Star Citizen: Reverse the Verse LIVE - Facial Customization System

The attachment problem. So the attachment problem, it's super interesting to talk about. And it's not the first thing you'd think about within this. And maybe we'll talk some body customization questions later because that's what the big challenge is within there, is the attachments. So morphing a face and doing all these things is complex enough. But then when you tack on attachments to it, so now you've got hair, now you've got scalps, now you've got glasses, now you've got helmets, now you've got monocles, now you've got top hats, now you've got all these kind of things, you have to make sure that those still fit every single face that you can have. And like we showed in ATV, there's a huge range of human anatomy. And generally, that's why they have multiple sizes of hats. That's why they have multiple sizes of helmets. But guess what? We can't make that. I can't go to Josh and say, cool, you've got this awesome idea for a helmet. Great. Give me 30 of those with just slight adjustments to every single one. So we needed a much more dynamic system.

We could put those in our shops. We could have a size one helmet and a size two helmet. But you could only get the one that's for you. You would have a hat shop.

You would have a helmet shop. You would have all these things. So we needed to make sure that we could implement a system that could adapt the shape of the attachments to fit, basically. So that's what we did. When Andreas Rohr came on last year, who did part of the demonstration there at CitizenCon, he was the one that kind of cracked the case on this, which was using some concepts, to be fair, from the film side that he had done most of the work within, but then applying them to a runtime implementation. So these were all these new deformers that we've created that are runtime deformers. We call one elastic, one is non-uniform scaling. And deformers are a method of moving verts around or deforming them, obviously, at runtime, based on some sort of input, and the input is the face. Now, what gets weird about it is there's different deformer types for different types of attachments, and Ivo and Andreas went through this really well in their presentation. The elastic type, basically this is one that follows per vert the offsets for every single vert against every other vert in the attachment. It wouldn't work on a helmet because the whole thing would contort into such a weird shape that you wouldn't be able to recognize that helmet anymore. So it was really important to bring those deformers online so that we could actually roll the system out, and that's why there was a big, big delay within it. We were still using the system behind the scenes; we couldn't apply it or give it to the players without solving the attachment problem.

So the short answer is yes, but not exactly as you're asking for it. I just don't want to do it in such an overt way where it's like, oh, what's your age? And then you just have this slider, and then eventually it sort of just adjusts the face. What I'd prefer is that there was different levels of, basically different variants to the skin tone. So for example, you might have that pretty standard Caucasian sort of skin tone, but I'd like to see, and I know Chris wants to see this, an old version of it, a younger version of it, a sun-damaged version of it, a scarred version of it, or basically variations within those tones so that you can basically get that sort of result. But whether you're setting yourself to a 70-year-old, because I've seen 67-year-olds that look quite young. So I think it implies a lot that, oh, you're 70, so you're going to have liver spots or something like this. And that's not a guarantee. But I'd prefer that we had variants within the skin tones. And making those is not a nightmare by any stretch of the imagination. Because we're kind of in a data-driven setup now, we can add those as we need.

Yes, but first we have to, okay, so hair. Let's talk hair. Okay, hair is hard. Hair is really hard. And not just hard for me. It's hard in real-time rendering. Hair is very difficult. And to be fair, nobody's really solved it in a new or unique way for the last, let's say, 10 years. It all uses alpha cards, and whether you use alpha test or alpha blend, there's pluses or minuses to the whole thing. So the difference between alpha blend and alpha test: you have a card, you have an alpha channel that's basically saying this is transparent here, not transparent here. Alpha test only does a singular test of that. And it says, basically, is this vertex going to render or not? Is this pixel here or not? And it does one check, and it goes through, or it doesn't go through. Alpha blend does multiple. It actually blends it rather than doing just a test against the singular pixel. When you do alpha blend on a bunch of the hair cards, the softness of the hair can look amazing. The problem is performance changes depending on the distance that you are, so how much of the screen space alpha blend actually takes up. So it's a bit of a rendering thing. So we showed at CitizenCon pretty much an update to not only the hair shader, but the method that we make the hair. So this has not at all been applied to the current character creator. So that's step one, is take those hairstyles that are there now and make sure that they're using the updated shader and the updated pipeline so that they're all going to look a lot better.

