A New Dimension to Below: Personal Discovery and Adaptation

[Music] hi everybody I'm Dan John Cox this is bringing a new dimension to below personal discovery and adaptation I'm a 3d artist at capybara games and specifically today we're talking about below blows a game I've been working on for about two years not the entirety of the production but specifically what we're talking about is first off how below kind of got its visual style what is involved in that and in a somewhat technical manner as well sort of my personal journey starting with the team and integrating more 3d into below the difficulties and tricky elements that come from switching from ubisoft style production into something much smaller much leaner and a little bit crazier but first off what is below below is a procedurally generated roguelike game with an extreme difficulty level where if you are hit by a single enemy you can start bleeding if you can't stop that

bleeding then you die and you'll have to start the whole area over again the world however stays persistent and will continue through every different life that you have when you die and you start over you start over as a new wanderer a brand new person in this dark and foreboding world trailer it's always nice [Music] [Music]

[Music] so that's below there we go first off if you are looking forward to below in some way don't worry this talk has no spoilers in it there's a lot of secret stuff that we're hiding and you won't get spoiled on any of that so first off let's cover all those visuals or what happened before I arrived so none of what you're about to see is pretty much anything that I wasn't particularly involved with but it does set the stage for how things work in the first place so first off to understand below style and visual aesthetic you gotta understand cappy's history so they worked on sword and sorcery critter

crunch Clash of Heroes and super time force all of these are very 2d and animation based games with beautiful animation and 2d art no 3d so when you look at Bulow's style in the first place in some cases you'll see lighting and things like that in some 3d objects but it's not quite so simple in fact pretty much all of below is two-dimensional when you see it in reality some objects are 3d the character itself is 3d as well for a variety of resource intensive reasons but that's basically the set up that you begin to work with in below is that you've got a three-dimensional character on totally flat backgrounds but it ends up looking clean and interesting and has a beautiful style in and of itself but every background that you're seeing basically is 2d and that works really well except when you want lighting which began to become more interesting element to put into the game so how do you get lighting onto a

hand-painted background well you hand paint normals which is a super time-consuming and intensive process something that you'd often think you should not do because it just is laborious but that's what Cappy went for they wanted lighting so they'd paint into the backgrounds basically each artist would just lasso over certain sections of the element that they were dealing with and just from this sphere here pick a value that's appropriate and fake all of that lighting we also use flat cards so when you have to walk behind objects some would be 3d in this case and some would be flat little shapes 3d there and these other pieces being totally flat now for shadow casting reasons that becomes really really hard to work with some objects would be floating above like this and particles as well would often be flat as well the effect in the end looks very convincing but ends up being something that's still pretty difficult to deal with it from a production standpoint on top of that we've got to this Zelda

style skew Zelda is a huge inspiration for Cappy and for specifically below but when you look at the structure of a level and the world you can see that the world itself is kind of top-down almost like a fisheye lens but the characters are almost perpendicular to the camera they are actually just pointing in almost the same direction you're not really seeing just their head as you'd be seeing in that actual reality for example even in the new games where they are still skewing the character a bit towards the camera you still see a lot more of links head than you do his sides which is very different from what the original games look like which was the aesthetic that we were going for early on so you can see here you see most of link himself and same for our Wanderers you see them basically pointing towards the camera which means in reality our character is kind of tilted weirdly towards the camera now there's a lot more math going on Colin white who's here as well did a talk on the actual rendering of how this process works and

it's incredibly complicated also difficult to deal with but it basically means that the character is pointing towards the camera objects can also be pointing towards the camera at different angles avoiding your head clipping into certain objects is incredibly difficult zelda does the same thing but their characters are all and their characters and objects are all at a basically similar kind of skew so that nothing starts clipping with other elements makes it more manageable but basically in the end the zelda skew means that we've got a level that has a different kind of perspective to it than the characters which is highly interesting to look at very unique very different from what you normally see but in the end where do I come in why would I need to be there well first off below started off mostly as a non lip game with maybe a little bit of a fake light source around the character and still really dark but lighting didn't play a significant component until they started to add in more light and see what that does for gameplay for example in this case if we look at this scene without any lighting the gameplay is very very different than

