress that up to, first of all, make sense, so sparks shouldn't come out of, I don't know...

A ladder.

A ladder, yeah. Why?

A high-tech ladder.

Exactly. Or water shouldn't be dripping from a gas container. That doesn't really happen.

There's five of us?

Five or six. There's Caleb in Germany. He's handling a lot of planetary effects. So the sandstorm demo you've seen, I guess, at Gamescom, that's his work.

What's his title?

Senior VFX artist?

No, he got promoted to... Why can't I remember this? Principal. VFX Principal. That means he is often leading the technologies. He finds out about new stuff and makes it work in our game.

So Oli, he joined recently. He is handling cinematics, so Squadron things mostly.

And he's a good simulations guy as well. And that's a really nice addition to the team. An easy example is if we need to break a wall of bricks, Oli can simulate that to look really realistic, which is needed for all the cutscenes.

And he's using Houdini, which is a great piece of software. He's doing all sorts of crazy things like procedurally generated destruction for cutscenes and things like that. We also have Aron. He's also kind of new.

Yeah, he's a junior VFX artist. He's been here for a couple of months.

He's helping me out with atmospheric effects, and I guess he was also doing some weapons before.

Yeah, he's helped me with a lot of weapons stuff as well. So he's jumping everywhere, which is good. I'm very envious of him for having that position. I wish I was in this team when I was 20 years old.

Yeah, it's a good place to start. And there's you doing weapons.

And you doing ship stuff.

We all do stuff on the side. We sometimes do things out of our box. And there's Mike, our lead.

Or even director now.

Director now. He's directing us in the right direction.

Yeah, and I think he makes it possible for us to work. He takes care of it, basically.

Is that Mike Snowdon?

Yeah, that's the one.

The funny thing I want to mention is that I never, ever met another VFX artist before I joined, and I think that applies to a lot of us. Usually a company has a single VFX artist for a whole game, and we have like six VFX artists.

Yeah, this is actually the first time I'm in a VFX team.

That's the first time I'm in a VFX team too.

Yeah, but I think I would say, yes, there are plans. The question is kind of big, but I mean, if a big ship lands on a planet, I think we would definitely want to leave an imprint there, but don't quote me on it. And I think, I mean, you could go to extremes, like if you're crawling on the ground, that you would leave some markings after. But I mean, the tech, we can do that. But that level of detail is... there's so much stuff we would do before we can do that thing. Like, for instance, footprints for each ship, and stuff that we maybe show, like weapon decals. Like if you shoot glass, there should be cracks in the glass. If you shoot ice, something similar, stuff like that.

It's one of those things where it's one thing to be technically possible to do it, but nothing exists in a vacuum. We have to see how expensive it is with regards to every other system in the game, because there's only so many things we can do at a given time.

The same amount of time that would be required for that could be time spent on something that yields more.

And also sometimes the technology is there, but you just start doing that and you realize actually the technology is not good enough, or it's just too slow and things like that. It's fine to have five decals, but what happens if you have 5,000 on screen? That makes a huge difference if you start adding them all over the place. That might change things.

Yeah, so you're saying having the technology is not enough. It has to work in this game as well.

Yeah, for a long time, every weapon had one impact effect. So laser cannons had like a red explosion to them. The ballistic cannons look like a more ballistic effect. But now I'm really trying to add for 3.0 that if you shoot ice, especially the materials that we use on the planets... So if you shoot ice, we want to see that red laser explosion, but with ice as well. And the same for sand, rocks, metal. I'm not doing all the materials this pass, but, oh, here we have ice or snow. So the middle one is the ballistic Gatling and the two on the side are the new laser look. I changed it. It used to be a red explosion, but now, I think we have another video, the new laser has more contrast to it. Yeah, here we go. So what we see here is sand and stone, basically. So they're pretty similar, but I think it adds so much more to the scene when you have this kind of stuff instead of just one impact. So it was fun coming up with a way of doing this, because actually what happens here, the ballistic impact and the laser impact, they are both showing the same ice effect. They share that. A common way is to bake them together, so that the ballistic has an ice version and the laser has an ice version and the neutron has an ice version and so on. But it's very time consuming to keep that up in this game with so many weapons and so many cannons. So we're using a system where the ammo, the ammunition, they share the surface-specific particle effects.

