Change Is Good: The Importance of Iteration Within a Character Art Pipeline

[Music] good to go alright so I've been told to tell you guys before I start to turn off any noise making devices this is being recorded so anything that's not distracting from you know the presentation would be a would be great also after the talk if we could fill out your evaluation forms that's that's really helpful so please do that so first I just want to thank all of you for coming to my presentation it's a huge milestone for me and something I've wanted to do for a very long time and my apologies if I stumble over a few words I'm more than a little nervous here there will be a time reserved after the talk forum about to show you if you have any extra questions we're going to try to reserve time at the end of the

talk the room that we'll be in is room 3000 which I believe is right across the the hall let me start by introducing myself so for those that don't know me my name is Gavin golden and the lead character artist for insomniac games working on our upcoming Xbox one title sunset overdrive prior to this I was the lead character artist on BioShock Infinite for rational games and it's following DLC burial at sea while working on these projects I was mainly responsible for the look and feel of our characters in the game managing a small team developing the maintaining best practices communicating with other departments deliver what they needed on time working on this project taught me a lot of lessons one of these lessons that I truly believe in is the importance of iteration during development I feel that artists and especially younger artists are told told earlier in the career not

to get married to their work but aren't really given working examples why they shouldn't be in this talk I'll be using my team's work on BioShock Infinite as an example showing where we came from and we we ended up how had settled on the first idea that we had it would have looked like a totally different game and probably not has done so well I'll share the struggles of development the process is taken and the idea is developed along the way and how we ended up with our final product it's gonna work this thing Greg all right during this talk I'll be coming the process of iteration using examples from three different departments how their needs changed over time and how it required us is a character our team to change narrative the storyline of our game the driving force behind the final decisions made overall art direction aesthetic choices made an environment side that influenced our character's appearance and the compact design more intense than any

game irrational had ever made most importantly the lesson that I learned from this project hoping to show with these examples is a need to embrace the organic nature of change during game development not everything can be planned four features had cut once they are tried out and new features added entire directions can change and is our job as production artists to adapt those change and deliver high quality assets before we get into these examples though I want to get some backstory this is how about Jacque Infinite started there was a dark gloomy world with an Art Nouveau inspiration this is essentially a flying rapture low visibility populated by mutants and fantastical machines with a touch of horror these worlds were so similar that the development team actually started referring to the worlds flap sure eventually we felt they were not pushing ourselves far enough that the game would feel like a Bioshock 3 and not a BioShock Infinite which is a

totally new game and had plenty of fresh ideas and a unique storyline here you can see a great comparison showing what Emporia which is one of the final levels in BioShock Infinite originally looked like it's filled with carnival imagery harsh lighting and heavy flow of fog it was a totally different world and one that definitely belong more on Bioshock 1 this is another example of how extremely different world was this is a shot a level that would eventually be dubbed town center I was just an average environment for the game we intended to introduce the the player to our setting you can see in these images we were holding on to a Bioshock one routes into rapture in a lot of ways we were holding back and just playing it safe it would have been easy for us to make this game as a lot of the team had already made it years earlier we wanted to do more we want to embrace the bright blue sky and the idea of replacing New Year's Eve with the fourth of July this is the town center

that everyone has seen it's an impressive reveal to a city in the sky above those thick clouds from previous shots we're praising replacing clowns for founding father statues and following the idea of having horror and broad daylight it was a huge departure an idea driven by our creative director Ken Levine but embraced by the entire team it was fresh it was a game we needed to make and we knew that as salt as soon as we saw the changes side by side it felt new it felt we were no longer making just a sequel but a game that would stand out on its own and more importantly stand out from other Bioshock titles even though it was the right decision that doesn't mean it was easy at all a changes big came well into development after lots of assets had been created and caused all departments to rethink how they were working and what their end goal would be initially there was some hard feelings lots of cool assets were getting lost and everyone loved the feeling of Bioshock 1 but it was hard to

argue with the new direction it was more ambitious it was refreshing and it challenged us to move outside of a comfort zone and a major change came with Elizabeth Elizabeth is arguably the game's main character and was the entire story of BioShock Infinite surrounds she didn't seem human she had to show a range of emotion and the player had to want to help her she needed to be interesting and not feel like a burden during gameplay she was a major part of the story and not a friend and not just a friend that you had to protect from point A to B she wasn't just an escort mission without question Elizabeth became the backbone of our game and one of the major design pillars during development that was always the case here you can see all of the different phases of Elizabeth many of which show up at some point during her game but a lot of versions that were completely there completely finished that we just had a scrap because they