We also had to update it to unify them so that we could have them on all of our different head shapes and all of our different characters.

And make the attachment system work for that. So they use that elastic sort of technique. The trick is getting it simulating on top of that. So once we've got the visuals where we want it, then we'll get into the simulation side of it. And that has to do with the complexity of these attachments. So now think of how this works. We have this hair, and we have an adaptable face, and we have this runtime deformer that contorts the hair to fit. And it's not just a uniform scale thing. We can move the scalps. We can move all of the areas of the head, so the hair has to move up and down. And now add joint chains into that, because generally you're going to want to simulate these things with either joint chains or some sort of cloth vertex simulation. Or you go crazy and you go TressFX, NVIDIA hair. But all those systems use some form of vertex simulation to some degree piped down onto the GPU. So whether we want to implement that, that's probably a bit further than we'll go. But on the joint chain technique, we need to be able to update the offsets of those joint chains so that the simulation happens from the right place all the time, depending on the attachment changing. And the constraints.

Actually, I have a question for you. The blending that we're doing on these attachments that hair is a part of, are we able to do that in multiple steps, or is it just the one step? Like, my head is this shape so my hair has to attach to that, but if I put on a beanie or whatever, are we able to then also clamp it on the secondary step, or is that problematic?

That's a bit problematic, and I know what you're saying. So what we do instead, we do subgeometries. So that's basically different forms of that geometry depending on what the attachment is. A great example of this is the current Snoopy cap. And we finally fixed this thing. Well, kind of. It's coming back. So everybody's seen it, I'm sure. You've got Mohawks sticking through helmets. Now that we have long hair, it's more like the rule than it was before. It used to just be the exception. The one hair, the Mohawk, would always go through the helmet. We had set up a bunch of logic, but there was a code bug within here that the tag wasn't being respected. So basically what we do with this Snoopy cap: if your hair is so long that it would clip through the helmet, we instead put a Snoopy cap on the hair, but that's all adapted based on what hair you have. So when the helmet comes on, just to go to your point of putting on a beanie and changing it, we do this with the Snoopy cap. So when the helmet goes on, if the hair is so long that it would clip, we put the Snoopy cap on you.

But because of the way we do that with the subgeo change, we don't necessarily have to pick the same Snoopy cap for every hair. We can have that hair sticking out the bottom of it or whatever else.

And that's all just asset work. And generally, with the character team, I talk to Corey a lot about what he calls hat hair. So it's basically, when you put a helmet on, what's your hair look like? When you put a hat on, what's your hair look like? And it's just how many of those they actually want to make. For every single hairstyle, if you now have to make a hat hair version, a helmet hair version, that's already multiplied it by two, or three anyways, the three different forms of each hairstyle. And you just don't want to make so much complexity for them.

What I'm imagining is Pacheco's hair, but then you put a hat on top of it, but the dreads are still low.

Yeah, but again, what's that look like in a helmet? Because if you've got dreads, okay, cool, you put on a helmet and you've got dreads hanging down here. What's that look like in a helmet? Actually, what would that look like? It'd be really weird. You'd have this bunched up stuff in your helmet. So I realize I'm hand-waving, and put the Snoopy cap on and that volume of hair just kind of goes away. I don't know if you could fit that into a Snoopy cap. So, yeah, I guess short answer is yes.

Oh, man, okay, so, yeah, and I don't have a rigid plan for this, but we've all seen it. When we were starting on the character creator, we all took a little look at all the, there's a million of them. But one of the best ones, not overall character creation, but for hair specifically, is Black Desert Online. Those guys, and I think I have a pretty good idea how they're doing it, but I need to solve our joint offset problem first with the attachments. But I love that you get control strands within their hairstyles. This is super cool because then it's one of these 10x things. If we can make it so that we're only having 10 to 20 hairstyles and basically the players can modify them to give them the kind of style that they want or tweak the style, it's going to save our content creators a ton of work if we can get such a system online. And I think the Black Desert Online one is actually very intuitive, how it works, how you see it anyways. You put the hairstyle on, you've got all these control strands kind of coming off it, and you can adjust those control strands individually, and I think that's super cool. I'd love to do something like this, but it's not right now laid out in the plan as an absolute must-do.