what you'd have and the tone of the game is very different from what you'd have once we add lighting back in this seems obvious but it's a it's more than just a shift of aesthetics it's a shift of gameplay and tone and that's there to become more important but becomes really tricky to do when you've got just two-dimensional assets when you've got cards that are supposed to fake cast shadows in situations like this real cast shadows really changes the feel of what you're walking up to and sort of the believability and reality of that situation so that's where I come in they needed 3d to work with so they needed to bring on somebody with experience doing that I worked in a variety of different Studios medium-sized to super small to medium Ken I've taught for about eight years in college teaching kids 3d art and I worked at Ubisoft for about three years so I feel like I had a pretty good sense of what a proper workflow was I've had to train people in doing that and had to make sure that I had a good understanding of what good proper workflow was and we've already got 3d in

the game basically it's what I was told so more should be easy which seems reasonable so chapter two what's hard is easy and what's easy as hard or basically our workflow was a bit crazy when I got there there is no editor in the game whatsoever the game itself is all you have to work with so when you want to move an asset you can't you need to do it through scripting and editing and nothing that's dynamic and immediate it's also a procedurally generated game no editor is pretty hard to deal with when I've gotten really used to one so our basic process at least for me was I would have to model something in Maya fine then I would have to compile it with a batch file that would take about thirty to several minutes then I would have to code or script it through text which is that my brain doesn't do that and then I would finally be able to test it in game but of course I would have to assume that the object spawned in the game

because I have to load it literally into the game hope that doesn't crash the game and that's kind of tricky to deal with so starts getting a little bit harder to work with now this was the first asset that I got into the game my computer's kind of slow so the gifts are running not as fast as I would like but it gives you an idea of what I was working with now for this object for example this is the Lua code that went into it which is lots and lots of lines of code which again my artist brain not super used to dealing with and I still really like technical stuff the other problem was my is King when it comes to our workflow so everything that goes into Maya is gonna go directly out into the game engine and I don't get to do anything in-between the problem was also Maya doesn't match what I was seeing in game when I started which meant for example in this case on one side on the Left I can see what I would be seeing in game that there are little skew and to the other side I'm seeing what I would actually be editing inside Maya now I can see them simultaneously but if I tried to edit anything I would immediately get an

overlay of whatever I was trying to work with so I can't see the silhouette so easily because objects are covering each other that would mean I would have to start shading something inside of Maya so that I could understand what the shapes and planes were so that I could hope that it would look the same in game because I'm just faking some lighting as well but if I was trying to edit it in a in Maya as well I'd be getting stuff like this which becomes really confusing to deal with and on top of that there was about 20 steps that needed to be done inside Maya or stuff would really break down and not work and you'd see nothing so you'd export you'd do a whole bunch of work you'd compile you'd script you'd look at it in-game nothing would show up and I would get pretty frustrated and confused and sad but there were valid reasons for this process first off cap had been working with the engine that we were working with for a pretty long time was pretty comfortable with it the artists themselves are very flexible and willing to deal with a lot of difficult stuff to get an interesting and result

this was not something that I was used to on top of that as you try to build up an interesting idea of what you're working with you'll change that workflow and sometimes you don't want to necessarily retool everything that you're doing before you get to the actual result that you wanted so there were lots of good reasons to have what I felt like was a crazy workflow but it was still hard for me to deal with on top of that certain things look simple to model you see a mountain here and when I would try to jump into modeling something like this I'd start to zoom in and see is the shape I'm dealing with a curve or is it a bend some of the shading that's implied suggests like a single plane to a surface and I'd think okay wait this seems a bit more complicated I have to flatten things out and do a lot more work than I was expecting it's fine but let's let's look at the normals of this see if we get a better idea of what the actual structure of this should be with shading that's the normals I can't make one big flat plane for a huge section of a wall that becomes really really