weren't hitting the mark both from a narrative and aesthetic point of view in this section we will touch on all these different versions and cover the choices made along the way to go from Gibson Girl to femme fatale like many game companies we start from a piece of concept art ranging everywhere from thumbnail sketches to fully rendered illustrations our concept artists try to convey the story we want to tell in the form of guidelines from modelers just to get approval before jumping into deep into production I'm building a character in this case our concept artist tried many different angles and the results a more classical look for a companion character inspired by a Victorian era outfits our team went to putting this version of production of Elizabeth who at this time didn't even have a name I'm into production model came out great but her purpose had entirely changed the decision was made to have Elizabeth speak and though that may seem seem

obvious now she was originally intended to be a mute a quiet mysterious character that would only show up during key moments in the game it became clear that this wasn't engaging to players this didn't help us tell a story because well she didn't say anything without having that dialogue it was extremely difficult to inform players and to have them interested in helping Elizabeth along our journey this required us to go back to the drawing board our story had evolved into a fairy tale like arc where the play would rescue a young girl from a tower who had been there since birth well the purpose of that mission was dark and surrounding mature content sorry so I can surround mature content the character needed to seem innocent she didn't seem like your help was needed and that you as a player wanted save her this man dropped the formal gown putting her hair down and making her seem young and aloof to the world outside of her prison we

started to embrace the idea that Elizabeth was completely unaware of what acceptable fashion was wearing a nightgown and boots with a roughly cut hairstyle that she did herself this implied that she wasn't used to being around other people and certainly wasn't expecting company the problem we realized though as the model was complete and functioning in game is that none of that information was obvious it didn't read at all it was too obscure out of context she looks like she could belong in any time period an outfit too generic to speak to a specific time period not detailed enough to stand out as a major character when put into a crowd Elizabeth would just disappear another thing that we had to consider is that Elizabeth was now our main character she was the face where we were doing she'd be on posters she'd be on game covers and hopefully be recreated in other forms of media she was the face of her game and we had to

ask the question do we really want the face for a game to be a girl and a nightgown the answer was no we needed to push harder and create something that stood out an outfit that was more iconic and unique to our title we had to give up details from our older versions and just built her outfit completely from scratch this tilt back and forth from our concept artists and modelers trading paint overs and feedback to really nail something special at this stage we had also gotten more comfortable with our tools and we're ready to give Elizabeth more polish this meant developing technology just for her our team created custom shaders implemented a prett a px cloth authored unique textures and took multiple passes on an animated rig to literally make her one of a kind these changes also came with a pressure for our up-and-coming reveal a video of our game would be seen by the public for the very first time and it became all hands

on deck to make sure that it was impressive Elizabeth Elizabeth would be front and center showing off her ability and giving a fairly good example of how you would interact with her during gameplay reacting to minor feedback after the reveal and addressing minor polish that we had been meaning to work on as an art team like revising her facial features costume tailoring and her hair rendering we had landed on her final Elizabeth as you see like a like I had mentioned before she broke all of the rules and had a budget that blew most of our assets out of the water a poly count almost three times higher than the average character in BioShock Infinite skin materials and hair materials seen on no one else and are tailored specifically to ideal viewing conditions just for her a completely unique rig for her character with a facial rig having cinematic controls and at least double the amount of draw calls is everyone else in the game having

these specs on every character in the game would have had our rendering engineers hating us but for Elizabeth it was necessary she needed to stand out she needed to be the best looking asset in the game in the end assets were tailored around her to ensure that she looked ideal lighting was was revised to make her look good and budgets were adjusted around her just to make room for her asset and when you compare what the final Elizabeth looks like compared to her initial burden even in earlier test cases it is a night and day difference it took lots of days from different departments to get her just right and it shows embracing changes that needed to be made to create the best end result once we had settled on our Elizabeth model we felt that she was strangled way from the original idea that she was an innocent person trapped in a tower the corset version of Elizabeth had all the notes that we needed - for later in the game but wasn't fitting the first

