Fallout 4's Modular Level Design

all right everybody we're going to get started here uh the usual first few words go ahead and move in towards the center if you haven't already uh turn off any noise making devices cell phones Etc confetti cannons just leave them under your seat and remember to fill out the surveys at the end that's a a huge help to us lets us know sort of you know how we did and it's most honest and Anonymous feedback we can get um all right so Off to the Races so as you can see this talk is about the module level design of Fallout 4 you're the right place hopefully and before we really get going a word of caution this talk has a lot of ground to cover we're going to go fast uh we are going to make slides available you can find those online after the fact so don't worry too much about getting everything down on the screen just try to keep up with us and we'll get through this together okay so modular kits modular level design when most people think of them the first thing that comes to mind is a Lego break and the comparison makes sense Legos are small objects we work together as part of a system to create

something original which is maybe exceeds the bounds of the original imagination or intentions of the creator of the brick itself in a game like Fallout 4 we might think of a structure like this something made from a series of interconnecting pieces working together as part of a system to make something original that the game developer or the player created outside of the original intent of the person who made the kit so what you might be expecting today is a module of kits 101 sort of a primer and that is and isn't what we're doing here today this is actually the second time that and I have talked to GDC about this specific topic we originally here in 2013 it's part of the level design in a day workshop um and today we're going to recap some of that talk but we're also going to focus primarily on new ideas and New Concepts and lessons that we've learned since we were here last a couple of brief words about who we are uh my friend here is Nate perky pile and he's an environment artist Bethesda game studios my name is Joel I'm a level designer Nate and I have known each other for about 12 years worked together on and off throughout that time and between the two of us we've worked on several games and of the

games that we've worked on together all but one of them made extensive use of a modular system for its level design environment art which is just to say that you know we've been to this Rodeo we've had several projects in many years to try out different approaches find things that work screw other things up and just sort of learn that's not to say that everything up here today comes from Nate and I ourselves you know we owe something to our colleagues in the industry who have also shared their thoughts whether it's Lee Perry darus kazimi drik U who also interested in you know systemic level design um but also the entire team at Bethesda game studios and not just the other level designers and the other environment artists who use this process most directly but the studio is a whole because with thda game studos is a place that has you know created an environment to cultivate this kind of workflow and allow to grow over uh over the space of time that we've been working there and before and I think it's important to have this kind of context in mind in this and really any GDC talk because while it's on its face about a specific process or spefic specific technique

that you can hopefully use and apply to your own work it doesn't exist in a vacuum it comes from who we are as a studio what our priorities are our culture as well as the priorities and the goals of the games that we hope to create with it so today we're going to Endeavor not to just give you the how of some specific techniques but also as much as possible the why so you can understand our thinking how we came to these processes and best understand how to apply them to your own situation so modular level design is something that's been around in games for a long time it Bethesda all the way back at dagger modular kits were used to build procedurally generated levels maroin had the sort of standard kit approach where everything snaps together on a grid Joel and I worked together on blood rain 2 and that was that same kind of kid approach and it was really obvious to the player that this is a kid everything was like north south east west Oblivion had the same kind of kits but there was a lot more custom pieces for Destruction and things like that so it worked a bit better but you could still tell it was a kit in Fallout 3 we had standard kits for most spaces but here we started to try a of new things

like directional kits that only snap together in a certain way asymmetrical hallways Skyrim had standard kits as well but we tried even more new things like kits that only went in a single Direction so you could have really specific architecture or the cave kit where there was all these separate pillars and walls and things that could be mixed together by the level designer to really Define the space and it wasn't just dictated by the artist but today we're going to cover Fallout 4 and how we allowed the level designers to have even more flexibility and a lot more visual variety so yeah as uh as promised you know this sort of the 101 section of the talk uh we're going to go over some fundamentals and again I'm going to go kind of quickly through them so when we start thinking about creating a kit the first thing we want to decide and understand about the kit is its footprint footprint is in essence the grid the size of the pieces and the si uh the shape on which they're going to snap and tile with each other this is where the comparison to a Lego brick probably makes the most sense the design of a Lego brick is highly specific and that specific it is what permits it to

work together as part of a system it also has an effect on what you can build with a kit and what the look and feel of the final construction is going to be like so a typical footprint might be an equilateral cube of volume and this is a nice footprint to have because it's going to tile with itself cleanly in all directions and it's going to give you a certain type of aesthetic a certain proportion of space like this hotel hallway we see here it's worth playing around with the different types of footprints you can have for example you could double the height of that same basic footprint and give yourself the same flexibility and tiling on equilateral horizontal plane but giving yourself a different ability to create a look and feel with additional headro you can play around with other axes as well uh such as making a wide kit kind of like the one Nate was showing from Skyrim a minute ago when you're getting into these situations you want to be careful though about how a footprint can lock you in and do things like limit your ability to tile and snap back on itself it's a point that we made in 2013 so with these like wide Footprints we often call those a one directional flow kit cuz it flows in One Direction it can't tile back on itself

it's okay to make these kinds of decisions but you should try and think about the ramifications that it's going to have and decide how you're going to cope with the limitations uh and what you can live with something else really critical to understand something I think a lot of people misunderstand where they start in this workflow is you need to treat the footprint as a maximum extense of the piece not sort of as a starting point so this is another example from our prior talk now if you look we have two pieces identical pieces of a hallway kit pushed right up next to each other now if you focus on the floor you'll see that the artist has built all the way out to the footprint on the tiling Edge which you have to do that's how the pieces are going to meet and tile make the kit work but the artist also built all the way out to the edge of the footprint on the floor on the other non- tiling Edge where the walls are what this means that the walls of that kit are going to be co- planer they're not going to include they're going to have Zite flickering problems you can't do that even worse there's a little Al Cove detail built into the kit even at this gray box level and those alcoves are now going to intersect and protrude into the neighboring pieces meaning you can't use a kit in this very basic way if instead we had use the um footprint as a maximum