Yeah, it already works. We even showed it in the CitizenCon demonstration with Ivo and Andreas. We showed it specifically on Captain White's beard. The tricky part is we just don't have many beards right now. That's all. It's just content. What was so interesting about the whole deformer thing too is what I expected to be the hardest in making these runtime attachment deformers actually ended up being the easiest, and the stuff I thought was going to be the easiest ended up being the hardest. The helmets I thought were going to be the easy one where we just kind of scale them, whatever. That was way harder than I thought. But the beards and hair, typically things that touch the face are super hard to deal with just because you've got blend shapes in there. You've got all kinds of stuff going on that you can't just skin to because you've got different topology. But the elastic deformers, the same ones we use for the hair, are actually the simplest ones for us to make. So what it means is that every single vert that moves around will move that same attachment as he speaks and everything. So we get much, much better results even with this attachment stuff than we had ever before, because it was just skinning. And actually, even though you might not have noticed, even in the early Squadron sort of videos and vertical slice stuff, blend shapes would clip through the beard and stuff like this. You kind of didn't notice because we tried to hide it a little bit. But this system works so much better for that. So making beards is actually much, much easier than it ever was before. What's also nice about that is there's other things like beards that go across your face. So if you can think of a bandana type thing around your face, that totally works and you can see the speaking happening below it, which is just incredible to me. That's not an easy thing to pull off. A doctor's mask is another one that I like to see, which is the exact same thing. I showed that in the internal demo, that it all moves around. And what's cool about a doctor's mask, too, is it's got these sort of loops that go back around the ears. And because of the way the deformers work, as the ears move around, that whole stretch stuff moves around. So it just makes having the singular asset, mark it up one time, and it works for everybody.

Yeah. We actually just started talking about that the other day when planning improvements for this next quarter. And conceptually having undo history and stuff like that is definitely on the radar. I think it depends on which level of undo we want to do. I think we were talking about having an undo stack. So what we do is just save off all the DNA values, which is actually a really small set of data. It's like 200 bytes. It's a really short string of characters.

So much of the complexity comes from the fact that we have all these different types of modifications. We have changing the DNA deformer as well as changing the actual items that are attached. And that's when it starts to get a little bit more complicated in terms of how we're going to manage that history.

Yeah, because actually all the eidonomy stuff that we have right now for hair, eyes, soon-to-be beard and stuff is actually totally separate from DNA. This is actually an item being attached to you. So we need to have this history of your changes be sort of agnostic to the data that you changed. Like, okay, if you changed DNA, we need to say, here's the value, refer to that value of the DNA matrix. If you had an attachment, it's like, okay, well, what item port and what was there?

And the smarter the system can be about that, the better we can have it update. If we don't have any idea what the change was and we just say, okay, here's the last state it was in, we have to reinitialize the entire character. He's going to pop off screen for a second and then come back on. It's just a way worse experience, and you don't get to see the actual change happen so you can judge it. So you have to just hold more of it in your head at a given time. So making it smarter is worth doing because that'll make it a lot more usable.

Yeah, I think so too. That main point is there's a couple of data sets in here. The main one is kind of the anatomy, the shape, the geo.

Yeah, and how the colors are being attached, how all the materials are being attached. That's totally separate from the matrix.

So I know I made the joke in the ATV where it was like, oh, I see blonde, I see brunette. Actually, that data's not in there. It's just a joke.

So we have the DNA matrix for the anatomy, and now we need a string for the eidonomy.

Basically, we need some sort of data to represent what hairstyle, what color, what skin tone.

In addition to it being useful, I think that there's just something that's really beautiful about any time we expose the data in our game, because for once we don't have to fake it. We have so much going on in this game that we can just say, here's the actual values of it, and have that represented on screen in any number of ways. And I think that it's very easy to end up with something that's kind of beautiful and engaging just on the data side of it. In so many things, you'll have the doctors manipulating all these aspects of it, and it's all just hand-wavy. It looks nice, but it doesn't mean anything. And here, it's actually tied to something that's real. And sometimes that thing that's real is still abstracted away, like the way that we do our blending. Even just having my manipulation of the world result in actual changes to it, I always find that compelling.

Especially as a programmer, when the data is written in hex, it doesn't take too much of a leap to say, oh, I can actually look at this and see what it means. This head has this much weight in the face. It doesn't take too much explaining to say, oh, okay, this is what it means. I really see what this code is doing here.