impression that we wanted to give we had to show that the character progression throughout the game and sell that the story a sell the story that the Elizabeth that you meet at the beginning of the game isn't the same Elizabeth that you leave at the end of the game we decided to have different outfits for represent different stages of her evolution as a character the first outfit would push the shoes considered young and an innocent person who is unaware of the outside world a concept artist referenced children's uniforms at that time to help show the ideas and we put into practice print a brand-new model this led to another problem that couldn't be predicted with level shifting around in the game area is being cut or shortened and key story moments being tightened tighten throughout the game we realized that the Elizabeth in the corset the one that we had spent so much time developing and perfecting was slowly taking a back seat and turned in terms of screen time the younger elizabeth is now becoming a more major asset and not

just something brief that you see in the beginning of the game with more attention and focus this time we needed to spend more effort into making her asset more appealing the initial take while accurate to the clothing of the period was looking too frumpy and almost too childish after trying different color combinations and passes on trying to improve the fidelity of the model we needed to to do a complete overhaul around this time Kennett seen the redesigns of Disney Princess is done by an artist named Claire Hummel we had always considered it is need to be a great reference for Elizabeth and one of the capture that same sense of wonder and the designs that Claire had done so we just contracted her to to create a design for our younger Elizabeth this this new design had brought Elizabeth to a more relatable age had a cleaner color scheme and still felt within our turn-of-the-century time period for a clothing with our Elizabeth assets

locked down we could focus on taking her further in other ways we created damaged states in each of our models that would be swapped in throughout the game this helped tell a smaller story and keep things fresh and interesting and show a minor progression with her changes through modeling textures it caused by things like like major combats in the game crawling through vents or over rubble we made these model swaps whenever Elizabeth went off camera essentially desponding one version of her and replacing her with the new version it helped make Elizabeth feel more alive and was it clearly fairly quick thing to create that really went a long way another level of polish that we added tool is their ability to spot things throughout the game play custom animations and comment on them on the things that you saw in the world this also involved emotional animations they'll be added to our animation sets which helped make Liz seem like she was reacting to the world around her finally

after we had shipped infinite it was time to look at DLC we wanted to deliver something that I've been requested by fans of the franchise and elaborate more on an alternate rapture that had been explored in previous titles this meant converting the roles of our characters from infinite and to something more fitting for rapture namely creating a femme fatale version of Elizabeth that would be right at home and the move in movies and Pulp Fiction of that time in many ways the dlc elizabeth is a more compressed version of the lessons we learned in the core game that you will see in the next few slides we completely started from scratch and had a healthy pre-production for a time for approach of that size our concept artist George lacera referenced a lot of fashion from the time period and took stabs at different takes on a femme fatale for Bioshock bringing a narrative animation and modeling into the conversation we wanted to settle on something that was both readable as Elizabeth but also reused her Reagan

animations from the core game after deciding on a concept we move we move forward of creating a new version of Elizabeth once you put her into the game we realized that much like the younger version of Elizabeth like from before she just wasn't looking right we took the idea of her being clean dressed in white and taking lessons of makeup and lighting from that time a little too seriously after trying a few different tweaks trying to polish the textures and pushing the model around I just decided to start over essentially there were just too many small things that just built up into a complete rework it would have taken longer to rework the character to address all the feedback rather than just recreating a costume from ground up at this stage we kind of went rogue the concept artist myself and can work together at my desk to speed up iteration time and to get all the features that Ken had requested into the

asset as well as addressing art related bugs that had been bothering us things like how her hair was rendering and sorting we updated her eye textures to introduce new new parallax rendering the cut of her dress how she deformed her a predefined rig and things like that we decided that it was quicker just to work at the modeling stage and review it in game since we were basically trying just to hit that last 10% the concept of the character was correct it was just a matter of execution this meant a lots of color choices lots of different tags on accessories and generally lots of iteration on smaller details in the model lots and lots of iteration so many color choices ranging from Scarlett Elizabeth in trying to reuse the color scheme from the core game and attempts to attempts to make different accessories to give her some kind of points of interest here you see what the final product is that we we felt as a

team really nailed the feeling of a femme fatale Elizabeth was a nice translation of a popular character from our core game into a different time period a different setting that was familiar to our fan base in this next section I'll be covering how a massive change in art direction caused us to rethink how our general population caused us to rethink our general population and how embracing a standard a standardized system allowed a very small team to develop assets quickly and consistently making better friends as her art direction involved it required us to take a hard look at our general population and how they were fitting into our changing world there was always a complete facelift bringing our heavily stylized characters from an Art Nouveau setting to a more relatable period-appropriate look that you can see in the final game using examples that span over four years of development oh sure how these changes how changes such

