Repopulating the Earth: Character Production on Horizon Zero Dawn

[Music] the massacre was only the beginning who did this to us who attacked the tribe all we can do is pray to the goddess no praying won't save us we have to find the killers so you will have to leave the sacred land where the trail leads I will go [Music] hey law of vanora what brings you to Meridian there's a threat to us all men

in masks raising up and shoot machines and I've seen their leader from the depths I will raise an army a storm of shadows to take back what's ours what's it doing to that taking control you can see the corruption coming off that looks come a long way since the citable Eloy who are you and interested pudding [Music] all right guys let's get started so you know the drill by now about 15 minutes after you leave this hall you get one of these nice little emails with the feedback and the bit I'm really interested in this is a little white box so I am far too close to my material I

don't know whether this is really clever or really obvious so if you guys could just give me a few notes afterwards we really appreciate it but without any further ado here's or what you're here to see horizon zero dawn so we're that game with the dinosaurs [Music] and not only that more specifically where that game you play as a young Huntress and you fight those robot dinosaurs with bow and arrow and we're setting a beautiful expansive open world which spoilers is actually the ruins of our own civilization reclaimed by Nature this is populated by a series of new

human civilizations humanity is now scattered and struggling to survive and as a lawyer the player character become embroiled in the conflict of this new world and work to solve the mystery of what happened to the old world so these tribes these are what we're going to focus on today the title of my talk is repopulating the earth and it has a dual meaning in the in the fiction the tribes they don't really know anything about our shared history they've had to start from scratch they have their own mythology their own cultures and Agrella we had this parallel challenge we had to go from a blank slate no design no story no tools to this world populated by a great many diverse living peoples so to begin with though let's let's take a look at how that blank slate came to be because gorillas for the longest time are the kills and guys and some people the studio had done literally nothing but 15 years of kill zones so a little bit of creative fatigue was setting in and a small team

of just over about a dozen people they split off and and walked on what would become horizon so this was a pretty small team but like the work they did was absolutely pivotal they came up with this amazing premise that basically just won over everyone who saw it we got fantastic prototypes like this that said a very clear expectation of what the game could be and through all this inspirational work you know everyone our team was like really sold on this potential but at this point it's it's just potential and this tiny team they can't actually make this huge crazy ambitious game all on their own to do this they need a lot more people so Killzone Shadowfall wraps up and then a 200 person team rolls on over and have a video of what that looked like so you know everyone's really excited everyone has loads of ideas too many

ideas everything is up in the air we have this atmosphere of paralyzing uncertainty and that was a point where I drew on the team so you know 10 slides in higher than count and I was characterized director on horizon zero dawn and getting in on the beginning of a crazy ambitious you know creative new IP is many people's dream project including my own especially if that project does end up being successful like horizon was but of course at the beginning when we were starting we had no idea if it would be successful and we didn't even know if we could pull it off and I had to deal with the worst possible one-two combo that uncertainty which I just mentioned but also at least this huge scope of this expansive world full of people so summed up you know we we didn't know what to make but we knew that we had to make a hell of out of it so much like this guy you know once I got over the initial excitement I'm like where the hell do I even begin this the story is that the it's the talk is a story of how a this burger how we manage

there's two contradictory requirements scope and uncertainty and I go do this by breaking down the last three and a half years of my life into ten little stories for you all so this this first story let's take a look at this contradiction we don't know what to make but we know we need to make a lots of it and at this point at the beginning you know the projects is a blank page all we have is a premise and we just have to figure out the rest and this wouldn't have been so bad were not for the fact that everyone else was also sitting in front of a blank page the story isn't written we don't know how many quests are going to be we don't have any idea which characters people are going to need and all the tools you know they're all still geared towards making Killzone games so we these guys they're kind of a making a quest is going to be a long time until we actually have any information to act on but we can't just sit around and wait I wanted to get some characters done but I didn't want to do anything or work so for these early characters that we make to be useful you know what what form do they need to take formless shadeless like water now you put water

into a cup it becomes the cup you put water into a bottle it becomes the bottle you put it in a teapot it becomes the teapot so they have to be like water you know we didn't know exactly what shape hole they had to fill so they had to fill any hole or put another way you know we didn't know exactly how the characters will be used so we need to start with the characters which were the most versatile but normally this isn't always how things are done you see specific characters like the lead cast for example or some of the most important content you need to make and obviously is like some of the more visible content some of the most visible content so it's often what people will start with but you know specific characters like these this is risky if you're doing a doing a design which is geared to communicate one thing one idea what if things change and you know what if things are cut you know there's this wasted time and effort the more important someone is as well to the story you know like these lead characters the more likely it is they're

going to change as taking a roster example so rosters a lose surrogate father he's pivotal to the first act he was part of the story back when your bill was to working out the premise you would think roast you know he'd be a safe bet but not all you see uh Sasori evolved so did Rost he mellows out a lot from this ax-wielding berserk you see on screen and and leaving what we know about roster side you know we don't really know anything about the culture that Ross comes from you know this is kind of important if you're doing a key character like Ross because he's individual he's um he's special and if I don't know what the average person from Ross tribe looks like you know how do I know how to make this guy stand out so instead of starting with individuals like Ross we started with a generic guys generic is a bit of a dirty word I say generic you're probably thinking things looking the same things looking boring like this poor guy that's not really what generic means it merely it's merely the opposite of specific it means versatile so let's look at this from