extense and left a buffer we could use it just like this no problem at all pivot Points are also very important um pivot Points judge how you're going to work with the kit how it's going to snap mathematically uh and they have to be really nailed down early on so if we take example of a couple of Kit pieces from Fallout 4 uh here we have two the one in the middle of the screen is a basic sort of floor ceiling piece and it's uh pivot point is at the bottom center which is the most common place to have a pivot point but the highlighted blue piece that little clamshell on the side it has a pivot point at the bottom Edge both cases are going to snap mathematically fine based on the grid the footprint that these ones are using but it does affect how you work with this piece as long as it's mathematically accurate it is to a certain extent a matter of personal preference but once you've nailed down the uh pivot point that you have on a certain piece you want to make sure that you don't change it because the minute that you change a pivot point you're going to undermine any work that level designers have already done with those pieces something else that we haven't always

been great about but we tried to improve in Fallout 4 was standardizing our transitions the best example of this would be a simple doorway early on in the game we chose a single wide and a double wide doorf frame size we made sure those Dimensions were well own to the team and we used those consistently throughout all the different kits so every kit didn't have to solve doorways its own way this gave us the ability to use door art interchangeably but also more critically it allows us to mix and match as level designers between different kits so if we want to go from a basement to a brick to a hotel we can do this fluidly without having to get custom art to handle every single case it's also useful for our character artists and animators because it gives them a known quantity in terms of the smallest choke point that our characters have to be able to navigate looking at some sort of alternative and sort of you know beyond the basics uh techniques one thing that we've talked about before are layered inserts we use with a base kit best example of this remains from Skyrim if we take a look at a very simple cave layout that's made with a traditional grid system it doesn't look like a cave at all we have this unnatural grid

Corner in the edge of it so what we do is we bring in these layered insert pieces that go floor to ceiling and what these do is allow us to Shear off the unnatural corner by themselves these pieces have bad texture seams and gaps but we can easily cover those up with Boulders pillars free floating pieces and even after like the most rudimentary lighting paths we have something which begins to feel much more naturalistic we relied on this extensively in Skyrim where we had to create so many caves and natural Interiors but we found ways to apply it Forward into Fallout 4 because it was such useful part of our workflow in Skyrim being able to do things like create sweeping architectural curves and lines inside an environment without having to make custom art to handle every time we were interested in trying something like this something else we rely on extensively which really boils down to an editor feature is the ability to specify a local snap parent some other editors have this feature and call it a custom grid so when we're creating a new space we will tend to work off the origin snapping along the world grid coordinates and that's all fine and it snaps perfectly but sooner or later we're going to want to rotate off at

some arbitrary orientation and start tiling in a New Direction what we can do in our editor is we can point to this new snap reference and essentially treat it as a new origin and then snap along with kit logic and mathematical Precision on whatever arbitrary direction we want this allows us to create sort of Meandering flowing layouts that are dictated by the needs and the whims of level design rather than always being locked into custom art to get us off the grid and back on again all we have to worry about is the areas where we transition from one grid to the other one specific type of uh kit we use a little infrequently is called the pivot and flange uh we use pivot and flange kits in Fallout 4 we also used them in Skyrim uh where this example is from and this is what we call the ratway kit it's in the under the city of uh rifton and Skyrim and this is a one directional flow kit basic hallway pieces which snap together and the Pivot Point on these exists at the end and we're able to swing them around and within a certain threshold create these gaps and these overlaps we have an archway piece which seals that in and covers it visually and Night Lights nicely and that's good for creating

space which just feel like Once Upon a Time long ago they may have been built on a straightened level but they over time have settled into a more interesting organic shape now this stuff we did cover most more or less in 2013 and we went into a lot more detail with it so if you're interested in more on this stuff by all means that talk is available online it's free it's in written form um it's been localized in different languages go check it out um but today we didn't want to just give that talk again with some updates because we're always kind of looking forward and thinking about where we can go next with our workflow we never assume that we figured out the best way to make the game we were making and even if we had you know the needs of the next game are going to be different and need to affect the workflow it goes hand in hand and I think it's important to have this willingness to experiment with your process to always be striving to improve but you do want to be methodical about that have calculated experiments and how you approach it if you have a consistent set of goals in mind that you're working incrementally towards you will tend to achieve them over time there'll be some dead ends and missteps along the way but you're going

to improve if you choose a a vector and work along it sometimes we can get into workflow ideas and tool set changes which seem interesting but if they don't have the long-term goals in mind we can end up with something that's different but not necessarily better and if you truly misunderstand what worked well about a workflow before you start making changes to it you can end up with something worse off than how you began so yeah once a time once upon a time we work on a game called Skyrim and and then we transition to Fallout 4 so we think about the overall World Direction the World building the sweeping fantasy landscapes in the setting of a game like Skyrim are inherently different in several ways than the dense compound mixed use urban environment of Boston and the surrounding Commonwealth so as we start thinking about the world and what needs to be in it we think about where players are going to go where are they going to meet people Adventure what sorts of environments do we need just to create the texture of the world and and do a good representation of the location we have chosen and as this vision of the world begins to coales and Take Shape we

can take one of our first really concrete steps forward in production which is defining the kits that we need to create in order to support building all of these locations so that's what we did we kind of went through all the locations and came up with the kits that we thought we'd have to build and we apply our conventional logic to this which is to analyze which environments are going to be built with these kits how robust how important how extensive are they going to be how much time do we have in a schedule and then we start building out you know when each Kit's going to be available to work with now it's important to understand that every kit has a certain amount of overhead has baggage and it takes a lot of time kits are a commitment you know if we have 10 kits that are of a certain complexity and five kits that are twice as complex it takes longer to create the 10 less complex kits because you have to figure out how is the kit going to work what pieces do we need to establish getting people up to speed on how the kit is going to function reacting to requests feedback bugs and the ongoing Visual and cosmetic ownership and iteration of the kit itself and so knowing this and looking at the

long list of kits we need at the beginning of Fallout 4 is about when our allergies kick in we're really cognizant of time at Bethesda game studios even under the best of circumstances our games take years to make small inefficiencies redundancies in our workflow these time syns can really set us back and critically what they do is they prevent us to getting to a point where the game is playable the faster we can get to playable the better because we strongly believe that the first time a piece of content is playable is really the first time you can meaningfully critique it and decide how to iterate and having all these kits we had to build was going to delay us getting to that first playable point with all of the different environments that required them so it's through this lens that we returned to that list of kits and we looked at them differently and what we realized was for the most part this list was defined by the usage of the space before the apocalypse and several of these kits had something in common which was that they followed similar construction techniques in terms of how they would have been built in the real world foundations supports pillars uh