difficult to deal with and seemingly impossible so I would start to try to translate these things like this and this was the first version that I was dealing with now some surfaces get a little bit more jagged and and don't work so well I always have to flatten these big flat surfaces so that they look like the original I don't want to be the guy that comes on to a beautiful project integrating 3d into it and start making the game look significantly worse by just having a lot of angular phases and things like that that don't feel as fresh and vibrant as the 2d illustrations that existed in the first place and the geometry itself is fine but nothing terribly complex and flattening out all of those surfaces starts to become a big pain because it's hard to get those angles right and a lot of lattices or FFTs we're used to try to get that stuff to stay flat but minor edits become difficult to deal with on top of that below is a small game and I'm used to working with like broad stylized work but below is really small

and a lot of the things that go into it take up only a few pixels on screen such as this was the first character that I was dealing with this was just sort of a work-in-progress piece to try to get a sense of what the design of this character would be and I thought okay well it's got some detail it's got some shapes simple good fits the style but on screen it's pretty small and if you're dealing with that if you shrink this TV down that is the in-game resolution of it it gets real small and what happened is small little details on the character like the strings and straps and stuff like that start to flicker because if you've got anti-aliasing turned off which I'm switching here things start to flicker too much and they don't look like consistent shapes they just look like little details that are flashing on screen so that doesn't work so well up close it starts to look good but then my Art's not really working in game the way that I would hope so we redesigned the character to these big broad shapes something that seems a lot more manageable now even when anti-aliased or at they're just alias it starts to look

the same looks fine up close it's a relatively simple design but it's got some nice impact not designed by me but it looks good so that's satisfying but the actual modeling itself less so it's a really simple model which is fine that's what the game needs but I came into the production with a certain set of strengths and a certain hope that I could give the project something useful and when I was working on a model like this it kind of felt like well with the design already supplied almost anybody here could do this even with a little bit of skill and modeling or very little experience so it started to feel kind of frustrating like I wasn't actually helping out the team all that much on top of that a lot of my job became just trying to translate the concept art into 3d very literally because the concepts were beautiful and they worked and if I'm reach angel that seems unnecessary all that works already been done I might as well just use it so for example here that's the actual texture sheet for this piece it's

literally just taking the concept and sticking it in and I'm just kind of putting geometry on top of it but again that kind of feels like almost anybody could do that and I'm not sure that I'm giving this project much worth which starts to become kind of frustrating it and a little bit sad for me it was all really really different from what I've been used to and really hard to adjust to honestly occasionally I felt like a failure like I wasn't really supplying anything of great benefit to the project but Cappy is incredibly optimistic almost deliriously so certain things like painting and normals and these crazy skews that we were working with would be things that I would just simply say not to do but they are what make the project interesting and they are what make the visual style really incredible it their optimism is really infectious and something that when I'll kind of let it come in it became a lot more manageable to work with because really games are not really only made in one way everybody here has dealt with

different ways of making games and dealt with what seemed like stupid decisions but sometimes those are not that bad and sometimes they add charm so overcoming we're getting into it basically I started to try to embrace the strange embrace some of these seemingly stupid decisions because they're not stupid with context but they can make things difficult now sometimes smooth in that process out helps but embracing it first also really alleviated some of my frustration and and sadness in a way so first off the workflow some of it just didn't work and just got in the way and just made things unnecessarily hard so things like we got an in-game editor we got some Maya working faster we got a prop scripting tool useful environment shaders to speed up the process and I got practice by just being able to do this process more and more I started to understand where we could cut corners and start to break things up which we you get in any situation but it's hard to guess that before you get there so first off we got our a Magoo II editor

which was basically just some minor tools we could use while the game is running so that we could move things around which we couldn't do without scripting and then reloading the level so that hugely useful big time-saver and helps us understand what we're dealing with on top of that we could see what animation states were running we could open up the script itself by just clicking it in the game and that just kind of smooth out some of the process and made things a little bit more pleasing to work with on top of that we got a light editor normally lights would have to be scripted outside of it through a text editor so you'd have to type in the location of where a light is gonna point or where it's going to appear in the scene that's really hard too but because lighting became more important with more 3d we got it in game and that just helps a ton super simple like in hindsight but super effective and helpful for us as well as post process editing in game all those kind of things just smooth out the process they don't change things fundamentally we still have other hard problems but makes a few things a little bit more enjoyable on