another angle what characters to the design team need at the beginning of production they don't really need specific characters like Ross because some ideas aren't yet that specific the still figuring things out they need to have flexibility to change their minds so we thought bigger you know instead of designing a person we designed people's you know we started with the cultures the the tribes and we designed these guys as a unit you know we'd map out all the different kinds of character we thought we'd need in each tribe and we call these key designs archetypes you know what you're looking at on the screen these are the archetypes for the kiya tribe and why do it this way well the trick here is that provided the tribe as a whole is versatile individual archetypes can be less versatile if you're thinking about these guys in the unit so that guard you know he's fairly limited in his uses but that's okay providers the rest of the population and kind of cover what the quest team might need but if we don't know anything how do we go about identifying which of these archetypes to design well we we don't actually need

that much to go off we only need a few very basic assumptions like this from the writers in order to get started I mean we need a hell of a lot more to get finished but we don't actually need all of this information to start and with this little bit of information you know we can get stuck in we can start designing those first archetypes once we have some in the show you know we bounce things back and forth with the writers and we just continue that until we have all of these guys and this worked out pretty great because whilst all the quests are their own little stories these stories are taking place inside of an established world and because we got started early we got in on the ground floor with the writers whilst are still doing this well building these archetypes ended up being a part of that world building so when the quest team you know when they needed to start nailing down what kind of characters they needed to use they tended to be these kind of characters you know what a coincidence and for every character with a name we'd built a unique character from those generic paths as generic archetypes many of our key characters

like nil or like Petra they they just kept using this generic path those paths they just kind of fit the character so well that we didn't need to change anything but of course there are many many other characters where that generic content wasn't good enough there are some simple examples you know sometimes the character it needs look injured but more importantly we would have a you know large number of important story characters where kind of generic looks wouldn't fit their characterization I mean let's go back to roast this is how Ross looked like for most of production this guy he know he even snuck into one of the trailers and we kept using generic Ross you know we kept keeping him generic because things are staying in flux for the longest time and we really waited until stuff was only really coming together which was really only a couple of miles before alpha before we started selecting the characters like Ross who needed that bespoke content the most and then we were start doing you know very specific designs who really nailed who these characters you know really were in Ross case you know this is about late making

him look a bit more paternal than we originally envisioned but it was also about communicating his weird relationship with the rest of the tribe you see Ross he is a very kind of conservative very traditional Norah but he's also an outcast from the tribe so his design is to strike that difficult balance between you know keeping North traditions on one hand but also separating him from the rest of the tribe and if we didn't wait until we had all of that information you know his design would not have been as effective his characterization wouldn't have been as strong and you know the S idea of who ross was wouldn't of been as clearly communicated to the player and of course as a bonus because these bespoke characters because they're the last characters we made they are also the best characters we made because we were able to take and apply everything that we've learned from over several hundred so that's how I steered the art design towards first star characters that's not the only way the characters needed to be versatile because of course when you're done with a character when they're all checked in you lose total control over how they used so stop me if you've heard

this one before you've been asked for an acid out of blue but don't worry guys it's a simple requestors small he's only seen far far far in the background he can be super simple don't spend too long on him and then one movement morning you boot up the game you see this and this can be frustrating and hilarious but look at it from the designers point of view he's probably just addressing some kind of critical feedback or maybe this wasn't even the designer who made the request may be some other designer these are character like this they found this guy in the database and just figured he's a placeholder you know these guys they shouldn't have to work out exactly which characters they use in which context is not fair to impose this on them so once a character was available they should be free to use it however they see fit so this this didn't really happen authorized him because we didn't build the art at different quality grades every character had to survive close up scrutiny even this generic dude whose name I don't remember

so every character that we made was off the same high visual and technical fidelity so this guy you know his design is not as perfect to his characterization as Ross was but I can zoom in on this guy just as close as I could do him Ross and also you guys could do the same with further mode so another anecdote you get a request for a cinematic asset so it's gonna be closer to the camera you make it super detailed and you know super expensive and then one day you boot up the the game and someone's fill the city full of them and the frame rate is now a single digits but again not really an issue on horizon because every character has a very very similar performance footprint you know every character gets very cheap as soon as you're not in their face because we're extremely aggressive and extremely consistent with our approach to LEDs so our crowd NPCs who walk around and those worn off individuals like the guy you just saw you know they're using the exact same pool of models but the game is clever enough to use some contextually a crowd NPC you know the

game knows only to pull in those lower Lots where it's the same character in a conversation the game knows how to pull in the full set by investing all this time upfront making the characters technically more robust the team could use them however they wanted and that meant that we weren't constantly spending time maintaining them which gave us free time at the end of the project to do more bespoke characters like Ross so how do you get started when you don't know anything well bespoke content is extremely important risky so start with more versatile content work up to more specific things later the the more versatile your content is the more likely it is to end up in the final product and though it takes extra time upfront making content versatile from a technical perspective increases its usefulness and makes life easy for dependent teams so that's the that's the first and the the longest story the other ones are shorter I promise you that's how we tackle the

uncertainty but what about the scope you see you know we've got the player character which has a lot of outfits and we have Story characters and we have quest characters we have enemy characters and we have all the background characters and on a game like Killzone you know about this much work to do and on horizon is to be more like this you know it's a whole lot more stuff to make but that's not the end of it you see we're a first party studio we're there to show off what the hardware can really do so it's not enough just to kind of make all these guys they need to look not only better than anything that we've ever done before they need to set the bar for open-world games so too much stuff not enough time so has anyone here ever worked on a project where they felt they had more time when they needed anyone you know you're looking dude so how do we get more time we didn't you know we just made sure that we squeeze as much as we possible out of what we did have and I'm gonna explain this with some fun graphs so this this is an abstraction of