you know plaster brick the finishes and the building materials might have changed but the general function and the general proportion of the spaces more or less were the same you we try to think of kits as solutions to specific problems and to make sure that every kit justifies its own existence again we're looking for redundancies and ways to iron these things out and what we realized is that all these kits were solving a similar problem in terms of the type of construction method used to build them and we could consolidate them down until one larger much more robust kit that we' simply call the building kit and the building kit is going to be the primary example we use to analyze today so like Joel said the building kit was used out to build a lot of spaces for Fallout 4 but this has its roots all the way back in Fallout 3 with the office kit by Robert wisuki and this was a sort of standard kit everything snapped together on a grid but it was called the office kit for a reason if you wanted to build an attic or a basement you'd either have to use a completely different kit or you just couldn't do it because it didn't exist

so we wanted to take all those different construction types and unify it so we could just start by building one level designers can keep working and then we can swap it out later as those come online and this had a lot of our standard stuff that we have in all the kits small Halls medium Halls large Halls so on but the way we built a piece this time was different previously when we would build a piece you'd kind of Frankenstein it together in Max but then if you needed to say update a door frame you'd have to go through each one of those and update it so it was a ton of redundant work so you would export out a piece and here's your piece it's one big package deal there it is but this time around we assemble the pieces in the editor and it ends up looking like this so it's composed of all these component parts of the floor the wall and the ceiling so your actual kit ends up looking like this instead of just building all those possibilities you're spending time on visual variance and the actual capabilities of the kit this lets us do a lot more things with destruction we can just swap out parts of it for a different kit entirely layouts can do whatever and if we want

to build a new piece we can just do that in the old workflow if I had a corner piece and I have a door on the left and a window on the right that's a piece what if it's a door on the left and a double door on the right another piece every time you add functionality to the kit you would have more and more pieces and you'd easily have hundreds if not thousands of pieces but now you can just swap it out and that's the main thing we've done is we've taken the kit and broken it apart and that's given us a lot more flexibility yeah and I want to take a few minutes to talk about how this shift to a more granular style of Kit affected workflow for level design now the notion of going more granular isn't something that may seem really you know Earth shaking or or you know inventive to an outside Observer but it had a lot of repercussions on how we built and it's something that we had experimented with at small scales before but this time we kind of went Allin you know and the main benefit the obvious benefit is it gives us a lot more flexibility but there are reasons in terms of you know Tech and Logistics and workflow that we hadn't gone in this deep with it before and the

one we'll focus on right now is the workflow because working this way is inherently more timec consuming so if you take a really basic blocked in space like this simple room with a couple of hallway stems using the traditional sort of clamshell style of Kit we can create this space in 20 clicks right 20 objects dragged into the editor but if you take this very primitive layout and you explode it into the more granular subparts that Nate was talking about now the object count balloons up to 123 objects from the original 20 so this is one of the moments where we have to think about how our existing workflow has made us you know strong at what we do but what we can do to make sure we expose ourselves to the power of the new workflow that we want to embrace and Nate mentioned this in passing but we did this uh how we did this was with an editor tool called a packen which is essentially a prefab so we could use the more granular pieces and collect them together on the editor side into a prefab and those prefabs were named with the naming conventions we were used to in the traditional style of Kit organized in a different part of the and we could quickly rough out spaces in

our existing workflow which we were comfortable with and then once we knew that the layout wasn't going to fundamentally change and we hit iterated and agreed on that we could then as needed explode out the pieces into the subcomponent parts and do visual variations and one-offs as needed this all kind of speaks to the priority that we put on things being ready and being ready at a certain time uh kits are absolutely fundamental to the way level designers work at Bethesda we literally don't have primitive geometry tools in the editor it's the only way we're able to work and because of this we frontload these into the art schedule we work with production and environment art to make sure that they're some of the first things made in the game now level design doesn't mind that they're of gray boox quality early on as long as we have the proportions and naming conventions that we need to work with they're going to visually get better and better throughout the game the main thing is that early on we know that those Kit pieces are pretty solid that they're trustworthy uh this gets a little bit into how we schedule how we think about that you know like anyone else who's thinking about schedules we try and be aware of our dependenc where where and when and how we depend on

other members of the studio to do work but we also want to make sure that our creative process is built in a way that flows in the same direction so our production reality and our you know ideal creative process aren't working against each other which can so often be the case this is actually something that I've talked about at GDC prior um in 2014 this again we'll make this talk available to everybody at the end but all you really need to know about this for right now is that levels are built in a five pass system and each one of those passes is Tailoring made based on our expectations of dependencies on where the game is uh at a certain point in the project as well as how we think we can be most effective in iterating the level uh creatively and so in the context of this conversation pass one is where we're focused on layout which means this is the first point at which our core kit elements need to be usable but the probably the more important subtlety to this is in the subsequent passes we also need to make sure that the layout we did in pass one is reliable that we're not having to go back to the drawing board and redo the first pass every time we do a subsequent

pass because we're just going to Hemorrhage time and fall behind if we get into that trap over and over again and when you're in sync with the rest of your studio about your priorities in this kind of workflow you're able to rough spaces out very very early on this is a screenshot from conquered Museum of Freedom Fallout 4 and this comes you know this is in 2012 this is at a point where Nate and I have been working on the building kit we're in agreement about the basic functionality the pieces that we're going to represent even though it's pretty rudimentary and if you compare this to the shipping version of theace face you can see the similarity immediately and that's because Nate and I were aligned and What mattered to us most and probably you know the nicest thing about it is we we were out of each other's hair like once we knew how those pieces were going to work I was able to do you know level design type of things and focus on all of those elements and Nate could focus on the artistic side of things and we didn't really have to conflict with each other in the later stages and the years of time that passed between first and and final and so you know I find myself talking about priority a lot like when