as these gave us more ground characters in the world more refuted a more repeatable population and a more consistent art style first I'm gonna take a drink though if you don't mind the main issue we face when deciding to move from a darker Victorian style environment to a to a brighter turn-of-the-century setting is that our characters no longer fit both the tone of the game and overall visual fidelity given how long the project was in development gradually over time it was very clear that assets that were created at the beginning of the project and how they look compared to the ones created without a year or two after changing our games setting required a complete overhaul on our characters to take them from a mutated appearance there's something more relatable we needed to reinvent our characters to properly fit

their new home and not see what they came from entirely different world well it's always fun to make monsters and we were happy with the work that we had originally created as a team we decided to start over and recreate every asset to support the driving art direction on the environment side originally we followed a model similar to Bioshock 1 where all civilians were actually aggressive characters who would either attack you on sight or just be doing crazy things like eating trash and swatting at flies that didn't exist artistically they were very similar to splicers and followed a much more stylized appearance with exaggerated exaggerated proportions and overly expressive faces with the change in direction it was also decided and became pretty apparent that this world would have many repeated assets we were actually creating a population for them

to live in the world and using faces like these whatever had obvious repetition it was very easy to pick out features on certain characters as you can see up here and that was a problem that we wanted to avoid creating assets in this way also meant that artists were free to model in texture however they wanted as long as the final results looked good the final result artistically was interesting and it did look good but it meant that every asset was created uniquely which led to inconsistent poly counts UV layouts and texturing everything from scratch with the workload ahead of us and really only having three team members we knew this was not sustainable these all assets also had unnatural proportions that was more fitting in our stylized world than our more realistic one that would eventually be Colombia our textures were also authored in a much more painterly way and the overall tailoring of our outfits were both too damaged and not

period appropriate clothing we really wanted to move away from our civilians looking crazy and one of them to appear like people in the photos from that time much like the problems that we experienced where those civilian heads the the bodies were also very inconsistent dropping our assets into a crowd of the textures we created were so distinct that it became very hard to repeat the characters the color choices the patterns and types of models were too obvious and even if they looked it on their own had very obvious repetition this method of working also left us with no clean way of kid bashing assets every asset was essentially created uniquely and it become very difficult to reuse any amount of work from one character to another we started making baby steps in terms of quality to match the environments being built in game this man building hired detailed sculpts mo more photorealistic textures I'm

bringing the proportions to a more realistic space these baby steps were in the right direction but we really needed to push it further this is an example of our general population at a time generally characters were still looking very distinct they were still too stylized and it was very easy to pick out specific heads and quality wise we still had older assets that were just throwing on for overall look spawning characters in a test level during an art review and having one incredibly mutated face show up every now and then see more like a bug than an artistic choice we're also soft wearing similar problems with our bodies some characters were still wearing ill-fitting dirty clothes and the costumes were very very easy to pick out in the crowd these also became distracting environment shot games the environment is generally an explorable area that we

wanted players to experience and having characters stand out too much just doing too much attention rather than having them fade into the background so is becoming a pretty big problem the solution of these problems was to create a universal system this meant creating a base character that all future characters would be derived from our heads had shared topology UVs and facial rigs this allowed us to rig heads in more or less one click of a button share elements of textures easily and were use elements of the models from one character to another by establishing these constants throughout the game it was easier to make swooping changes and to control the overall quality of our assets in game this workflow had us working mostly in ZBrush by having a sculpt that we could pull from that had our universal third order and UV layout we could basically drop it draw it down to the lowest subdivision level press a

button and have it raid read a set ready to go one unreal pretty quickly by having a shared asset that we pulled from also allowed us to work much quicker and also establish an expected final result that was easy to compare new assets to rather than taking a shot in the dark and trying to have assets match a loosely defined style we had a very strong foundation that served as a quality bar artistically and technically it set a standard by changing to his workflow it also removed most of our tedious work from our day to day jobs by not needing to unwrap every head create every low poly model and create a unique facial rig for every head we could focus on making art and pushing the quality of our work by establishing rules it actually gave us more freedom defining restrictions what we couldn't couldn't do it allowed us to focus on quality and keep always under the hood consistent as a rule expanded and certainly the scenes