production capacity from the beginning of a project until the end and the vertical axis you know that's capacity that's how many artists inside and outside the studio which I can keep busy making characters and you see you know there's a ramp up whilst we're figuring stuff out in pre-pro and there's a ramp down as we put the brakes on new content and transition to bug fixing and this is important because this shaded area in the middle this is literally how many characters are we gonna be able to make if this red space is how many characters were making the negative space is how many we aren't but could be and our potential capacity is quite high I can get hold of quite loved artists if I like so this adds up very very quickly and our open-world like well game like horizon the scope is something which has value to the player experience in a very direct way you know not just in the sense of his more has more different things for you to look at if we can do more characters we can do more distinct cultures we can do more

variation within and without we can make each culture and the world really feel alive if we can do more characters we can do more bespoke content we can do more of those important one-off moments and and story characters like Ross so like my plan was I'm gonna claim as much of this white space as possible but I mean that sounds pretty obvious right but it's something I had to constantly sell you see remember our old friend uncertainty uncertainty creates this tendency to be cautious and it's very easy to start just eroding this volume with lots of small very reasonable decisions so I hypothetically you know let's say there's this cool new procedural texturing tool so let's experiment let's get the work for walk down before we commit to full production and you know that's not the end of the world but now the tools guys they want to X refactor the export pipeline so you know let's wait for things to settle before we get stuck in and you know it's early days writing and design teams I don't really know what they need let's wait for things to sell a bit you know that's wait for some asset request to come in and you see all of these things

they hand up very quickly you know so how to solve this well I didn't really solve it I just had to learn to tolerate that uncomfortable feeling that came with the uncertainty and just commit to things regardless we had to just get stuck in now I'm not dismissing those examples you know we did research some procedural technology we were constantly paying the other teams to get that updated designer story info but we didn't need to wait until we had this information before we got stuck in a scaled-up but you might be thinking you know it's not this risky you know I couldn't you waste time by making things that you're not gonna end up using well not really because there's an opportunity cost at play you know this time is wasted by not being utilized if we do make something in that time and there is a good chance we're being used that on the whole we're gonna have a lot more characters and of course what is the best way to increase the odds of that content being utilized like Bruce said by making them like water you know by making them as versatile as possible so on horizon we had zero wasted

characters probably should have done upon there and everyone we made they were used in the game not always how we might have originally thought but always in a way that can added something to the sort of player experience so that's how we could go from this to to this you know we made smart choices about what to make it in what order which got us starting faster and earlier and we did more work upfront making the content more technically robust which freezes up at the end to do extra stuff so how can you find the time to do everything well playing that white space you know make sure you're using all the time you have available to its fullest potential learn to tolerate that feeling of uncertainty and ignore it if it's just you being irrational and don't wait for new information don't wait for your technology use what you have now and adapt later so you know that's how we were able to finish everything but if you're if you're serving other teams like we are you know if you're providing

other teams of content it's not enough to just get everything finished before you ship because the rest of the team is still working in order to do their work they need characters I need characters now and we need to get them what they need when they needed it or even before and this is pretty difficult for our content creation teams because most there's other teams that never use in some mode of agile development you know they're going to be doing continuous short iterative cycles punctuated by play tests and this is pretty incompatible with the ideal flow for our content creation were you're doing these long slow methodical processes which are not really fitting nicely into these agile development loops so you know any artists here have been pressured to cut corners and compromise quality to squeeze a model into a sprint anyone yeah and any producers here have been frustrated by artists giving you multi month estimates refusing to budge so what do you do well if you zoom in on those blocks you know you get a little

bit closer to them you actually find that there are a mix of tasks and some are agile so anything design related for instance you know you're you're testing stuff out as quickly as possible you're seeing if it works and you're more or less you know just rinse repeat that process whereas things like game restoration these things are really waterfall you know the work is not very subjective it takes a predictable amount of time and the goals and outcomes pretty clear you know you can you can pretty much do this in one go and fortunately those fast add your friendly bits they're mostly the beginning and the slower methodical bits the most at the end so what we did is we worked in those early subjective bits into the sprints everyone else was running which kept them fed and that meant we could take our time on the rest so let's let's look at this in practice you know making a high poly model is about solving a couple of very different problems so we'd work to get the shape and the form and the volume right you know essentially you know nailing the essence of the concept design in 3d and this process you know it's very subjective in

these loss or iterations and is suited to agile processes whereas a high poly Austin his material expression needs detail these construction these since they are over a time consuming they're all very careful these things are not suited to agile making a design change at this point is some its destructive its time-consuming so what we did is we broke the task into we speed scope to solve the design we handle the rest of the separate pass and the great thing about this was it got the character to the rest of the team extremely quick because we very very quickly converted into a proxy model one get into the game now these morals they they look pretty crude but they actually clear all of our dependencies it's visually representative of those characters for design and play testing so these characters are recognizable they're distinct even if they look a little bit play though and tech animation you know they can do rigging they can do simulation animators are getting a good representation of volume form and weight but the most important thing is that because we made the poly

counts we made the shaders and textures all these things representative of the final content the fire at this proxy model and the final model performance wise it's essentially unlike for like replacement so tech and our more relaxed you know there's not going to be this performance apocalypse at the end of production when all this content gets integrated all the end of once so that meant they were willing content be integrated much later than they would have otherwise and in which high volumes and that's how he got you know even more stuff than a vendor production so this approaches brewed because it gets us off the critical path really quickly when other teams have a proxy which fills their every need this removes a pressure on us to to deliver that final content lead times an hour less of an issue so we get more flexibility with deadlines we needed some finished characters to meet internal milestones or two for demos or for trailers but other than that we pretty much just had to get the rest of the cast on the disk before they started printing them and that meant that we