thinking about this workflow and preparing this talk for today and those priorities don't just apply to sort of the meta scale of workflow uh in our scheduling but also to the individual elements of a kit if we're very general about it we can sort of lump all the elements of a kit into one of three categories you have your low use count hero pieces your set piece objects which you know only happen once in a while you have your utilitarian core pieces your walls floor ceiling stairwells all the things you need to create a basic layout to establish flow and blocking and then you have your variant pieces cosmetic variants alternates uh pieces which are maybe you know part of the core kit but they're used less frequently or spe uh special requests to accommodate needs uh of layout and again we're cognizant of time so we want to be cognizant about when when and how we build the kits and these different elements take priority uh going back to the 2013 to talk again briefly one of the points we made then was that the hero piece is you know numerically the least important piece of art we're going to build now there's this thing about set piece

art uh if you're a producer they're expensive to build you kind of want to get them out of the way early if you're an artist they're better looking they're more interesting to work on and if you're a level designer they can sometimes feel like the thing which is going to make a moment work which is going to make the level sing which incidentally I think is kind of a yellow flag in the strength of the level the thing about this hero piece is that these all kind of combine and make them seductive and they feel like the most important piece of art and sometimes get from OD in the schedule and we argue that we should do the opposite that they are the least important thing your humble wall your humble doorway these things you're going to see dozens of times and they need to be Rock Solid and bulletproof and they need to be that early on you know there's a stigma attached to words like generic to words like repetitive and I get it but when we're talking about this kind of workflow I think it's shortsighted to get hung up on this to get fixated because it blinds us to the potential our generic pieces are by their nature versatile set pieces aren't they're by their nature built to solve a

specific problem with very little hope of being reused ever again so the flexible pieces you have in the hands of the right level designer become problem solvers they become you know the the tool that you can use in many situations they are the unsung heroes of level design if you take a scene like this this is from the sagus Iron Works uh and this is sort of the penultimate moment the scene at the end of a a quest and a and a dungeon space the level designer here Drew would have loved to have a bunch of custom artst sell this moment you know he cares very much about how this shot is framed and how the player enters a space but this is actually you know a space where you're not seeing a bunch of custom art this is a whole lot of generic art working together in concert with good composition lighting and staging to create a moment which has all the trappings of a set piece without actually being a set piece and the important thing here isn't just that you can do this but that it spares the time the time that we didn't have to spend building customer for this is time you could use somewhere else on something

else that could have been more broadly applicable to the game as a whole so I think you saw this coming but I'm inverting the priority right I think the utilitarian core needs to happen early and fast the variants are important too and the heroit is the least important thing and we actually push those way to the end of the schedule and this mentality is all about getting out of the way of progress making sure that you're not blocking anybody else's ability to do meaningful work and I think if you're the artist it behooves you to minimize a road boox to not be that bottleneck for one it's going to get you in a position of having real examples of how your art is being used as quickly as possible and you're not relying on your own guesses and assumptions in the test space about how it's going to be used it also allows you the most time to iterate on the visuals of the kit and this particular point I think some will find counterintuitive because I think there's a tendency with some artists to want to figure everything out about how the art is going to look how it's going to work at every nut and bolt in place before giving it away for somebody else to work with but the problem here is that then you

become the bottleneck right you're holding up work and that pressure to figure it out and get moving can force you into making decisions you might rather have more time to think about but if you just focus on the proportions and the functionality which is all that level design really needs to move forward you can then get out of the way let them start working and have basically the duration of the project to focus on the Cosmetic aspects of the kit now for level designers this can also be a shift in thinking because you're going to have to use placeholders you're not going to have your fancy hero piece early on uh I actually don't think this is a big problem it's actually a useful exercise because level designers then are you know incumbent upon us to prove the actual need uh take for example the Boston Public Library like many of the spaces in downtown Boston it's a beautiful building filled with beautiful things you can spend an afternoon in this uh in this place filling a notebook with pieces of art that you believe you would need to make the space but knowing the way we work the level designer here darl he didn't have that luxury he needed to make progress he needed to get started on his

iterative passes so he's using generic pieces he's recreating the key rooms the areas that we feel are important to the space and while that's happening a couple of other things are happening the game is progressing the kits are getting more robust their functionality their visuals those placeholders are becoming a little bit more cemented into the layout and we're finding the places where the placeholders work and where they don't uh we're also finding more about the gameplay of the space as darl iterates you know we have guesses and assumptions but until we can actually play they're just guesses and assumptions and what we realized by the time we were ready to do any custom art was that this space this reading room was the most important room in the library what's more is we couldn't recreate this room with placeholders the barrel Vault ceiling that you see at the top and those big reading pane windows there's nothing in the kit that could handle that and those are the custom pieces that we made for the Boston Public Library the thing about this again is if darl had sat down and made a list of the custom set pieces he thought he wanted and prioritized them and we made them at

the beginning these might not have been at the top of that list we would have possibly built art which wouldn't have been showcased well may not have even been necessary and what's worse is we might actually have gotten the time to build these pieces thinking they were less important than they really were so visual variety is one of the big go goals with the building kit and one of the big ways we did that is all the architectural variance so with this setup we could take any room in any dungeon and just swap it out like I can start at wallpaper and it become The Deco kit it could be a concrete basement can make it into an attic wood paneling a brick Factory this ends up being a ton of pieces and just one of those architecture types is you can see it there but because all the logic is the same across all these kits I was able to bring new versions of this online in as little as 2 weeks you spend most the time doing just destruction and little cosmetic things you don't really have to worry about the logic it's there so that really makes it a lot easier this lets

us do a lot of things we could never do before like this where in the back there's brick walls goes down to this concrete basement there's these barewood platforms suspended there and I can swap out any one of those elements because it's all broken apart and we can do things like this because the walls floors and ceilings are all separate it can just open up into a larger space that's not a custom transition that's just what the level designer did for this particular area or things like this where it's stacking vertically and there's The Deco kit on the bottom it goes to the wood panel kit above that that previously when we did vertically stacking kits the logic just gets insane and it's all the naming conventions don't make any sense people get confused and just don't do it but now you can just put stuff on top of each other and it's easy but you can still do really boring stuff like this but even this one we couldn't do because it's actually wood panel walls in the back but the floor is The Deco kit that would have been a whole new export before now I can just swap it out or you can just blow the whole floor out the walls are just doing their own thing and these little damage platforms can be moved around around so every hole can look different