So it's more generic this way, faster.

So the experiment was, will it still be interesting even though they share the impacts? And I think it's definitely worth it, because if I make one change for the laser, sort of, that's going to get carried over to the ballistic impact.

Yeah, I can nerd out a bit on that. The game has a decal handling system, but there were big improvements made to that. But one side effect of that was that we couldn't spawn footprints the correct way, so to speak. So right now we're spawning them as particles. So it's a particle, but it works and looks like a decal. The other system is much better, because if you stand still, you take one step forward and then one step back, then that footprint, if it's handled correctly in the decal system, since no new decal is being spawned, it can be there for a really long time. But now with the particles, we can't guarantee that. We have no way of dynamically making a particle live longer or shorter. The only thing we can do is one set time or a random time, so to speak. So all the decals for a while will be quite short, because it's very expensive to have long-living particles in a game. We basically want to get rid of them as soon as possible. But in the future, when we get the functionality again, I think they're working on it in Germany actually, then if you shoot at a wall, for example, just with your P8-AR, those bullet holes will stay there for a very long time. But if you're playing a multiplayer match and there's a lot of you shooting, then they're going to stay there for a short time. But that system will handle it dynamically. So depending on what's going on, it's going to stay there for shorter or longer.

So for now, we're using the particle system. They won't persist for very long, but we are working on a decal system that will take over these things, and then they'll be able to persist for much longer. At least that's what we're exploring. We can't sit here and say that's absolutely going to happen, but it's what's being worked on right now.

So basically the system is there, but the decals aren't visible.

Obviously we want them to last as long as possible, but there are performance constraints here, so we are going to find the sweet spots.

Well, quite a lot of them, actually. But probably the favorite one I made is the de-res effect in Star Marine. When you get yourself killed, your body will turn into those small cubes. And we have a video for that. That's the one. So you probably can see that if you've seen the latest Tron, and that probably goes back to our conversation about movies. We are watching a lot of things, obviously, and that's heavily inspired by the new Tron movie, when they are fighting with discs and the other guy gets killed and he turns into those small cubes. So that's the obvious inspiration. I found it really cool to make, actually.

It's one of those things that only Star Marine allows you to do, because it's a video game within a video game. You wouldn't want to be able to use this effect in the Persistent Universe, but in Star Marine...

Yeah, that was the point of it. Like, okay, so it's a game inside the game, so what can we make to make it kind of like a simulation? So that was one of those things. And also a high-tech explosion. That's the second thing I made, and I'm really proud of that, and it was fun to do. So for some vehicles we are using the, that's our internal term, low-tech explosion template. So all explosions which have smoke and fire, those are so-called low-tech explosions for us in our department.

And they're often orange, yellow.

Exactly. And we needed a new type of explosion for high-tech vehicles, and we decided to go with plasma blue. And that's the result of this work. So we are going to use this template to iterate quicker on new explosions for high-tech vehicles, mostly.

Yeah, so for example, if I were to make a new big impact for a plasma weapon, I would probably start with what Michael did in this explosion and get inspired by that.

Exactly. So now we have default settings.

We would know what the bar is.

Yeah, that's the starting point. It cannot go any worse than that.

That's a really good question. And we want that definitely to happen. It's an obvious thing that you're expecting while flying in the atmosphere, but unfortunately it's quite complicated to make it look good. So we are looking into that, and actually I think a few weeks earlier we briefly discussed the so-called signed distance fields in one of the ATVs.

About three weeks ago.

Exactly. And basically by having that technology... and by the way, signed distance fields are just voxelized versions of objects, basically. You can voxelize anything, starting with that, ending with the vehicle. And that is a very GPU-friendly representation of an object. Creating a 3D mesh, it's not really GPU-friendly in some ways, and signed distance fields are just a different way of representing an object. And by having that, we can do a lot of very systemic effects, such as sonic booms. And we want to use that technology to improve all the atmospheric flight effects. So atmospheric entry, debris going through the atmosphere and leaving a trail behind. We are going to use them for a GPU particle system in the future to do all sorts of odd effects. But at this point, we don't have them. Definitely we want to have them, but as always, we'll see what the future brings.