top of that changes to Maya where we started to get a lot of scripting a lot of these - actually all of these tools were done by Nicolas Steven our animator who learned Python and 3d while working on below crazy guy but basically some of his workflow stuff that I was able to request and get help with was basically being able to edit in the game view very quickly so he made a little window that I could see where I'm actually seeing at the game scale how many pixels something is gonna take up and being able to edit that without having stuff blocking my vision so that I could move things around change them and work with them stuff that seems relatively simple but just makes for that little bit of improvement in quality of life on top of that we've got things that help some of the other artists who are being integrated into 3d such as a UV projector where it would take our game view and basically project the UVs so that an artist would be able to send that out to photoshop and just paint a background now changes to those backgrounds would then become a problem

later on but that's where useful shader changes can happen so for a scene like this these scenes are usually loaded into streaming memory so they're a lot more manageable and faster to load in than we would normally be working with so 3d assets for what I was trying to work with load in slower and load in a little bit heavier because they can't stream which meant a room like this I would want to go as low as possible on my texture memory but I would also want to have all of the color changes there's light shading that's been put into this in the concept itself and I would want to maintain some of that again I don't want to reduce the quality of these scenes but that's a lot of texture work to do and any kind of changes if I just hand paint this become awkward so if I wanted to make a room like this with new texture settings one of the things that we got was I can model out a scene now where normally to get texture on it I would have to create a big complicated atlas of just various color values with texture incorporated into them so that I could block out big broad colors and so

that they would have texture in them but planning these is a lot of work and you can see here in this one for example there's a big empty space that's really not being used and there's even a bit of concept art in it because it just wasn't being used that's a huge waste so the thing that we ended up doing was getting a much smaller kind of texture that we can put in of just color values and gradients I can apply that to the background just you Ving it in and then we'd get some basic shape form which works then when lighting is put in it starts to look a little bit more like what the original concept would be and the kind of style that was being used then I can use vertex colors multiply it on top to basically add in just a little bit of the painterly effects that would be used by the artists where they could darken out certain edges or brighten up certain ones have soft transitions and then finally UV project on some actual texture into it if you can see it they're basically just a little bit of texture work that can be applied in in the exact same kind of style that was originally being used for the concepts and then we end up with something that's

much closer but also any edits that I make to this are super easy to do we also got things like a prop scriptor which meant complicated assets I could press a bunch of buttons and a script would come out the other end that doesn't mean that I don't have to script anymore because it doesn't hit all our edge cases and we have a lot of edge cases but it means that the workflow becomes a little bit more manageable and I feel like I can do a bit more with the art and we can integrate more artists into the process a little bit easier as well and Gaughan's now because I've been teaching people art for about eight years I have a specific mentality about how things should be done and things like n-gons you generally want to avoid you don't want to have too many because that can change what your shadings going to be in-game when you send it out the triangulation you can get all screwed up it and not quite be the way you want so when I wanted shading like this it'd be really really hard to get because I have to flatten out all my surfaces but what I realized at one point was I could remove edges and just have big horrible

hideous and Gon's of varying shapes and dimensions and it would render like a flat surface because for whatever reason in Maya and in game that would just render like a flat surface and none of the in-between triangles would get rendered or calculated in the normals so big flat surfaces could be with crazy actual geometry if you look at it from the side but for me in the end result it would look the same as if I'd done a lot more work than I actually had to plus any edits to this become a lot more manageable and things don't have to stay flat and be difficult on top of that I started to discover something that was really fun and interesting that I didn't expect which was basically 3d pixel art the art that I'm making doesn't get actually converted into pixels and I don't get to paint over any pixels but the 3d art basically will only take up about 36 pixels on screen in some cases these are small little cabbages and turnips and flowers that'll exist in the game and I had to make them to fit really small on screen

so they had to be really broad and big and bold and being able to have some of the tools to be able to see that it makes my job a lot easier in reality the models are awful they do not look good up close they're janky and they've got like holes between them but because they don't rotate I can get away with a lot and I don't need to necessarily get super Precious about the kind of art that I'm making but I can start to find the fact that they only take up a small number of pixels on screen really interesting in a strange way that I would not have expected or thought that I would enjoy in-game you can barely see them they're tiny little objects right here in the center where they don't take up a lot of space but their shapes have to be big and bold bolder than I'd ever dealt with before and way simpler than I'd ever have to dealt with ever have to have dealt with on top of that finding new challenges and stepping out of what I thought was my intended role for me coming onto the project it felt like okay Dan you're the guy that's supposed to add 3d to the game you're an experienced three guy so you better be good at it but at certain