in this case it forms this ramp that you can walk up the level designer is deciding what to do that hole could even be a 100 yards long I didn't go into Max and make this 100 yard hole piece that would get used once but the level designer did this and it works out really well and we can do things like this where the wall is angled at 45° or it can go on these gentle curves and we when we did this in Skyrim there was a normal kit behind those walls and it was all this wasted space you couldn't use but now the walls can just do this material swaps are something else that we did and this isn't an architecture change it's just the textures and these don't take very much time and it lets you get a lot of visual variety so I can take that same wallpaper room and instead of changing the architecture I can just make it crazy 50s style wallpaper and this boring blue one another interesting pattern old people wallpaper just Chang the green stripe to a red stripe intricate green deep red whatever most of those were made in like 2 days we can now go and take every single room in a space and make it feel

unique because the walls floors and ceilings are all separate those can all be material swapped independently not every kit is the building kit though and one of the first things we ask when we're building a kit is you know what new functionality does it provide is it doing something new sometimes the answer is well no let's just make another building kit but that wasn't always the case and one of those is the industrial kit and we wanted these big blocky concrete things with these pipes rising up and lot of intricate detail work it's definitely not the building kit and this ended up looking like this so the way this works is that there's these Big Blocks that form the core and just like Mass out the shape of the building but then there's this intricate framework kit that has frames of every size and every one of those sizes has variants that have X cross frames V cross frames a cross frames there's a silo setup where there's a bunch of different widths and these are vertically stackable so we can make it as tall as we want there's a bunch of different pipe kits some are big enough you can walk through others are just tiny details then there's this free room kit

that we can put on top of those concrete blocks and you can go inside of these take cover we can put machines in there ends up looking like this everything going on here is because the level designer wanted it to be that way I'm not dictating this as the artist that catwalk is there for a reason it'll look different every time here you can take cover behind those pipes that room is tucked away in the back this space has a totally different feel with his spiny little pipe moving through it and all the frames are material swapped to red or could just take parts of that kit and mix it with other kits here it's being used with a steam tunnel kit there's a lot of Versatility there the utility kit is another kit that we built and this was all about this intricate pipe work and then it's snaps together in a certain way and there's a lot of baked in detail and the hallway kit the left side is different from the right side and these can move independently of one another and we also have these long curving pieces to get away from that rigid nature of the kit and that ends up feeling like this so using this kit we can have a really simple hallway like like this and we just throw a couple trash piles on the ground and it's pretty much done we

don't need to do this big set dressing pass to make this look interesting but that kit kind of wasn't worth the time all we actually cared about was the pipes cosmetically and architecturally everything going on with the walls was pretty much the building kit and in some cases that's even what we did here it's utility kit on the right building kit on the left and some of the pipes you see are a separate Pipe kit but it was a really nice kit to have it's what we call a glue kit cuz we're never going to build a space where it's all utility kit just this giant building full of pipes instead we just use it to spice things up a little bit and it was really useful to have the steam tunnel kit is another kit that we built and this is a lot like the ratway kit from Skyrim where all these things are jammed together organically so we don't actually need that many pieces but the main difference is that the floor is broken off and it's not baked in so you can have a really narrow space like this or make it taller make it taller so the level designers can really play with the verticality you can have these raised up areas on the side and drop down in the middle the walls in the back can be angled however we want you can blow a

hole through with these cinder blocks and then bring in the catwalks or I can just take those cinder blocks and put them up against the walls a decorative element and bring in industrial framing on the walls and have a totally different feel it's really flexible another kit that was built by one of our artist Claire Struthers for downtown is The Deco kit and this was built with a lot of the principles of the other kits where there's all these different parts little trim kits the lobby kit the foundation so it can mix with the street and one of the main things this kit let us do is the these interior exterior areas where you could just move right through it previously this kind of thing would have been a load you'd go up to a door activate it walk around for a little bit activate it again and load back out now you can just move right through it and this kit was really flexible and let us build buildings of all sorts of shapes which works really well with the organic nature of Boston streets yeah and I do want to talk about Boston for a little bit because for one there's a tendency talking about this kind of workflow and level design to FOC Fus on intimate spaces Interiors combat Arenas these sorts of things but

exteriors can be built modular as well and in fact Boston downtown was the most ambitious thing we've ever attempted in the level design uh scale at Bethesda uh huge in in indust locking and in order to start talking about and understanding how we approached Boston we have to go all the way back to Washington DC in Fallout 3 now when it came to figuring out how to represent DC's downtown for a number of reasons we had decided to block it off so all the downtown neighborhoods were these little sort of cloistered areas these walled Gardens that you would load into which were separate from the open world of the game they all kind of had this consistent look and feel where you can see the the buildings and the big warble piles on the edges they they were walls they were literally walls and that the meaningful gameplay happens in the negative space in the interior and the in the ground level layout now this made getting around downtown really complicated this is a map that users made to help them understand how to get from place to place and I think the complexity of this map is indicative of some underlying

issues with the implementation that we chose so going into Fallout 4 and figuring out how we wanted to represent Boston it was important to us that it was part of the open world and felt like the Beating Heart of the Commonwealth we would have long sight lines and clear lines of view we'd have Avenues and streets which were actually useful for navigating the city as you would on as a pedestrian we wanted neighborhoods which tapered off naturally Into the Wilderness and when we started thinking about how we're going to represent a space like this we can get into conversations about how accurate must it be are we going to try and recreate the city street for Street Building for building brick for Brick it's sort of the can I visit my college dorm mentality of of you know City design and we knew that was never our priority we were always more focused on authenticity can we understand what makes Boston special and distill it down into a core form and represent the Fallout version of the city in a way that would feel familiar and sincere to Boston but also like a real real location with a real internal

consistency and a sense of place to those who have no familiarity with a city whatsoever so with this mentality we approach the actual process you know of building the city and where we begin is with the coastline one of the first things we create for any of our games is a rough pass of the world map now Boston is defined by water the Charles River Boston Harbor the Atlantic Ocean these give us hard boundaries hard edges to most of the areas of the city we're going to to recreate we can take this and compare it against the scope and scale of the game map that we're going to build and now we know the square footage we're working with now we know the you know the U how much we have to condense downtown and then we can start focusing on the skyline the skyline of Boston is important it's not as iconic as a Skyline of you know Shanghai or Paris or London but it's a really useful Skyline because it's asymmetrical and it provides an orienting elements for players and in order to recreate that orienting element we have to focus on the anchor buildings that Define the skyline and get their relative