could give every single character every single one of them the time it needed to be excellent so how can you work a jar without compromising quality well the trick is to work out which part of the workflow agile and which bits aren't and only apply agile methodologies where it makes sense Muse proxy content to meet the needs of dependent dependent departments as quickly as possible because if you can quickly meet the needs of other people's sorry if you can quickly meet the needs of our team you get all the time that you need to make every single asset good so proxies you know they're they're a great way to handle this content requests that we were getting a lot of requests most of these quests you know they have at least a half dozen characters pretty much all of them come with some kind of special requirement now that's just quest you know got enemies we got player outfits all kinds of other stuff so you know that's where that's where all these guys come from as I'm sure you've experienced yourselves you know everyone always needs everything at once and you can't do it all at once so you have to

work out some kind of order and the most obvious ways prioritize is by how important things are this is pretty easy to quantify you know it's it's critical path stuff for things which are central to the experience you know on horizon that's you know core mechanics player character main narrative but the other thing that we had to account for is our old friend uncertainty so to avoid wasting time we started with the things which were most important but also the most certain and let's look at some of these requests that we got for Eloy we got a lot of requests so this is every feature addition gizmo that our asked at AE and all of her costumes at one point in development and clearly not all of these made it and thankfully we avoided spending a lot of time Lee sings by prioritizing the most certain but how did we know which bits were more certain or less certain well with experiments you do kind of get 6 cents for like how what is gonna work but you have to be really careful about relying on you

intuition because prototypes and play tests they have this tendency to constantly surprise you I mean that's kind of the whole point of them so my approach was that for any asset didn't really matter what we'd immediately offered to create some kind of proxy content and it's normally that to one of the following outcomes you know often the feature requests it just worked as intended awesome but sometimes you know the requests of owls reate waiters as we need to we wait for things to settle and then we make the final content sometimes a proxy though it just goes unused maybe the feature got cut maybe another team didn't have time to work on it but it doesn't really matter because we didn't actually spend that much time on it and sometimes rarely the requests he just doesn't take us up on the offer to make a proxy and I found this is generally a good sign that feature is never going to see the light of day so those things we do prioritize and I found that informal techniques like this are much more reliable than just blindly following production documentation so this is an experience I'm sure most you can share I've learned to be a little bit skeptical of these tracking systems you know they're often inaccurate or out-of-date the requests normally lack

context or missing some kind of critical information and this is this is my favorite example okay so those kills unlevel and it's centered around a hundred foot statue okay and your objective is to infiltrate the complex inside the hundred foot statue and at the end of the level is a big set piece you know where you blow up that under the statue and not a single producer identified that we needed to build the 100 foot statue it was it was just so obvious that everyone had just assumed that someone else attracted somewhere else and as hilarious as that mistake was it's perfectly reasonable like trying to log every little bit of work on a project like horizon which has a million moving parts and like hundreds of independent actors doing the wrong thing it's insane you know it's unreasonable to expect someone to do it perfectly so I gave them a hand by being proactive in so whilst I did pay attention to those tracking systems I didn't ignore them but I found it far more important to pay attention to the ground truth which was to play the game every single day you

know so I put my trust and what I could see in the game I trusted my own judgment on how the existing content was going to work out and doing this you know I was able to identify work that we were going to need to do and start it and sometimes finish it long before anyone even noticed there was an issue and I found the best way to do this the best way to approach this is you have to shift your perspective you have to stop looking at the game as a developer and try and experience it as a player you know it's important to taste your own dog food so I caught literally hundreds of things this way and if I hadn't as since we're just not being in the game so how can you prioritize well start with the things which are most critical and also the most likely use project planning tools but with some healthy skepticism and play the game every single day track issues and tasks yourself and solve issues before people even become aware of them all right so we're at the end of point four I've mostly been droning on about um

production stuff I recognize a couple of you in the room they're probably thinking a bit feel a bit misled you know and this is on the visual arts track where you're talking about production well my philosophy is that process and outcome they're part the same thing you know they kind of exist on the same continuum and if you want to be the person who is directing the outcome you can't wholly delegate process and on game development that process does involve a lot of strategy and planning you know collaboration team building so in my experience the most effective approach to creative leadership is a holistic one which does incorporate these things like planning and production alongside creative direction but enough enough production for now let's let's let's do some creative direction everything that we've talked about so far it's been pretty even methodical you know and then keeping we have a pretty methodical approach to our visual design our design is essentially an exercise in communication whatever you're creating must effectively

communicate your intent but as we have hundreds of people working together how do we make sure these guys they're all seen to the same hymn sheet well we have this focusing device which we call the four pillars and these are the key themes they're our main ideas around which the premise is built you know this these these first four videos I showed you the this beautiful post post-apocalyptic world you know these are animistic machines the diverse tribes and of course a boy so anything that anyone asks the game it needs to enhance these pillars and not detract from them and this is how we work out if something is horizon II as of themselves so these are not things I can directly apply to the characters so that's that's examine them and deconstruct them and see if we can translate them into something that's a bit more applicable the whole point of that beautiful natural post post-apocalyptic world is that it's vibrant life persevere is it's

optimistic so to reflect that pillar we want characters who are equally dire who are equally vibrant diversity is something that's also fairly self-explanatory an architectural example in horizon is of course the machines who are this new and diverse form of life a horizon is it as a game celebrates the difference and the unexpected so obviously we want tribes to be very different from one another in ways which are complex and unexpected contrast between the olden on you this one is super important in my opinion this is the key to making things feel horizon II so these kind of animalistic machines that kind of contrast is the arc typical example of this but it's a critical theme that it needs to come across through the characters as well and finally David and Goliath the sense of overcoming impossible waltz you know we need to show those tribes and daily as underdogs in order to really sell the power of those machines so having these ideas outlined it gives us some sort of