Yeah, but those are in the works. They are definitely coming. And I mean, we have them already, but we still need some ways to hook into them. But the cool thing about sonic booms in this game is, I remember from working with, I think it was John Pritchett, who we did a lot of atmospheric entry stuff with. He sort of designed how that works. So I imagine the sonic boom will actually be physically correct, right? It's probably going to happen when that ship reaches Mach 1. And that makes this game cool.

Okay, firstly, these are all lies. All lasers are red. So what we did was, every maker had its own laser color. So one of them was red, one of them was green, and I think maybe one was blue. But we sort of painted ourselves into a corner then, because what color would a neutron projectile be? Because we think it's important for the player, like if I see a ship firing, I want to see on the projectile what kind of weapon damage it's doing. So we decided, let's make all the lasers the same color. All neutron weapons are going to have the same color, but they will look different depending on the maker. But that's for now; they look the same, but we have plans for... we want to really... I mean, it's not going to be super visible, like that's a Behring or that's a Klaus & Werner, but we want it to be distinguishable. It will be at some point. And on the FPS weapons, we're working with what we call a purity scale, that will guide us. So a maker will have a certain value on that scale. So then you can see, we have one maker, the pirate one. What's that one?

Of FPS weapons?

Yeah, or maybe Apocalypse Arms? So maybe that one looks a bit more rough than a super polished Gemini weapon. So we will maybe add some smoke elements to the non-pure version. So that would be one way to distinguish them from each other.

Can we look at that? It's also my favorite one. And I want... okay, so now, look when it goes from charging to charged. There's a wave there through the barrel. I think that's such a nice detail. I think the sound guys did an excellent job on this weapon. Everyone did. This weapon with sound is just so nice. That trail there is not final. I think it looks better now. The video is quite old, actually. But this weapon was so fun to do, because previous to this, there was a charging and overcharge system, but it needed more functionality. And we got a new hire in Germany, Thorsten Knuck, who came in, and he's an old CryEngine guy, or Crytek, you say? So he could get in there and solve a lot of stuff. So we worked closely, and together with all the other departments, we managed to make this weapon. It's totally unique. So one thing that you don't see is it actually has, I think it's 16 muzzle flashes firing at the same time. That's not a usual way to go about it.

Staffan showed me the setup there and I was like, what's happening here? It's not normal, but it is for this weapon.

Cracks in visors is something that you should expect. And burn marks... In this game, the suits you wear, the loadouts, they are so intricate. So making a bullet hole the normal way might look really odd, because imagine if you have a set of tubes or something; to smack a decal on there would look really bad, because it just wouldn't look real. And the same if you have an ammo magazine and you get a laser burn mark that's circular, and how that would look on your magazine pocket and your skin, and it wouldn't work. So it might be that we go with burn marks instead that look good regardless of what you're wearing.

But that's also one of those things that probably sounds simple, but making decals on a character, that's a whole different...

Yeah, they have to be skinned.

Exactly. Technology, that's a different thing, because the mesh is moving. The objects on your armor are moving. Normally, I guess, when you're creating a decal, you're just placing a decal on the ground and it stays there because the ground is not moving, and that's fine. But for a character, there are things that you need to consider, and all that's just more complicated.

It's absolutely something that we want to do. That's for sure.

Yep. Definitely.

What are the differences? Or what are some of the differences?

Well, we have some effects which are looking different, and some of them have small differences. So let's just say an explosion. In space, often you have explosions with debris, and debris in space doesn't have gravity, but the same explosion on a planet would look different because it has gravity. That's a simple example, but we are having a whole new set of atmospheric effects coming on, like contrails and engine trails we've shown a few weeks ago. Let's focus on contrails. First of all, you don't see them outside the atmosphere. And in the atmosphere itself, the effect looks different. So based on the density of the atmosphere, the effect can be more or less visible. On the moon, where the atmosphere is not thick, the contrail will be just very subtle. You won't even see that. That's what you're expecting. On the other hand, we have engine trails, which are only visible in space, but there is a transition zone when the atmosphere gets thicker and thicker, and that's the region where we are fading out those engine trails. So there are different rules we are applying for different effects, and we do that where we think it's needed. So let's just say we are also discussing having different engine effects in the atmosphere, because we think that the same force applied to those particles in the engine, they should just look different in the atmosphere, because you have density, you have humidity, you have all those different variables. This should just look different. For some effects you will see in the future that there won't be too many differences, but we want to do as much as we can to make them look different, like react to the actual atmosphere. So that's your answer.