positioning correct so this is another old screenshot of the game in the early stages uh Raph one of our environment artists he went through and used really temp buildings to rough out all of downtown and what we were focused on here is roof Heights and density and specifically on those anchors the cluster of skyscrapers in the financial district which is on the East End of the city and cuts off into the harbor the sort of solitary you know nature of the Hancock Tower in Back Bay and then Diamond uh Diamond City or Fenway Park kind of abing at the West End of the city and we did end up sticking to this original pass Pretty Faithfully all way into the finished game we were able to add a few more anchors few more landmarks for players to you know fixate on and this is something that you know players familiar with Boston were able to pick up on from even our earliest uh trailer right people were able to take the background shot and make accurate guesses about some of the buildings we're representing and the position of the camera so with the skyline ready to go we start thinking about the thorough is the streets what is the skeleton what are the bones of the city and when

you're representing any City space you know you don't have to recreate the city street for Street in order to get the feeling of navigating the city correct if we're focused on creating say Manhattan you're going to be thinking about long blocks streets Avenues it's a city planner City this is not Boston Boston was created by filling in Marshland and Paving cow paths it's a notoriously nasty City to drive in but a really wonderful pedestrian City so that's how we experienced Boston we went to Boston we spent time on the streets just moving around the city and just kind of experiencing it and and existing there and getting a sense for what are the streets we RI on for navigation what is our mental map of the city what intersections matter to us what points of interest where are they in relative uh positioning to each other and then with this sort of sense of priority about what we're going to represent we can then go back to our condensed version of uh Boston in the world map and lay these streets out and we did lay all of her major streets out we knew what they were going to be named and how

they correlated and we stuck with these layouts with very little change all the way to ship once you have streets you can start thinking more specifically about neighborhoods like any major urban center Boston has distinct neighborhoods with their own culture their own history their own architectural you know uh quirks and when we're starting to think about the neighborhoods that we're going to represent and try and focus on we can start thinking about ownership as much as we can we try to make sure we assign folks to uh areas that they're going to own from concept to completion and so each neighborhood was split off as this own chunk of work that specific designer and artist would work together on and they'd be responsible the onus was on them to understand that neighborhood and how to bring it to life and how to do it you know how to do right by it so with all of this we're ready to start creating we're ready to start iterating so this is all kind of pre-production this is all ideology and getting ready to to attack the city now I mentioned it before but we have a process for this um you know I talked about it a previous year and it's this iterative process that we go through and we build things out it's

worked really well for level design and it only made sense to apply this to downtown you know we have the neighborhoods we would treat these as super levels and we would just give them more time and attention and but put them through the process that worked for us it didn't work it didn't take long to realize that building the city was going to be really hard we knew it was going to be hard we knew it wasn't something we had attempted before and there were going to be you know wolves in the woods and and unexpected problems but there are a lot of unexpected problems it seemed at times like everything we knew about level design didn't apply our best practices in pacing and you know setting up encounters how our Ai and detection worked uh how we optimized how well does the editor run how do we organize how do we work together nothing was going as planned and all these hurdles were adding up and taking a lot of time from us you know there was more revision rethinking and reworking creating Boston than with anything else and it was taxing the schedule of uh metrics and best practices that we had in place but

while all of this is going on and and all this is going wrong there's something else true which is that we're building the city modular in a way that we also have never done before now it's not like we' never built a city SPAC modular before for example in Skyrim White Run is built with a kit but it's built with the white run kit it's only used to build this city and the only person using the kit is the same artist building the pieces and using them to create the city in Fallout three the level designers were using building kits to create those downtown neighborhoods but those kits were pretty chunky and that's just kind of where we were in terms of the state of art for us as a studio at the time but that was okay because even though we didn't have a lot of flexibility to create different shapes to accommodate our layout goals uh remember how I mentioned before that there were wall Gardens that made it really easy to adapt if we had a downtown uh space in DC and we wanted to try something else with it we could just make the change to the layout and just move the walls of the garden to accommodate it was really no big deal but in Fallout 4 everything is connected

everything's part of the open world we cannot make a change to one neighborhood without affecting the neighborhood next to it and potentially the one next to that and creating a ripple effect that undermined work throughout the city but like I mentioned we were building the city far more modular than we ever had before we were taking our lessons from interior kits over the years and applying them to how we built downtown the pieces were far more granular which gave us a couple of freedoms one when there was so much work to do we had more people able to do the work level designers and artists both had a tremendous amount of power to react to issues if we needed to change the the shape of a building we could do that we could help an artist an artist could help us and we could make changes without having to make big custom pieces and get Turn Around from the art schedule out of Max l dot and also it let us respect things like the Corky Street layouts of Boston we didn't have to conform streets to the buildings we could conform the buildings to the streets and this meant that when we had a problem that required us to make a change we had the ability to do that we

weren't locked in to bad decisions we had inadvertently made early on with pieces that were far too static to be flexible to iterate with so while working module this way you know it had it its own difficulties and problems for the most part it did decrease the pain and Offset you know all the problems that we were running into with Boston and let us maintain a focus on the quality of the City without having to make a bunch of compromises and how we uh approached it because of all the hurdles we ran into kind of Saved our bacon so now you're going to get the grab bag this is the stuff that didn't quite fit into the other categories but that we wanted to cover one of those is the plug-and socket system and this is how we transition between kits previously when we transition from say industrial to utility that was a piece we made that was it if we needed to go to a different kit we'd make a new piece not very flexible so the way we solved it this time is on the top we have these sockets and that's just a whole of a set size and then the plugs are door frames and other things so as long as the holes match we then put the plug in there and