and the game were developed the need for larger crowds became stronger we needed cheap assets to fill up areas our general population assets were correct and looking great but the amount of joints verts and loaded textures on screen were too much for consoles to handle the solution was to create cheaper versions of these assets that had limitations essentially making much lower vertex counts heads and bodies baked down to just one material and a heavily reduced rig and no artificial intelligence we took to calling these characters chomps a really dumbed down version of our main character assets and these were these assets were surface transfers and basically an LOD of our major asset over time though these characters were put more into the spotlight and their quality just wasn't holding up at all and to give us one last push to reach a desired quality level a strike team was formed to tackle a large chunk of our

town center level and bring it to a final version this involved all disciplines that aren't working together on a specific area a character concept environment art banded together to set an exemplar for what following assets would live up to in this area our crowd assets are chomps were a big focus for the the character Department to improve the quality of our assets we develop sets of characters per level rather than cramming a bunch of characters onto a retrofitted retrofitted texture sheet like before we created characters in a modular way with the intention that it would be repeated a lot throughout the game these assets became easier to manage as they were just fewer pieces to update and generally cheaper because they were purposely made to share single assets across at least four characters another friendly character or characters

that came up during development where the loo test wins it's interesting to think that these characters started off is just a name just a brand and something that you never actually saw but it turned out to be one of the most popular characters in the game the original direction for these characters is that they were identical they were literally the same asset but one had a beard and the other one was in a dress obviously this is a horrible idea it was one of those ideas that seemed fine on paper but just look bizarre in practice the information wasn't reading and said just gave us two characters that looked off basically with the importance of this character increasing and wanting to move this assets and the quality of our main characters we needed another pass we decided to contact Claire Hummel again since she did such an amazing job on Elizabeth we wanted to apply the same lessons we learned there and applied to

these set of characters and it was a huge success the new version took a more simplified costume made bolder choices with the tailoring and had great similarities that implied these characters belong together and finally to ensure that we delivered a game with the highest quality possible-- I booked at least a month of Polish time from my team to get one final pass on the characters in the game over time and like I mentioned before we developed techniques that improve her assets and left some assets needing to be brought up to a desired quality level we had developed techniques for baking in subtle lighting information corrected topology and weighting methods that gave us better deformations having his time allowed us to make sure that every asset had this exact same treatment and now some pretty staggering before-and-after images what are care of our characters most of these are from our initial push to drop stylization compared to the

final asset so it's about about a two-year difference here you can see one of our Asian civilians that was part of general population and this is one of our Caucasian civilians seen in Colombia many of our civilians from the Finken area which was shown in a teaser video were pushed on with too far and - looking comedic leap or eventually we pulled back and made this information more subtle and this after it shows the the process from beginning to end one of our dockworker assets oddly enough this is a the first and last character that I worked on the project having started with a burly start up character that involved to a more period appropriate this repeatable dockworker that properly fit into Colombia and finally an example

of how one of our female civilians changed from a more fantastical unrealistic outfit into something much more cleaner and period-appropriate for a lot of our civilian outfits we actually looked towards the 1920s for tailoring even though it was out of our time period the style of the 20s were creeping into the late 1910s this meant that better fitting clothes and more interesting colors and patterns and here you can see us going from a completely made-up outfit on the left to something more Victorian and then something more Edwardian I'm in this last section we cover my my favorite characters from BioShock Infinite is the the Vox Populi and the founders combat AI at a rational we we greatly improved the appearance of our combat AI characters from Bioshock 1 to BioShock Infinite after collaborating with our designers we were forced to rethink how we handle

character art and a combat scenario this often required new models re-evaluating silhouette the importance of color between factions Anna tiered armor system unique to BioShock Infinite initially there were no factions our characters were literally spot splicers and a Victorian and they were regular torn up clothes they were heavily mutated the first concept as you see here is that everyone was against you so every character you met was trying to kill you as we populated the world with regular people and the story behind Colombia and its struggle developed we really needed to split who was trying to kill you from who just wanted to talk this force us to started thinking about who enforced the rules of Colombia who is trying to stop you on your mission and how to break them away from the crowd

the combat system in BioShock Infinite challenged our team the way no other project before it had though you still had plasmids and guns under the hood it was a completely different game and BioShock Infinite we had actual roles within factions and both needed to be defined you needed to be able to pick out the shotgun guy from the machine gun guy and they needed to stand out from civilians that you couldn't easily detect a threat infinite also had huge draw distances that required us to create characters that were more noticeable from a distance and given how many different types of characters were on-screen required to share assets and overall proportions initially we followed the same system as as Bioshock 1 had and tried to apply the mindset to these requirements for BioShock Infinite we found that the idea of anyone being anyone is horribly flawed and just confusing if any character can pull out