I have a simple addition to that. In space there's no gravity, of course, but on a planet there is. An addition to that is we have to make everything work with everything in between as well, because our gravity is going to be gradual, or, it's a gradient. It's not either on or off. So you think you're done with something and then, okay, so how does it look with 1.5 gravity? Like, oh, okay.

Yeah, I would say my surface impacts, or surface impacts in games overall. But to me, it's really important that not one impact looks identical to another one. I want randomness and dynamicality. Yeah, that's a word. So a lot of work goes into something that I imagine some people maybe just don't see. Do you have an addition here?

Everything. But really, for some effects, like the high-tech explosion we've shown today, it took seven weeks to finish. So it was iteration after iteration after iteration. And the funny part is that I was trying to add more and more and more, and the end result was really muddy. It just didn't feel right. And Mike just took my explosion, he removed half of the things I added, and he said, yeah, that's your explosion. I was like, yeah, that looks cool. How did you do that? And he said, I removed all of the things. So it took seven weeks for him to remove half of my things to make it look good. And that's the good result we have in the end. So yeah, there are a lot of those things out there.

Yes. The current shields are not done by the VFX department, but with the signed distance field, we will eventually take another stab at it. Not another, our first stab at it. So the new thing that we can do: if you fly a ship, someone shoots both your wings off and then hits you again, the new shield will take the actual current shape of your ship. So the shield would no longer show like a ghost wing or two ghost wings.

Using the signed distance field.

Yeah. So the signed distance field is the best way for us to get the shape of a ship. Normally it's often done with a low poly mesh, but the signed distance field suits particle work even more.

But that shows how generic those things are. I mean, maybe generic is not the right word, but they are very flexible. We can use them for sonic booms, we can use them for new shields, we can use them for atmospheric entry. Having one technology, we can do that much.

It's a big problem for atmospheric entries. You have this ship coming in, but maybe it has this shape on it, and then you have another ship that looks like this. So we want to have a system that we can apply to every ship, because it's not feasible for us to make a custom atmospheric entry effect for every ship. But a signed distance field would allow us to get this shape and especially design it so we can spawn glow particles from here and from the wings or whatever. That would be very difficult to do without.

That's a very good question. So sometimes... We try to go for realism first, but for the railgun, for example, if you look at what I did for that, it doesn't look anything like a real-life railgun. But I really tried to make it look like it does in reality, but the reality version is honestly quite boring, especially the trail, because there's nothing there, and the projectile is just a bullet. And I asked Paul, I did a mock-up for him, like, what do you think about this? Because our railgun needs to be charged as well, and then you have to indicate that the weapon is charged. So already there we're crossing a line, I think, because a real railgun can be fired without you being able to see that it's ready to be fired. So then you might as well go for that extra thing. So it almost felt like the weapon itself, the Scourge Railgun, called for something else than the real stuff to look good. I don't know if I'm making sense, but that's one example. Sometimes the reality isn't cool enough.

But that doesn't really apply to just us. VFX in general, you often see the reality and then you know it sucks, so you have to make it more interesting. To be honest, I have a friend who made VFX for... he's more working on Game of Thrones now, but he made VFX for a movie called Pompeii.

Yeah, it's Jon Snow.

Whatever. The thing is that they have this volcano exploding, and they made a first version which looked really real. But guess what? It felt like it's not finished, like it's wrong, just because the reality sometimes is not what you are expecting to see. We are always discussing what we should do.

Yeah, because we're so influenced by movies. That affects what we're expecting to see. I was looking at real bullets hitting actual steel, and it's the most boring thing ever. So sparks have to be there, because that's more fun, basically, sometimes.