points we'd start to hit some roadblocks things that that needed some extra thought or or care or just specific extra help such as this area I'm not showing anything else around it but you saw it in the trailer and this was an area that I worked on that I had a lot of design input on where we were trying to figure out what the the setup and mechanics would be for this area and we weren't quite sure where to go with it so I basically just requested to take it on and handle it so the enemies their design and most of the things that involved in whatever's around here is is a lot of my work in a way that I would not have expected and was extremely satisfying to do to just kind of step out of what I thought was my role even to the point of in many cases creating excel sheets which I don't like but I start to learn some really interesting things from doing them in a totally different way that may be completely inappropriate for my career maybe but is fun and interesting and

satisfying to learn when you kind of embrace jumping into something else on top of that our procedural generation works for certain areas but doesn't in others so for caves and areas like that where it's organic some of the way our object is objects get placed works but for places like this where we need structure totally random generation doesn't necessarily give us the results we want this area looks a little bit too barren so my first thought was okay well make more stuff walls and things like that when it generates in-game though it ends up being kind of like a cluttered room as opposed to something that looked like it once had structure and so in that case I end up trying to go okay well like without just coming up with new generation rules cuz that's really gonna set us back in other areas what can I do and in this case just taking large sections of generated areas and just hand modeling at large chunks of them so that we get that structure so that things look like something makes sense takes quite a bit more work something that I would normally say no like we should do it smarter but sometimes

smarts not interesting and sometimes smart isn't always smart it's just fast on top of that training the team an extremely satisfying thing that I was able to do this is a checklist of the things that we have to do when putting something into the game through Maya it's about 10 steps used to be 20 but each one of them at least we can check off to help make sure that the team when they're trying to learn a new piece of software and learn how to get stuff in they don't forget one of the integral steps because if you miss one it won't show up and those kind of things were helpful they help me better understand what our problem points were and they help the team better be able to integrate into a new process as well something I started getting really used to was creating videos where basically I could just record an entire video of what the process was narrate and just talk through it so that when our guys were jumping into 3d for the first time that they could understand what the steps were and how to best understand how to get stuff into the game and have it look appropriate for style the thing I

started to really discover was to embrace some of that unpleasant discomfort those moments of failure and those things where you feel like you're not doing what you should be doing or even doing anything worthwhile one thing that I found actually while I was doing this talk was this TED talk on how frustration can make us more creative and Ted Harford covers some really interesting scientific studies on people dealing with frustration and how that can lead to better results for example study was done where a group of friends would have to solve some puzzles multiple-choice in this case and they could work on it together and they did pretty good and then another group would have to deal with a few friends but also some strangers in that group and what ended up happening was they did much better much better than the group of just friends the problem was the group of friends felt like they did a really good job they were amped up about it they played they thought they did great they didn't the other group thought that they did poorly but actually did great which was something that I was feeling through the entire process on below which was not

feeling like I'm actually doing something beneficial or useful but that might not always be the case the thing I started to find was to avoid my own box the box I create around my process is the things that I think are the right way to do stuff and sometimes just stepping out of that into something bigger allows you to make something interesting even if it feels like you're doing something that you thought was stupid or thought makes no sense because it's just going to lead to problems later means that you're making something maybe interesting sometimes you've got to find that optimism that little bit that's gonna give you a reason to keep doing it and just rush into a naive idea because below and many of the games that cabbie makes are beautiful and they have a beautiful innocence to them where they don't care that maybe something might be crazy because sometimes that craziness makes that makes things much more interesting thank you

I I'm at I'm at 29 minutes so I don't think there's time for questions are there if there's if there's a question up there okay yeah so so we're done but if you want I can go to the wrap-up room or or we can chat here before the next talk starts yeah thank you very much you