I do think it's cool. And that's why I make that matrix joke, because it really is. You're looking at it, and you're like, oh, that's head ID 9. It's blended at 20%.

And so I would have totally wanted to have a copy-to-clipboard function. So you could just paste it and be like, this is my dude. You want to tweak it? You think I have too big ears? Here's the guy. Make your own ears on it.

And especially since we reset them every major release right now, I think that's a nice way to bring it back. But one big point about it is, what we're doing right now, you'll see it on the next roadmap publish, but we have a V2 planned. We've got lots of iteration to the actual UI, the actual interface. We've been talking about it a lot this week anyways, taking in the feedback from people and everything. So I don't want to go too deep into it, because I'm sure we'll do an ATV or something like that about the next iteration of it. But the whole mentality around that is simplify some of the interface a little bit, make it a little easier to get to what you want, take up a little more screen space, try to make things a little more clear. But the point being is that we're going to change some of the content that is in there. So one of the things about being able to output that string would be, between releases, if we change that content, that string is going to mean something very different all of a sudden, because our indices are going to change, the amount of heads are going to change, the hairstyles are going to change. So unless we're very additive in the way that we add the indexes, we're always going to basically clobber that data. So once we're in a state that we're actually pretty solid on our data, that would be a better time to do it.

We can also have an internal mapping, too, where we try our best to future-proof it and say, if we're going to change the order of stuff, it won't break it.

Exactly. So for the next release, it knows, okay, well, if they are importing data, that ID 1 is actually ID 4 or whatever we want to do with that.

Okay, so hybrids. So what's super interesting about the hybrids, and what do I mean by hybrid? A hybrid is a resulting head that we can reuse as a source head again. So where it kind of started was what we found when using it, and this was over the course of the last years anyways. What I found when using it is that there's a lot of redundancy within faces, within heads. And you probably even notice it right now that we've got nine male, nine female within the system right now. But of those, for example, on the female side, there's probably four of them that are mostly the same. They're very close to the same, let's say, archetype. So there is some redundancy within there. And what's interesting to see about it is it makes the average head that people end up creating look a little heavier on the sort of Asian, Eastern European kind of side right now.

Okay, I wasn't imagining that. I thought that was me.

No, that's what you noticed. A lot of people kind of look like that. And a lot of that has to do with just the source heads themselves. So what we did in this early iteration, because we didn't want to make this thing wait until halfway through this year. We wanted to get this in front of players. The main reason was we wanted to get that feedback, because we use it all the time.

Yeah, and interacting with it is so key to the actual feature that just seeing the results isn't sufficient.

So the heads within there have some redundancy. So what we want to do instead is make each piece of data that you are sourcing be meaningful in its own way so that there's no redundancy in there. So what we're going to be doing over this next V2 thing is, what we've done is cross all the Squadron heads and everybody. And there's so many benefits to doing it this way. We've archetyped all the heads. So there's a huge mosaic that basically has every single face that we've got. And what we've done is kind of align all the people into their own archetype. So to use an example, I'm almost scared to use it a little bit, but Sandi and Rhona Mitra have a very similar archetype of a face. But then between the two, there might be unique features of each. So what we're going to do within the data sets is, instead of just the nine male and nine female that you've had since the first character creator, it will replace all of those with our hybrids of meaningful data sets. So we'll take the unique ear of that guy, the unique nose of that guy, the unique brow of that guy, all within the same archetype group, and make that a single head.

Yeah, we're distilling it to the extremes so that you can pick between those extremes.