a gun and kill you not only is it confusing but you're constantly on edge and never really told when the combat begins or ends which really fight against the idea of us wanting you to explore the city of Columbia technically we also tried following the same path as Bioshock 1 by manually placing each and every character as a unique asset placing these characters also became very tedious and hard to iterate on essentially every spawner needed to be manually placed once that spawner was placed and our designer we need to access its properties and point it to the correct skeletal mesh while this may have worked on Bioshock 1 it quickly became too much to handle an infinite it was nearly impossible to control what assets showed up where where assets were being referenced and left a ton of room for human error to make this process easier we developed an override system the

override was assigned per asset and had a drop-down box that had a list of options these options were defined by artists that reference the list of characters by doing this designers could just select what kind of character they wanted it to be and we as artists populated what the character would look like this means we we only needed to update one asset and it updated throughout the game rather than making sure each spawner was referencing the correct asset this also made randomization automatic if there was a more than one option in the list rather than having each spawner manually tweak to be different from the one next to it the problem that we that this exposed those that that unreal store is all of the model and texture information into memory before choosing which one to render meaning the base match that we were originally trying to to replay or that we had on the spawner and would be

replacing restored into memory and then the one replacing it and all of its alternative options were also stored into memory together this made one combat character cost nearly as much as a main character the solution to this problem was the wrong box every base mash in the game is a giant pink cube with a placeholder texture that costs virtually nothing here's the same rig is all of our base humans in order to bring in animations since there was a requirement for the actual override but other than that is practically free this meant that all we were ever paying for is the asset being referenced in the override if an override wasn't present the giant box would appear and letting you know that something had gone horribly wrong and that an override needed to be assigned our combat characters weren't just bullet sponges they were also a big part in telling a story of Columbia this world wasn't just civilians with with

people who had a chip on their shoulder they needed a military force to protect them and the ideals of Comstock the city's leader even though they shared the same ring as our civilians having a uniform that they wore as well as having a unique headset there was only ever used on aggressive characters really helped us define what a threat would look like in the game another challenge we faced on infinite that was completely new to our team was the introduction of a second faction early on we had known that there would be someone else that you fight but had no idea what that would be exactly we asked ourselves who these characters were and the answer was a faction that rebels against the founders and symbolizes the voice of the working class was the Vox Populi much like the founders they had to stand out from the civilians in the area rather than creating a uniform for them though we decked him out and military gear from the time and used color as a visual

language for the factions that you were fighting blue generally symbolized the the founders faction and red symbolized a Vox Populi by using colors in the way in this way and keeping them consistent it helped to inform the player who was a threat and who was not so for example a civilian would never be wearing a red shirt and were generally kept away from those colors and left in a more neutral neutral area during the completion of infinite we push really hard as a team to deliver the product that we that we wanted to in this push we had lots of feedback from focus testing and decided that we really needed to push for Kombat characters further we needed to show progression more clearly and to find tougher enemies from the weaker ones working directly with with my art director we created new combat characters in a free-form way time was definitely a factor at this point so we

just worked at the model level looking at elements that have worked on older passes on the models and also wanted to preserve ones from the higher level characters there was always to create three tiers per combat role per faction due to how many of these characters are on screen and given how pressed for time we were we ended up sharing a lot of elements between characters using unreal attachment system we added fully rigged armor that was built over a base model and unique ahead attachments per tier since you were only ever adding attachments the major cost came from from the base mesh and each armor set was added on top of that rather than loading in and rendering three completely unique body assets here you can see working examples of this system in place it's three completely unique assets that heavily relied on kit bashing and the sharing of assets to speed up workflow and you can see the

opposing faction using a similar process just in their own style after exploring the Vox Populi more we were able to really have fun with their designs these characters were contrasted against the uniform standardized nature of the founders and were able to create their own costumes using the idea that these characters fought their way up to the top and purposely wanted to be scary and violent we covered them in red paint gave them devil horns and more torn up elements to their outfits not only did it look cool it also really helps show the difference between the two factions another element to our combat system was the idea of heavy hitters these characters way back in the dark cloudy flap sure setting were originally just going to be flying Big Daddies well that is still a great idea it fit more with what we wanted songbird to be these characters needed to be obvious threats in the combat scenario sometimes towering over a crowd and easy