metric for what is right you know so if I want to add detail X on a character and someone else they they disagree we can discuss this objectively over them what is you know just what our personal tastes are you know this isn't foolproof but we were able to skip a lot of those familiar circular arguments so let's run through this methodical design process for a tribe's visual identity now this is going to be the Cliff Notes version of how we approach the Neue tribe so let's let's start at the very beginning let's look at some of those you know those assumptions we get from the writers like I showed you earlier what do we know about the the North tribe that started pre-production well their core identity is that these guys are going to be survivalist hunter-gatherers and they're the first tribe the player encounters okay so in order to introduce this interesting contrast between the advanced machines underdog humans early we're gonna make these guys relatively

low-tech and because these guys the first tribe you know this is where the player is going to be learning with the robot hunting this is where they're going to these are the guys who are going to be doing the tutorial robot hunting so in order to be able to teach the player they need to be expert machine hunters but because they're low tech expert machine hunters we're hitting that david-and-goliath scene and we know where the game starts you know it's the Rocky Mountains this is really useful because from this we can we get climate we get material culture all kinds of other useful assumptions watch to start and our first main point of references is mountain men so these guys are appropriate for their climate they kind of been born yet survivor suspect this is a pretty good sight point we start doing our exploratory sketches in parallel and these guys you know they embody that hunter survivalist role but they're not really right they're not really hitting any of those key themes these guys you know they're too literal to the reference these guys are too sophisticated for a lot of tech culture so let's let's focus in on that material

culture this is a good way to tackle this these guys are they're hunter-gatherers okay so they're not going to do a lot of farming this means that they have little to no textiles and their primary materials for garments they're going to be hides they're going to be pelts they're going to be leathers but we have now a conflict with one of our pillars okay you see in order to have pelts you need to have animals there's this big discussion list you do you know should the world of horizon have live animals because on the one hand you know there's an argument to be made that the having living animals competes with the primacy of the artificial life but on the other hand you know we took feed and clothe the tribes so we reach a compromise animals can stay but so Astor's would not compete with the robots any animal over a certain size is now extinct so we have to make their garments from stitching small eyes together so we've we've made a bit of progress now but something still missing let's design work is now appropriate for the environment but we're still not really

embodying a lot of those key themes identified earlier and there's nothing about these guys as well that kind of makes some particular to the world of horizon you know they're only about a step or two removed from Vikings so we'd had our knows far too Klosters material culture and had kind of forgotten these guys were hunting robots these machines you know they're a threat but they're also an opportunity a low-tech culture is gonna have all kinds of uses for the materials that they can strip off one of these down machines so this opens have a worth of newerth of new opportunities to hit some of those key things you know starting by mixing modern and natural through adding some of these robot materials to the characters so it took a lot of exploration to get this right this this is one of our first attempts and these guys are really not very successful nothing really fit the character there's there's no kind of practicality here so we start thinking more about function you know why these guys using the robot boots and by this point you know the writers have established the the Nora they don't really have any particular reference for the robots so these guys are going to be using them purely

practically as as armor for instance but working this out is tricky okay because if I if I make these robot pieces fit a person - well you know it looks to deliberate it looks like they were designed for that person you've basically designed a mech suit you know you we need that contrast between the old menu but if the way that these things are applied you know it looks - improvised it looks on a mixin or look like they're careless and messy we're not kind of hitting that expert machine hunter note so it took a lot of ideation and eventually we solved the spice you know striking the balance by using intricate configurations of these smaller plates but we're still missing vibrancy because these natural materials you know they're only providing a very very limited palette so we start thinking about maybe how we can use these robot past this effect and we we start by just taking small elements like bronze or gold connectors and usually sings as accents as jewelry and then we also add in these cables especially these really really

colorful anodized ones you know they have this kind of silvery luster and then what we did when to walk these things with similar cables made from natural materials and we get to this juxtaposition of the same shape language but with different materials you know it helps to make the robot elements look less foreign so we get our contrasts we're in a way which isn't abrasive and that's thus the really abbreviated version but that's how he kind of worked at the main motifs for the for the nor of visual design and because we approach this methodically there were a lot fewer dead ends and we knew very quickly if something was working or something was missing and this this methodical approach you know went beyond designing the individual tribes because in order to make the tribes distinct you know before we designed any of them we roughly mapped out well the core identity of each tribe would be and what the main forces shaping them would be and by like planning out things like location climate cultural development technological sophistication all these things in advance we knew that even though we couldn't be sure exactly how

they would end up we knew that they would be distinct so if I subjective how kano what is right well you don't need to know everything but you do need to know what your work is really about ok and I don't mean in a in a mechanical way like what the plot is I mean the overarching themes the the ideas and the in the and the premise work through your designs one step at a time making sure that your conclusions are internally consistent and logical but more important that these things embody the key themes and communicate very clearly to a player so yeah once again that was the the super short version and I kind of focused on the successes to save to the save time which is pretty misleading because I I gave the impression that this was a lot easier than it was I really really care wasn't easy so when I show you that process as a story looks all fine and good because you get a nice beginning middle and end you know but