Yeah, exactly. Meaningful data so we don't have redundancy across it. So what it does is it makes it a lot easier to use, because each of the data sources are unique in their own way, and you don't have a lot of redundancy to sort between. And this doesn't mean that we're only going to have four or five heads. Even within that archetype and hybrid approach, the reason we can do this is that I can take very specific features of maybe an actor that we can't use the full likeness of and create that into a hybrid. And that's actually not an issue at all. We're making people that aren't actual people, but are combinations of people. So basically, we're using our own system to source our system, which makes some sense. The other cool part about that is, because we're going to have this ability, it does make a pipeline for us making guys like mission givers or new heads a little bit easier. But what's really cool about it, I think, is an interesting design idea that we're trying to flesh out between myself and Todd and CR. I call it the loopback or the feedback pipeline or the sins of the father kind of idea. Legacy, or heirlooming or something like this. If you create a character within Death of a Spaceman, after you've died a certain amount of times or whatever, you can't respawn and you have to create a new character. But I would love to be able to then source back my original character as a source head within the pool. And this is totally possible because, again, it exists as a combination of those hex values. We feed it back in as a source and then we blend against it. You can only do that so many times, because eventually you will just end up back at the average if you just keep blending, blending, blending. You'll just end up back at the average over time. But I like the option for characters to do that. So I like this concept of, if you were playing the game as a criminal or something like this and you've got a certain amount of reputation, that when you roll your next character, if he's kind of, quote, unquote, the son or your child, some of that reputation comes with you. Because if your dad's a horrible criminal, maybe you start life with a little bit of that, and you have to work against it, or if you just want to make something anew. But I love the idea of being able to source back the player head.

Would we do anything where we could source from different people, from other players? Your head ends up in my pool because of a certain reason.

That kind of goes to the copy-paste of the data. We could absolutely do that, because if you could share the data between players, you could feed that back into your pool. But, and by the way, a lot of this is theorycrafting still. I don't have a plan for that. Hybrids are in. That's happening this quarter. The sins of the father kind of thing is a little bit different.

Yeah, we're exploring all the potential consequences and ways we can really leverage it.

So we've talked about possibly having the gene pool or the head pool basically controlled via the org. So for example, the org could expose all the heads, all the characters that they have within their org as sourcing. It's a dangerous road to tread, though. If they just want everybody looking a certain way or looking a certain type or something like that, they can absolutely do that.

You'd be able to deviate from it, but there's also an element of family or whatever you have to it. Just things to think about. People spending time together.

Yeah, I like the idea of dominant traits sticking around. What if I have a player that has this massive crooked nose and I just never die, and eventually this guy's like 100? Traits stick around in the gene pool.

But that is an idea that has been expressed, that orgs could share individually within there.

It's about having different colored eyes. As far as the asset side of it, I don't see a reason we couldn't support it. I don't think that we do, and I haven't heard of any plans to do so. How do you feel about this, Sean?

I love the word. Again, I'm just like, different eye colors.

It's a pretty rare condition in the world.

It's way more common in animals like dogs and stuff. We already support asymmetry, so it would just be like locking one eye and then...

Yeah, I mean, actually, the helmet thing was a big part of that asymmetry decision, because we supported manipulating your face directly asymmetrically. And that was problematic, because even if you end up with a non-homunculus character, when the helmet morphs to match your face, the asymmetry in the helmet becomes really apparent and problematic.

That's a really polite way to put it. It looks awful. You have a Slaver helmet that looks like a face, and then you have half of it looking like it was melted.

On the material side, which is what supporting heterochromia would need, I don't know if we have anything coming down the pipe for being able to do any manipulations of material at that scale. That would be very similar to how you would handle alterations to eyebrows, makeup, flesh tones.

So we did build this into the character system from the get-go as an option, because I thought eventually we would want this. And actually what I thought eventually we would want would be holographic upgrades and things like that that you would put on one eye, or a contact lens on one eye or something like that. That's your AR lens or robot eye or something. But within the item ports of the character, it's actually split up into left and right eye. We use one of them right now. That's all. And it has the two eyes in it. So the easiest way to do it is actually just to add another set of eyes, and then we zone cull one or the other. So there's actually four eyes, but we're zone culling the opposite ones. And that totally works right now. We can do it. So I think depending on how much people want it, yeah, sure. It's not a nightmarish add, but it's not planned for this quarter.

I want to be able to make Cable from X-Force.

Right. No, that's exactly it. Things like scarification and having any sort of post-manipulation, things like having augmentations or prosthetics, those would all be handled on the asset side at this point. Same as adding a hair, which is a weird way to do it, I'll admit. If you're adding the extra eye in or you're adding a scar or something, it doesn't initially make sense that you would attach that to a person, but that would probably be how.

And then the color is somehow attached to the person, which is fun, too.

That's the sort of abstraction you have to get your head around in order to manipulate it correctly.

It's going to be the same tech as doing different kinds of eyebrows, right?