to identify these elements also had their own rule set and unique versions per faction their own rigs and higher budgets they were more fantastical and we're designed to purposely be action which man has asked if we would want this character sitting on our desk unless we learned during development was to take a step back and look at every character we have in a line does it show clear progression can you easily identify one character from another this led us to create what we were calling a dinosaur chart I was a huge image that we actually had printed and posted in the office right on my desk that showed every enemy you would ever encounter in the game and how they stacked up against one another this allowed us to really have a bird's-eye view and not work in a bubble after having our heads down for so long it was easy to focus on characters in a specific area rather than seeing the big picture by doing this it also helped us pick out problems across the board and tweak

assets and so they all reach a consistent quality level he's actually shots you can see the the end result of a lot of iteration on our characters imagine how the game would have looked we had stuck with our original plan you can see characters that fit properly into the world that we had created more relatable characters that seem like they had chosen to defend against you rather than just being plain old crazy clearly defined threats with specific roles in combat and tier level difficulty throughout the game and the second faction of enemies that we that were clearly different from the characters you encountered earlier in the game I took iteration multiple passes and a hard look at what we had in a game to make it right and sometimes tough painful choices I meant throwing away work that we had done in some cases were happy with and proud of for the greater good of the project and the end game

development is a team sport it isn't about personal statements and what individual assets you can point to and say mine it's about delivering the best possible product as a whole fulfilling that fulfilling the needs of the team with high quality work even if that means changing the work that you personally loved I don't know these examples prove that that we as artists cannot be married to our work and the ability a doubt to change and drop any personal attachment and Bracey organic growth of development over time as a necessary step in creating the best product possible not only for a triple-a development but for indie developers student projects and personal work as well be honest with yourself look at how your work contributes to the game as a whole and put the needs of the project before your own personal preference after reviewing these examples I want you to go back to work about your school projects personal projects or mods if that's even still a thing well with these lessons in mind you will lose work but you're not

wasting your time just because your own work ends up getting thrown away it doesn't mean it was a complete failure even though thrown away work is contributing to a bigger better and result not every contribution needs to be in the shipping game and no one expects that all of your work will be perfect to be blunt get used to it it happens to everyone on every project no matter what the size on BioShock Infinite the estimated the estimate is that we created around three games worth of content and that would that could sting but the end result is something something better because of it also making games is a team sport never forget this this is possibly one of the most beautiful things about our industry there are many moving parts with all sorts of disciplines moving at the same time racing to the final goal but we're doing it together other departments rely on you this as much as you apply it and rely on them and it's important to respect that level of communication and expect back and forth between

departments account for change in quality over time every project from the smallest personal project to the largest triple a blockbuster will have a range of quality in it at the beginning of the project you're getting the feel for it and you want to you want to see what works as you get better and everything is moving around you the are that you did monster years ago will look worse you can try to polish it as you see it but by doing that you'll either spin your wheels or end up in the exact same situation months down the road block out time for polish give yourself and your team one last chance to send out their work out the door and be proud of it finally great how great ideas happen naturally during development and it's very common for fun things that happen naturally through play testing and experimentation fun is what makes a great game is our job as production artists to try and let those things happen we are in a creative industry it would be incredibly boring and tedious if you sat down five years ago and

planned out every detail be patient and be part of the upcoming changes encourage your team to be creative and enable them to let awesome things happen so when the combat designer comes to you and says that the player will be ripping off heads with a giant hook you put in the extra time to make some meaty bits and I am done so thank you so you guys gonna like line up for questions if you have them actually have a few infinite steam codes for the first two people to ask me questions willing to just hand them out just come up to me after the talk and also having just wan a visual arts award that's one of the points you to the team that's responsible for it this would be all of

our Twitter using art members from BioShock Infinite so be sure to hit them up and aggravate them for me so when you guys outsource to Claire was she just a concept or did she do the concept and the 3d for the new characters like Elizabeth and the twins she just did the the concept art she was like a remote artist that did combat designs and basically went through the whole process with her so like starting with like thumbnail sketches and then like two final like rendered things she actually went a little bit overboard with it and did like more breakdowns on how cos he was from that time worked and I really hope like are our models but we did it all we did like a modeling oh cool thank you can you talk at all about any kind of iteration in terms of shader and material stuff considering your original