like being in the middle of that something else entirely it was frustrating and difficult at times I thought I thought it was never gonna end we generated so much material I couldn't possibly show it to you all this is just like a small selection I grabbed we see in reality design isn't about trying to find your way to some kind of end goal because you often don't know what the end goal is until you reach it and even then you know sometimes when you do reach it you're still not sure so our process it's less of a roadmap and it's more of an evolutionary tree normally we have some kind of jumping-off point like those assumptions so we generate some ideas ok and our concept guys they're great so all of these ideas are great but we have to choose what to commit to you know we have to start making decisions again not picking between good or bad ideas we're selecting ideas on how effective they are at encompassing the original intent as well as how complimentary they are with the other decisions that we've made you know and this process just basically continues you know we id8 we make decisions we id8 they make decisions and

at some point you just have to kind of stop when you're feeling it but committing to things it's super important because it creates this positive feedback loop the more decisions you make the easier it is to make even more decisions because you can base some of the decisions you've just made you know you get this snow-blowing phenomenon called the colon cone of uncertainty which is basically that as you make decisions over time the number of possibilities decrease which means that certainty increases and then it's easier to make progress and do useful work and on the beginning of a new IPI Corizon like those possibilities they're they're endless you know we could have done anything so it's very easy to feel lost and hopeless and it felt a lot more comfortable to play it safe and just not commit to anything but I just had to ignore this and start making decisions because it's literally the only way to make progress but those decisions they can be pretty hard because it often feels like you're throwing away great ideas this guy he's a he's a tech priest

from when the guys are just working the premise you know and the story changed a lot and this guy got cut because he no longer fit but his design was so strong and it has so many great ideas that just some of the themes and ideas just kept very emerging you know the the ropes and the colors they influenced are very similar Mesa Priest the power-on logo the stitch my sleeves you know we adapted this for the Eclipse court and this iconography just made a hell of a lot more sense for these guys you know and the cable stitched into his flesh you know we incorporated these into the Bannu Collins and the writing team then you know imbued them with a lot more meaning and they became a lot more important you know I didn't need to worry because if those ideas were really strong they persisted a good idea stays with you you know it gets stuck up here it's impossible to unsee things okay so this is a bunch of random noise yeah okay and now they see a dog okay and now you will see the dog whenever you see this picture you know it's now stuck in your subconscious forever I could show

you this image again of five years and you will still see the dog so how can you get to the end if you don't know the way well don't despair if it feels like you're not making progress because the worst thing you can do is be too careful the best way to crawl through that kind of uncertainty is to start making decisions start committing things and don't stress out about cutting things you know if something is good and it gets cut it will find its place sooner rather than later so making those decisions that's pretty important but you need to be sure that you're making those decisions for the right reasons I'm not making those decisions on my own we're making things for other people to use so there are a lot of stakeholders from lots of different disciplines and this means that we're getting feedback from people who don't always have an art background these guys they don't always have the artistic vocabulary to verbalize their feedback clearly you know they're not always able to do things like separate presentation from design but turning the same word as

artists we're not exactly objective about our own work either everyone has their blind spots everyone can be swayed by a quality finish or taste rather than content so as an example we present lineups like this even though you know these things are just about presenting costume design and getting feedback and costume design people were getting swayed by things like gesture facial expression even even silly things like you know the order in which things were presented you know we talked about this beforehand and people would make a conscious effort not be swayed but it's really hard and practically impossible even you know and the the feedback that we were getting was just less useful as a result so in a misguided attempt to to address this issue I shifted us to lineup so identical poses and models so the only difference was costume this said it didn't really work like I thought it would because before if someone didn't like a pose or something they didn't like that one specific variant but now if someone does like the pose or the

face they don't like any of them hello so those contextual clues like gesture they're irrelevant to the costume design yes but it's not just fluff you know they're communicating vital aspects of the character so now what we do is we go white or presentation to begin with and then we standardized as the design is closing you know as we're getting towards the end and sometimes like this example on screen will we draw the same illustration the same design under different conditions for av-test and then work out exactly what it is people like about avery specific illustration you know you can't defeat bias or brute force it's just something you have to account for and another thing you have to account for is people's expectations so before someone sees your work they will have their own mental image of what they expect to see and if what you show is different from those expectations even if it's appropriate and even if it's good it can trigger an adverse reaction so this is this is the first design for silence so for people who

don't know silence is this kind of morally dubious genius badass he was first described to us by the writers as his kind of technological wizard but also some hunter he has no problems taking down machines so I see this first concept you know it's awesome and it nails Who I think Silas is and the concept team they also think they've melted and won so we're really enthusiastic you know we go show it to the writers and they like it but they don't really see him as silent because it doesn't really meet that wizard description they gave us so but no worries you know we discussed our expectations and we do a bit more exploration and we're not really sure on this wizard thing so we startled you know we show them this they're not really feeling it you know it's not really a striking us at first design and it's not really a strong and that was it know as it should be so we think okay we're being a bit uncompromising let's pull the the wizard slider to max thinking that this is this is what Silas needs to be but it expectedly

they don't like this one any better and every time we talk about these concepts everyone's always doing it in the context of that first concept it was just like it left such a strong impression it essentially just pulled everything else into its orbit and it wasn't so much that it was wrong it was just that it wasn't really what they had in mind and he just took a bit of adjusting too and ultimately what we ended up permits you've been with was something pretty close to that first concept now mind those explorations they weren't wasted time you see we incorporate a lot four features from those ideations into the final but it just meant that we kind of got there and a fairly roundabout way so now what I do is I try to make sure that every design has a bit of breathing room by making sure that people to see designs long before they're asked to cast judgment on them you know I try and do this out of context so it doesn't feel like I'm slitting their opinion that very second and this is like silly simple things like you know print them out big and stick around the office slip them into random emails and meetings and this sounds a little bit manipulative but