Yeah, somewhat. And people have probably noticed this anyway, some of the skin tones for the females have makeup built into them. There was this character art pass where they did a makeup pass, basically, against all the heads, so they added some eyeliner, eyeshadow kind of things. But whether that's a separable thing, it's completely up to character art on that front. It's, again, not a huge nightmare to do. We have a great feature within the skin shader to deal with it, but I'm not sure that it's going to be perfect, because the thing we have in the skin shader is really good for tattoos, for decals, for dirt or mud, for face paint, basically.

It doesn't do the blending nearly as well.

Exactly. And it doesn't adapt the specular. So the biggest thing that happens when you put on makeup is you're changing the light reaction on your face. A good example is you put powder on so that you're not as greasy-looking. We don't really have that control. So if we put makeup in, it would look like face paint and not like makeup, because makeup changes the specular reaction of that skin. So we'd have to support that. That's all. So right now, to have a really good makeup, it has to be pretty much painted into the texture itself and not a layer on top. So we would need something like that. So if we do progress on the tattoo front or the face paint front, we'll have to do something in that way. But right now, we wouldn't just put the makeup in, because it would just look like face paint, and it would look like a bunch of game developers thought, what would makeup look like? And we have Joyce Platten here, who is a beauty specialist, let's say. She has podcasts and everything. So we had her come over a couple of times, and we talked through this quite a few times, and we even showed her the decal stuff. She's like, mm-mm. That's not at all what it should look like. Please don't do this. So, yeah. Face paint, yes. Makeup, not really. It's kind of built into the shader, and I know some of the skin tones already have it within there.

Ah, that's a great, great question. It's really one for Josh on the character art side. I do not know exactly what their quarter plans are; we're doing our quarterly planning over this week anyways. So we'll know next week, I guess, but we won't be on the show. Mostly, they've got to get all those hairs that are in there. For me, if it's my call, and it's not really, I'll talk to Josh about it, but we've got to get the hairstyles that are in there all updated to the new hair shader first. That's number one. Before we add any new content, let's make sure the content we have is at the highest level that it can be, and then let's add some new stuff. But if we just start adding new stuff in the old shader, it's just more work that we have to come back to, so I'd rather not do that.

And of course, we'll almost certainly have an update on that on ATV when the time comes, or back here on another RTV.

Definitely. Once we have the new iteration that we're working on the design for now, figuring out that new flow, and breaking down how much work it is and how much of the quarter I have to steal from these poor guys, we will show it and talk about it.

Yeah, it's the second one [the longer-term soon]. And it's less about whether we want to do it or not. It's more that this whole system is a data management problem. So right now, if you blended a male and female, because of the positional differences of the origin joint or neck joint that we actually align, when you blend a male to a female right now, it's really creepy. The face melts into the female. But what's super creepy is it keeps animating. So it's like a melted talking face. So there's some great images, and anytime somebody comes on a studio tour, ask to see the page of the things we've seen, because there's great images on it, and we are happy to show them under NDA only. So right now they melt. The reason they melt is that the offset is different. Again, the female's about four inches shorter than the male. The origin position is totally different. So we don't do the blend necessarily relative to just that origin, but that's what we could do to do that work. What it would mean, though, is either taking all the data from the female proto rig right now and adding that to the male, and then taking all the male ones and adding that to the proto rig there. But the issue I have with that is what I talked about in ATV, the performance wins that we got from this whole system. And to reiterate, this is massive. All the heads in that one proto file are cheaper than any single one face was before. So we now have all the faces for the same cost that a single face was before. The difference there is incredible. I don't want to make that much worse. I don't want to take away that benefit that we have, because as you add more heads into that, that gets bigger and bigger. And if we take all the female ones and add them to the same place, now that means every character that loads has to load the same pool of memory. But if you're only rendering males, you're only going to have, I think it's something like 60 or 70 megs at this point. If you're only rendering females, it's only about 20, 30 megs, because we've got a couple less of those. But combining them together, you'd always have the same block of memory, which would be easier to manage, but it's going to be more than you necessarily need to use.

You have to make that choice between, are we going to add 20, 30, 40 more female heads to this, or are we going to use that space to put the male heads that already exist in that same space?