it works I don't care what kit is on the other side it's one of those things that was really really obvious in retrospect but we didn't actually do it that way before so it worked out really well and the way destruction works and kits is worth going into also and there's a big problem with destruction in kits and that it's really repetitive you just have this piece or handful of pieces that get used over and over and it looks the same every time so our solution was to have all these different little pieces that could get used together there so then it can look different every time so for the wall holes in the building kit there's 11 versions of the hole but that's only actually half the hole the other side can be just a different variant or a different kit entirely so if it looks too much like one next to it I can just start swapping it swapping it swapping it just swap the one on the other side so I can have this line of holes swap them all and again the other one is being swapped independently so there's a lot more potential versions there and it's really flexible this lets us do things like this where I have this hallway it's the concrete kit material swwa to look like

cinder blocks and I can just Kool-Aid man through wherever I want I can even go to a different kit here I'm going from building kit to utility kit back to the building kit the way floors work is a little bit different though and it's easier to just see it and at the bottom you can have these Square pieces and those socket in with the normal kit logic but then there's all these free floating platforms and these exist in all the versions of the building kit and this lets us do things like this where we have these really organic hole shapes and that's not a particular hole piece it's just whatever the level designer decided decided to do and there's this ramp you can walk down or they can blow that whole floor out and then those platforms can be sagged independently so there's a more organic feel to it and we can do things like this where the whole floor is just caving in and in this case we didn't build a church basement kit it's just the building kit on the bottom and then we spend time making unique architecture on the top and there's not a lot of wasted work Dynamic destruction is a little bit different though and this is for like per piece destruction where you can shoot out individual

planks and it works with kits and they'll simulate so you can have things like this where you have a shack and you start shooting out individual planks or you take a grenade and Chuck it in there then a whole bunch of planks go flying out and again this works with kits so we use this for railings and other things like that decals and gbls are another thing that are really obvious but extremely useful so you can have a really boring wall like this and just throw a bunch of like rust stains and cracks and grebles or like fans and Vents and things like that and this isn't baked into the kit these are all just placed independently so if I want to take that fan and it's rusting move it over a couple of feet you can just do that now every wall can look different and there's a lot more opportunities for environmental storytelling layers are another thing that are really obvious but save a ton of time cuz sometimes you load into a scene and it looks like this and just I don't even know what's going on but you set up layers and you can start freezing things hiding them extremely useful that's not even the worst example if you go downtown that list is just really

long the mouse wheel swap is something else that we made and previously if we wanted to swap out a piece you would click on a piece bring up a menu and then pick which one you want hit okay maybe that's the piece you want maybe it's not really really slow but this looks like this CU you can just click and scroll to the mouse wheel it's great for finding variants and like that wall hole example I showed earlier I can just do that in a matter of seconds you can go through a whole dungeon in just a couple minutes big timesaver and this also works with material swaps helper markers are something we talked about in 2013 but we were being kind of sneaky at the time cuz we had real kits that were using it but we couldn't show it so I had to make a fake kit to show so this is how it looks in practice and there's all these editor only markers to show things like where the doors are how the walls line up and it snaps together markers for the ceiling so I don't have to flip my camera upside down just to click on it they're really useful but the modular workflow is useful for more than just architecture you can also use it for props and things

like that and one of those things that we built is the machine kit and there's this large library of parts of grebles and fins and tanks and pipes and an LD can just block out an area and say I want a machine here and then an artist can come and do that one of our concept artists John gravado he was really heavily involved in doing this and it's was a great use of his time because it was really towards the end of the project don't exactly need a lot of concept art then and this ended up looking like this there's all these component parts and those are the potential material swaps on the top and we can build things like this which are composed of all these tiny little parts so you can make all kinds of different things even vehicles and arcade cabinets it's really versatile and you can also use it with other existing art like this truck and in like a normal workflow each one of those things it could have been a custom asset someone goes into zbrush for like 2 weeks you get a custom texture looks great that's awesome but that takes a long time and using this setup John was able to make 400 machines we wouldn't have been able to do that

otherwise so think about how you can use the modular workflow for more than just kids okay so I'm going to wind us around into the home stretch um you know there's a lot of ground to cover in the talk today um and we kind of squeezed in everything that we thought we'd really have time for but the talk has bounced around a lot um just because we wanted to fit it all in and honestly you know Nate myself many of our colleagues at Bethesda we could sit and talk to you for you know another hour or two where the rest of the day about the ups and the downs and the nuances of this workflow you know and it is a workflow that has ups and downs it's not a shortcut it's not a you know a solution for all scenarios and with this and any other method or approach whether it's you know level design or production or you know best uh coding practices it's more important that you understand the you know more important than understanding the workflow that you understand the priorities that you have as a team and the priorities that your game has because any method like this you should understand before adopting because it's never one siiz fits all and

I strongly believe that when it comes to process and technique you know it's the goals of the game the Ambitions that you have as a team that should Define the process by which it is made and never the other way around thanks very much well we must have been nerv cuz we have some time um if anybody has questions we've got microphone set up here you can kind of queue up and we'll take a few you in the middle yeah I was wondering at what point in the design process do you factor in play tester feedback regarding design of levels or Aesthetics so we have those various iterative stages and all those stages can be independently verified uh so what we'll do is we have a few different processes we will do peer review so people will look at them together and offer feedback um we will do it within

the team uh we can do it with our QA Department uh we'll occasionally do sort of big batches of review like we call it level designer summer camp where we all get together and we all look at every space in the game and come up with a bunch of feedback and we structure the feedback based around each of the iterative passes so that we kind of know what we're looking for and how we can improve cool thanks uh over here on the right hello uh my question is what was the thought process behind choosing which pieces to give to the player from the kit that you already had you mean in terms of the workshop system yeah Workshop was a kind of a different system so it was just whatever we thought would be a useful piece for them to have without it being too overwhelming like we didn't want those kits to be like the building kit where there's 500 pieces so we tried to pair it down a bit more yeah the functionality the workshop kits is slightly different from the ones that we used um but I think it's fair to say that over time we started to like the differences between Workshop kits and

our regular kits that it actually got more similar over time than they did get different yeah thanks thanks uh question over here yes um now you mentioned that your process changed uh between Skyrim and Fallout 4 now given the increased Gran granularity and difficulty regarding uh you know recreating Boston in Fallout 4 how has the um collaboration between artists and level designers evolved if it did um well one of the big things that we have is there's an art review process since the pieces are so granular you can pretty much do whatever you want with them so we make sure to check everything and give some notes which they may or may not agree with but it's a bit of a back and forth and it's gotten more official I guess as it's gone on yeah the main thing is that you we have slightly different sensibilities in terms of what you know how important is it that something looks good and how important is it something you know works good uh and so by making sure that we're kind of both involved in creating the