art style is so different in the lighting situation that you ended up with yeah we actually originally started with just like an out of the box like unreal shade around Unreal 3 and eventually convert it into a physically based rendering setup so we actually like completely dotted our entire material break material breakdown and it's something that's more like like the metalness set up that you've probably seen online but before that it was a good traditional you know spec TIFF use normal kind of thing and like moved over to like PBR about about like a year and a half before finishing so yeah what would you do differently what's that what would you do differently if you had to do it over again um to speed up the iteration the speed-up iteration I think I would have I think I would have put in the standardized system a lot earlier I think even with like the heavily stylized setting we still could have done it and still had you know like that

kind of look but with this the more constant thing that you could predict I think we held on to like unique asset it's like a little bit too too long that uh that caused problems that we could have avoided did you notice there was a lot of burnout because of all the iteration no I mean you know we're all kind of on board for it you know like redoing like main characters over and over again it can get to you but uh but you know it's for the greater good and we all we're all sorry I got like a light ray in my face because like it's really all a roll up put on board for it and all of like the decisions were like communicated and it made it made sense and in the end you know it's like a hard to argue with right thank you yep given the themes in like the time period of the game when you guys were approaching design of some of the more like racist or like hateful imagery that

you have to sort of tone that back and keep control or like how did you approach that well I mean you know like a lot of games were generally we're made up of a group of people from different backgrounds and stuff like that so you know we all had like internal feedback on whether or not it was a good idea to go forward or not we have there were some things that we pulled back on that you know hasn't been revealed and things like that I was some original ideas that I'm not gonna say here but you know there are some things that you just try it and just you know totally if it's working or not or it's like wow that's too much you know you got to pull back I have examples but I'm on camera so I'm not going to say that Thanks oh hey hi I was wondering with all the iterations in the art you made when you're planning the project in time did you go

Oh operate you know like wait you were altering my English is not that you know it's fine and in this time schedule with all the iterations you were over time oh well we ended up doing like one short the end of the project but I wouldn't say that it was because of iteration it's just a more of a you know I mean new characters came online that we had to push for and I push you for like overall quality but not not through iteration we generally scheduled for that so if you're really making something I said well we've got to rework lids I'm just gonna block out like two weeks to hang out another Liz basically okay thank you okay okay so with the shared facial rig did you find any words limited creatively in making different code like different characters or well I mean for us slow generally like they shared facial rig was for all of like our human argument character so

there's only so far that you go right and like everything was you know like generally shared over like male like male and female kind of things so we didn't we didn't really go too far like too far out we we probably could have had more constants so like our eye positions and stuff like that I probably could have been you know consistent throughout people and things like that it wasn't like super limiting just because of like our hardstyle wasn't wasn't wild like you didn't have a guy with a Cheyenne giant head versus like super skinny had kind of thing and anything that required like I knew like a like Liz basically had like her own facial rig and like Daisy had her own and stuff like that so anything I'm required if we ended up breaking off but they were usually like anything I was out of the systems anything I was like a hero character or like super special good good that treatment so you think was helpful that you had a limitation so you wouldn't go yeah yeah I mean I think at first there might be some some

whining I guess about like you know having restrictions put on you but want you to find those guidelines you actually have a more freedom to work within them right so you have like an established you have an established like budget established like workflow final you know look to your art and stuff and within that you can just like focus on making art versus like I'm gonna try that oh that didn't work I just wasted time now I'm working extra you any man so it's kind of like freedom freedom through restriction I suppose cool thank you thank you all right I think all right go for it while you were working on the characters they were being changes made to the story and I guess new important characters being added did you find that that was like a real challenge no we generally like we would have like script changes but I mean just the way

that I would try to have what the team working is that they would be working on things that like if the script change it would be weeks down the road so trying to avoid you know having somebody working on something and in the day like a day after it's cut or change or something like that so trying to avoid that as much as possible and kind of having like a block in your schedule moving with with writing so it would be like a cascading kind of a kind of deal to prevent that pain thanks so the technical solutions you come up with to overcome some of the production problems what do you start initiating for those changes do you just like sit down with like a production person and say here's a problem and he would start fixing it how do you start yeah I mean generally like the the department heads like in our company would generally be communicating all the time so like me and with the like

character TD like shader artist things like that would be like talking in the same room to make sure that we're all on board and we'd have like actual production people to help us with scheduling to make sure that we weren't dreaming up something that just wasn't feasible so it's like a is pretty like organic I mean it was pretty easy to get changes to on just that we had to rely on the communication a lot so you know if like I ended I wanted to change something I really talked to you like the character Kitty and make sure that you know that what that was possible like adding bones or changing like our Universal weighting and things like that Thanks cool I think that's it all right thank you [Applause]