That's correct. And that's why I say it's a data management problem, because, for example, I'd really like to get a Squadron proto rig so that only the ones that are Squadron-related are loading in that proto rig. So it's a way of us managing that data in an interesting way, but it does take some work. So, yeah, further out. But again, if there becomes a lot of gameplay relevance to doing it, like if we do the sins of the father thing, or if we want you guys to pick your parents, like the GTA kind of approach, then yeah, that'll become more important. But at this point, it's not really a massive requirement. And I think you guys will get super far just with the hybrids already. I think that's going to make a big, big difference to the way that you guys interface with it.

Yeah, I think as a matter of priority it's more important to get more female heads in.

And where I always get to is just, I want a reason to do it. I don't like doing things for the sake of doing things. I'd rather have a goal or an end goal or a reason to actually do that tech, rather than just doing tech for the sake of tech.

Yeah, the eyebrows are baked in.

Same texture and same problem. Around your eyebrows, you do get a different specular reaction. And we actually have to paint out the gloss around the eyebrows, because it ends up looking really fake when light rolls. Oh man, faces have so much subtle stuff in them that if it's not there, it completely breaks down. But within the eyebrows, we paint out the gloss in that area. So basically, we make it less glossy around the eyebrows. The reason we do that is it looks a lot less fake. They're part of the texture right now. What I've talked to Corey about, it's the same problem as scalps. We have to make sure we remove all the base hair out of some of our texture scans. So when the guys were scanned for textures, they had little beanie caps on with their hair sticking out and stuff. We've got to remove all of that, because now we have a generic scalp that we can apply, and we need a scalp because it has directionality on it for when the specular rolls around. And then you put the hair on top of that, but you can't have other hair underneath, because your hairline then doesn't line up to what the scalp is, and it can look really weird. So basically what I'm saying is we're going to treat eyebrows the same as hair.

True, and it introduces errors every time you do that. There's no clear solution on that right now, but I quite honestly haven't had it be a massive problem yet. And at the point that it does become one, then sure, we'll do something about it. But right now, I just don't see it as a huge issue, especially because, again, we're going to randomize the weights and everything within there. You could eventually get to that state, but I haven't seen it happen.

With regard to character customization, it's a matter of introducing more extreme options, more meaningful choices to make. And with regard to the legacy system, it would be about managing it so that it does introduce elements of error within that blend. You wouldn't ever end up beyond that blend, but because we blend between four different heads, you don't necessarily have just the weight between these two people.

Yeah, so that's one of the things that we're trying to do for the V2, to make those decisions less segregated. We want to make it so that you're able to jump back and forth between all the elements that really impact your choices much, much more easily. So we want to bring the skin selection into the same screen as where you're doing a lot of the head selection stuff. That's some of the stuff that we're exploring right now. And yeah, that'll definitely help that process. Because right now, if you want to see what this feature would look like and you start going towards something, like, oh, the skin tone doesn't really feel right anymore, I want to go towards something lighter or darker or different, then you need to navigate away from the thing you were just doing. It's definitely a UX problem.

Yeah, it's like context switching. You totally lost the perfect framed shot of what you thought that person would look like with a different tone.

I mean, V1 of this character customizer with blending is all about how do we communicate how this novel system works to new players. One of the things that was really important is it is not a sculpting system. So we don't want to use any tools that operate similarly to what you might use in ZBrush or Max or anything: I'm going to take this thing and move it over here. That's not how the data works. It's not something that's going to be able to actually get you towards the result that you're expecting from that. And the moment you try to do something and it reacts incorrectly, you immediately have a break of that mental model and you immediately have a bad experience. So step one, not sculpting, we need to communicate that. And then step two was trying to communicate that when you're blending these things, you have the full breadth of choices at every step. So one of the initial things is, we keep showing you the heads, they come in and out and they do all that stuff. Because we felt that it was important to try to get across that when you go to edit the nose, you don't need to be using the same pool of four heads to manipulate that nose that you used for the brows. You can have a unique head selection on every single one of these features and blend them independently. That is a level of control and minutiae that, the feedback has proven, is a little bit beyond what is helpful. So we've sort of keyed in on the thing that worked, the thing that was nice about the way that we set it up, which is trying to make the coarsest, broadest choices early and refine from there. So each choice is how do I get closer to what I'm looking for, instead of how do I identify which thing I need to change. So it's focusing in more on the next step: I want to look more like this guy. I want to look more like this. I want this skin instead. So bringing that into the same panel, making it so that you don't have to go to a different step, is absolutely part of that process of how do I give you the clearest, broadest choice such that you can refine from there.