kit but then both involved in kind of vetting and signing off on the the spaces that we create with them we can kind of establish parity and sort of keep the peace between the Departments all right thank you very much thank you in the middle again hey guys uh first thing great talk and then secondly my question has to do with performance and metrics how did that going more modular affect those aspects of the game it screwed it up uh our rendering programmers were not happy with us when we made this decision did they hard cap you no um one of the main ways we solved that is that we ended up putting in an object combination system so even though it was broken apart and all these different pieces it would get chunked in and then visibility after the fact so it was didn't really matter in the end just when we were building it and running it in the editor yeah but there was a lot of there were a lot of tech that had to be kind of reconsidered to make this workflow possible thanks yeah over here so yeah I sort of have the same question I guess a little more depth but okay um yeah how

do you does basically just everything get sort of bake together the final game and yeah there is a baking process so static collections are one way that we can control essentially baking we can select a group of objects and say Yep this is a static collection now and those get handled a little differently in terms of you know how they how they render and how they olude uh we also you know we have a baking process which pre-calculates visibility from you know various directions in the kit and then we use that visibility to compute quicker on the fly at run time and I guess also like do this do any instancing I mean how how much can get baked like if you have like a Lev or like you know the church scene it's like you have a lot of duplicated Parts but then the upper level seems to be pretty you know sculpted I guess we're not actually instancing it's baked out as a large chunk and the only things that are not baked out are Dynamic objects like a tin can or something like that that wouldn't get combined okay thank you so uh you mentioned you had a review

process for uh how when you have conflicts between I know we we do environment uh art modularly as well and we had a lot of fights uh like unit abuse or like are you using this correctly you're not do using the unit you're supposed to um do you solve that by like a creative director stepping in saying what what is more important or how do you how do you do that usually I mean we kind of have ground rules like we generally and some of this just comes with working together for like over a decade right like generally speaking when we fight we already know who's going to win like based on the context like yeah I mean like at the end of the day like one side does have to win um and it's kind of case by case uh it's not too often I guess it happens once a while where like a third party has to kind of step in and decide but not often though I think the main thing is Art just has to be kind of okay that things will happen yeah I I think that's the

slowest part of adoption in the process is level designers are going to be artists with this you're going to make decisions and the important thing is that there's a two-way street right like one of our level or one of our environment artists is like you know was a practicing architect and came back into game development and like some of the stuff I do drives him up the wall but he'll tell me why and at the end of the day you know like say like I hear what you're saying but we can't accommodate or he'll end up inputting something better than the process so I guess don't be jerks right like just just talk yeah just just talk and realize you're going to win some you're going to lose some like the game is what matters not individual Egos and did those reviews help in um mitigating bigger fights down the line like if you have if you check in once in a while does that help rather than has that been a new part part of it yeah I think it helped a lot and cuz one of the other things that came from the reviews is that the L designer sort of got how things would work and it would get better and better over time so then the

notes would just get shorter and shorter yeah yeah and for instance like with the review process that Nate's mentioning like really that was Nate like for the most maybe entirely like going through and looking at all the spaces providing notes but like the level designer and Nate like both knew that the level designer had the prerogative to ignore everything right usually they they didn't because there was good feedback but Nate had a good barometer for like what was what was sort of misuse that he didn't like versus what was a bug and like a bug in as a bug right like this is going to cause a problem because you have used it incorrectly versus like I don't like what you're doing and here's a suggestion for how to do it better thanks a lot M over here yeah great talk um so usually you know when you're building well a large game uh you have to use kits from certain regions and areas so you don't mix and match you know tons of different uh you know texture maps and everything um you seem to mixing a lot of them was everything baked in at at the end or uh

was that some of the optimization issues that that you guys ran into uh without extending the question too much what do you mean by everything everything is baked in uh say say texture mats for so for instance an example in something like Lord of the Rings Online um you had an orc kit and you could mix that with maybe another race but you can't mix you know six kits together because you're going to pull in a whole bunch of different kind of texture pages so you're you're making more of a technical Point than like a World building F yeah yeah yeah yeah technical Point yeah yeah we we kind of you use a lot of vram but I mean you're only pulling in the parts that you're actually using since it's all broken apart so like in that steam tunnel I bring in the industrial catwalks and that's it like it's only that one texture so that helped but it's still a lot yeah but we we we can make I mean we have spaces that have like seven eight kits in it um is that what was that a two more questions or signning or or no more two minutes two minutes all right

right here when building the game so much the game with placeholder initially do you run into difficulty with people understanding the final Vision or people feeling that like working in vacuum or does concept art and previs play a huge role in that or do you guys work so much in this system that it's not an issue well I think one of the main things is that we don't like gray box gray box we we try and take those kits and put in the details that matter so you can really get a feel for what it's going to be even if it's kind of crappy and not done and that's made a big difference cuz when we Ted tried to just make a gray box kit that's just blank walls ends up not actually solving problems and it ends up being a problem down the line yeah all right hey guys uh once again thank you for the talk I found it to be really informative and I just have a brief question about like the kind of modularity and prefab sort of system that you guys were using you mentioned it briefly but it's something I haven't experienced in like previous bethesa editors that I've like I've experimented with so like you mentioned that you can

tie all the pieces together and then like arrange them in chunks I just wasn't sure how you would like coordinate with all the LDS and artists like how to make sure every prefab meshes with every other one like I said uh we have a convention that we're used to from prior games like uh you know when when the Fallout 4 creation kit drops out right you'll be able to load that up and you'll see if you're familiar with our modding uh recognizable naming conventions recognizable like piece shapes and so we'll have those static collections bundled up in one part and in a separate folder usually and then subcomponent pieces and it's just like we have the sort of institutional knowledge of the way we expect kids to be arranged and we just mimicked it okay thank you all right I think yeah we're out of time but if you've got more questions feel free to come up but we got to we got to tear down here for the next guys hey there thank you all