it's not really because if someone really really doesn't like a design a bit of time isn't gonna help you know if you're if you're doing something original and and with something which you know sometimes a little bit weird like horizon is in order to give a design a chance to be properly appraised you know appraised objectively it needs a bit of a climatization time and I found that if you do this the more original the more distinctive designs they're the ones that consistently win out so how can you keep that feedback objective well be aware that your design work is just holistically not just that one thing you want to focus on be aware that everyone brings their expectations and preconceptions so give your work room to breathe you know in my experience some more radical the more unexpected the design the more acclimatization people need to say it objectively so we're gonna stick with this theme of psychological blind spots so here is a story about time that I

found it's that common failing getting excited about a process whilst losing sight of the end result you know for a way I'm from we call this being too clever by half so now we have we have a whole bunch of distinctive diverse designs you're the concepts ready to go now the problem is like how do we make all the stuff cost-effectively you know how do we make these things these characters as efficiently as possible and modular construction there's a fairly obvious way to go so you know we start with this Lego approach you know we build huge live is a small pieces you know discrete chunks of the character which are individually versatile and then we just don't we combine these to make many different variants and on paper that's that's really promising but when you transition to 3d the reality is pretty different because our designs they use a lot of layering and a lot of overlap this makes it really difficult for these pieces to work together without intersections or without reading complications so in order to make these things work together we were constantly making design compromises in order to make things you know physically fit and

a good design it can consolidate you know a little bit of compromise but the difficulty with these modular systems is that every time you add a new piece the number of these inter dependencies increase exponentially so each of these lines that represents dependency between these pieces and the more of these dependencies you have the more compromises you are going to need to make and this looks like a lot but this is really just like three sets of pieces this complexity it scales exponentially you know here's temps as a pieces this only gets worse and worse okay so now you know I can only think of two ways to make this work option one is we do like hard rules on piece design which settle on standardized very very similar forms very very similar volumes but this kind of defeats all point because everything is just going to end up looking the same option two is like we try and defeat this with cleverness you know we we kind of create some tool some kind of procedural fitting system but the trick is like these things they

come with a lot of management overhead and require a lot of time to use and maintain and this is time I could be spending making more characters and really the most of the assistants can do is just kind of clumsily warp and fit pieces together and that's not really a compromise I was willing to make because I would lose all of these bulges compression folds you know these are important descriptive details that really make these materials feel unreal so to cut a long story short the problem wasn't a modular construction problem it was a dam problem I was being too clever by half and I've gotten so enamored with the potential of this modular process but I'd essentially just lost sight of the outcome which was simply to make as much variation as efficiently as possible so in reality that's dead simple okay cuz not every part of a design as value equally so now what we do is we just look at the set of archetype or designs like this we identify the artists which are most distinct at the most visual variation from from design to design and then

conversely when identify the bits which analyse variation which are most similar for characters character and then we break them up when we make loss on that first category and then we make fewer or one of the of a second so we have bigger pieces they have less interested interdependencies and this worked out a lot better you know so we can't make quite as many different variants with a set as I could in the other example but the variants that I can make are substantially more different looking for one another but even more importantly because there are no inter dependencies across these archetypal characters I can make every archetype look very different from one another so how can you avoid being too clever well don't lose sight of your objective focus on what you're trying to achieve don't get by seduced by cleverness for a songs like make sure it's the problem that's leading the process not the solution so these these techniques they helped us make more content but even then the scope of the game is still hilarious the overwhelming and gorilla

is not exactly a small studio you know horizon was about 200 people but we I think we like to pretend and we we would try and keep a small studio mindset we like having things like direct lines of communication we like having an open environment where you know everyone can be criticized but with every single extra person you add to a team collaboration becomes harder it becomes more formal so what we did with the horizon character team was keep it deliberately small you know we only had four full-time artists and two contractors so this of course means that we relied a lot on outsourcing and I could talk all day about how to outsource effectively but I have three minutes so I'm going to focus on you know perhaps the most important thing outsource effectively you need to outsource the right things and the right things are not always the most obvious things they're not always the things that you might want to outsource but if you want to make a lot of stuff with a small team need to be constantly asking yourself on every task you know can we outsource this and the trick to untangling this question is to consider

where it is that your studio really is adding the value okay because those are things you shouldn't be outsourcing and they're guerilla games you know we're we're game developer we're not a creative commune we're not an art school we're not a model Factory we're making video games and the meat of the game is the player experience where the storytelling the gameplay the technology the visuals where all of this stuff comes together you know this is this is what people remember this is what gives the game its identity this is what we need to own this is where we really are adding the value and everything else is essentially details on an execution so game meshes can we outsource this well good doing a good gamer sure it requires skill it requires practice and talent okay but the outcome is relatively objective and fairly easy to communicate you know the it needs to look good it needs to meet the performance requirements this means it's super easy to train it's easy to brief it's easy to review but they're super time-consuming to do well you know so so what value are

you adding by like tying up your internal stuff making these game wishes let's look at this from another angle here's a workflow from the perspective of the person making the game wish to see someone is handing them a high poly and then in the middle they do their thing and they hand the stuff off to like whoever's doing the UVs okay so this work is just flowing in one direction you know there are clear end points clear success criteria for every single handover this is not exactly a collaborative process these guys need to cooperate need to make sure that [ __ ] is getting done taking special notes but their interactions aren't really adding a lot to the outcome you know this is the difference between cooperation and collaboration collaboration is where those interactions make a team greater than some of its parts so there's no real reason to have these three guys sit next to each other because they can do their job more or less secretly the same as if they were hundreds of miles apart all right let's compare this to an example of something we diplomats to do you know a loads out and those hair can we kill me outsource us though because