So for the V2 stuff, what we've talked about a lot is, currently, with the UX and the way that people end up making their characters, we're not feeling like they're actually getting the result that they necessarily want or expect. And a lot of this is experimentation, but rather than being beholden to what your result is, because I think a lot of people play with it, and they'll get to a result that they're kind of happy with, but it kind of takes you there rather than you taking it there, if that makes any sense. In the V2 version of it, that whole mentality is kind of switching, where we'd rather you have very precise, as Calix talked about, ever-increasing complexity controls within it, but get the result that you want out of it a little more directly.

Yeah, if you go into the system with an idea of what you want, we want you to be able to get it. Currently, that's a bit more difficult because you have to sort of comb through the system for the tools that you need. And that was in service of trying to explain this novel system, but it muddied up those choices. And so we're trying to clarify those choices so that they can be a little bit more useful and available when players want to use them.

This is actually something we talked about. And then I was like, all right, that's great. We made a task and it was at the bottom of like 20.

And then Spence goes, Sean, do you want emote animations, or do you want the heads to blend when you press X so it actually goes away? And it's like, we've got to make a call. And I'm like, yeah, I want them to blend. And actually, we tried to get the different animation at the end in time. I don't think we made it.

Oh, that got in. It is in. It kind of cuts to the new animation.

Oh, really? We've got to maybe update the Mannequin blend time. I didn't know it got in, because we did get some animation at the end. Nobody liked the pose that we had at the end, because it was super slouchy. The female looks very slouchy, and with the angle of it, at that FOV and with that perspective, it makes her head look tiny on the body. It does to the male, too, to be fair. It's just more obvious on the female. So short answer is, yeah, we want to get that, but it'll become a priority call once we get closer to the end of the quarter, I'm sure.

Yeah, we wanted to expand the review step with a lot of those sorts of things: checking out your emotes, checking out different loadouts, checking out different lighting situations.

That's right. And I don't know if we've talked about it much, but the example I keep using, and hopefully people recognize it, is the XCOM photo mode, or XCOM 2. When you were making your soldiers, you had this great little photo mode that you could go into, where you could see different attachments on them, put different backgrounds, things like that. That was kind of the mentality around review. We'd also like for V2, and I'm not sure if we put this in the estimate yet, to have the different loadouts. So we'd like to be able to show you guys at the end of that, similar to how other character creators in MMOs generally work, where they like to show you the really, really awesome armor, like endgame, perfect, you've got everything. So I can imagine just having three loadouts within it would be nice. So basically, no clothes, civilian clothes, and awesome armor. So you can just see how your character is going to look with each one of them. I think it would be a nice thing.

Yeah, and conceptually, right? And Andreas is very keen to do this. And he's pretty confident that he can get there. Keen and confident. That's why I adore him for that, because I have my own reservations about it. I'm not sure. It's a technical and a bit of a visual thing, too. We have about 1,500 character items right now. When we did the facial attachments, there were like 50. Not a big deal to go over the facial attachments and put in the markup for the facial customization and the runtime deformers needed. A little different story on 1,500 to 2,000 assets. And that's what's getting harder and harder, by the way, about doing systems within Star Citizen and for the game: the sheer breadth of content that you sometimes have to update makes a very real impact on the feature. Like, are we going to do it this way? Because, holy, that's a lot of updates that we've got to make across it.

It's easy to balloon from there. You're just talking about character assets for what they're wearing. You're not even talking about what has to change about their animations, what has to change about anything regarding the seats and the ladders.

That's true. So I don't expect we'll do anything with the heights, but absolutely the body shapes, we do want this. It's just there is a technical challenge within it, and there are different ones than the face brought, but I have lots of confidence in Andreas and Ivo and those guys over in Frankfurt. And they are absolutely looking at it, but we can't promise it at all, because we don't have it working yet. But we'd like to. And if we get there, it'll use a very similar system with the runtime deformers. It'll be quite interesting, I think, once we get there.