this is not just an asset we can shuffle around as an R&D project it spans multiple disciplines and everyone is adding something different to the process you either the flow is not one-dimensional and predictable it's back and forth it has dead ends it has rika moments and a project like this benefits from people working closely together you know people who know each other well enough they can have a passionate argument about something without things getting too personal so a loss hairs is important you know anything a loan related is important because of course you know she's one of the pillars of the game but as important as she was it didn't mean that we had to do every single facet of her inside the studio and I talked about a speed model process before so inside the studio we spend a lot of time working at a line in 3d we get him in the game we get feedback we see how she feels to play and after an extremely long and collaborative process we feel like we've resolved the design the proportions of good the design elements are working you know she's nice and readable we're ready to do this kind of final high poly add in all that detailed material expression

you know everything she needs to feel believable can we outsource us yes and at this point we hand it over to one of our skilled friends because whilst this is extremely important things like material expression they're not very subjective you know they're not very collaborative detailing is high poly is a super skilled and is definitely fun and it's definitely something our artists are perfectly capable of doing but it's not necessarily the most impactful part of the production and by not doing this it frees them up to do some other more important things which I'll talk about in a second but just to kind of wrap you know what did we outsource in total quite a lot so using man-hours spent on the asset as a metric for almost 90% of our content one of our partners did the a majority of the actual a secretion you know at least spent more time in it than we did now that doesn't mean that they did all of it because we still have our fingers in every single character we're still keeping ownership of those identities of finding parts but it did mean that they spent more time on it than we did but

for 99 percent of all characters at some point in their life sigh in their lifecycle we had some kind of external development assistance you know so even if a character is owned internally like one of our leave cast we still get a partner to do at least some of the more mechanical less collaborative bits like movies or baking or lots and and just a reminder you know I said the internal team only did about 13% which doesn't really sound like a lot but it's 13% of our enormous scope so that's a pretty impressive chunk of work and it was pretty awesome work so I'm just gonna take a second shot off internally we made pretty much the same amount of characters as a kill zone game so outsourcing more didn't mean that we did less it just meant that we outsource more you know and through outsourcing more we gave the game the scope of content it needed to be a success so how can you decide what to outsource we'll make a habit of asking yourself if something can be outsourced because the answer is yes boss do you think the more collaborative the more subjective something is the more difficult is now source so work out where your team really needs to add the value and hold

into those things in particular these things they tend to be less content creation and more problem solving see our internal team they focus more on that problem solving collaborative aspect and then delegated a larger part of the actual content creation to our partners this leads into what I look for in a team member you see in my experience you can categorize game artists into one or two boxes you have the the artists who are focused on making great pieces of work you know the craftsmen and there are artists who love figuring stuff out pushing the boundaries you know the tinkers and the people in that first category you know they have they have ownership needs which are pretty focused they put that all into that one thing they're doing now that thing is theirs you know and we and we have a need for people with this focus but it's not always a very collaborative approach so I found these guys waffle not can work just as well remotely when it comes to the team go inside the suit inside the studio I want them had scientists okay these guys are the ones who at the

cutting edge of what the tech and what the tools can do these guys are the first people get out their chair and collaborate and these guys are the ones who are just like trying out all kinds of crazy [ __ ] on their own initiative so when it comes to ownership you know these guys say they think bigger yeah their own big broad problems like human anatomy and deformation or digital tailoring you know more muscat and more masochistic artists I really like tackling the problem of hair and the the character team they have an abundance of these ongoing projects most of our artists are owned at least a couple and their task goes beyond you know just kind of tinkering and researching and developing you know these guys are supposed to set the workflow they're supposed to teach others they're supposed to maintain the standard throughout production and that basically means that for every position I'm looking for someone who has an aptitude for leadership you know either an artist who has an existing experience who wants to keep at it but also people who don't have that experience but exhibit the right mindset and qualities these guys they'll each

coordinate a wider pool of artists so in addition to have excellent visual and verbal communication skills they need good emotional intelligence but most importantly these guys these an earnest desire to share their knowledge and to teach other people but and the problem is like not every excellent artist is or wants to be an excellent leader and if you put someone in a position where their job of all's like that and they don't want to do that that rarely works out so how can you handle ownership well be aware that people have different needs concerning ownership and high people whose needs are aligned with your own with with what you want to achieve ownership have broad problems to the right person can be as satisfying or even more satisfying than just owning individual content and if someone does own one of those broad problems they need to be equipped with leadership skills to educate inspire and delegate you know hire people not only based on their skill look for people whose goals align with your own and on that note the obligatory hiring

seek and with that out of the way I'm just gonna welcome back to this happy guy I didn't finish the burger with some grand strategic master plan I didn't calculate the optimal way to shove it down my throat at once and I didn't design some kind of magical burger eating tool instead I just kind of got stuck in and took it one bite at a time because more often than not those smaller day-to-day decisions those are the most important you know individually they don't amount to much but their combined effect is powerful how do you make good small decisions well ironically you have to think bigger you have to check every little call you make against what your ultimate ly trying to achieve every teeny tiny decision needs to be measured against what effect it's going to have on the players experience and with that I'm afraid I have run out of time they're about to kick me off stage anyone who answered ask me some questions or argue with me I like arguing you can meet up with me in this wrap-up room overlook 22 same building right there in the corner at 7:30 you see I foolishly took an appointment

right after this talk so I have one last lesson always check your calendar don't agree to anything without checking you can do it first and it takes only one really dumb mistake to undermine an entire presentation so thanks