Larian Cinematics: A Top-Down Look at Our Bottom-Up Approach in 'Baldur's Gate 3'

[Music] okay hello I'm Johnny [Music] Cash uh I'm always Blown Away by the incredible uh support that we get from uh everybody about Boulders Gate 3 um how many show up hands how many of you guys have played it [Applause] thank you so much uh I'm Jason Latino cinematic director at laran Studios I'll try not to talk too fast but I I did not think about how many people would be in this room with me when I was talking um uh uh this will be a a topown look at our bottomup approach to cinematics on Boulders Gate 3 uh I imagine many of you are here to figure out like how we put so many cinematics in the game we went from making a a Divinity original sin 2 with

node cinematics to making all of the cinematics um and we'll get there but first uh things being the way they are in uh this year I want to address uh the layoffs uh I have no great wisdom uh I have uh uh not much to offer on this topic but I will say a bit of optimism I did take this job off of a six-month layoff um I think I uh I I kept track which is maybe a good or a bad thing depending on how you look at it and I think I uh applied for for interviewed for about 67 jobs before getting this one um track your actions anywhere that you apply for uh keep track of that on the date that you you applied for it figure out if you know anybody at that studio if you just put it your application into a web portal um figure out like go on LinkedIn see if there are any recruiters that you can find reach out to them anything that you can make some noise and with the recruiters don't ghost them like some sometimes they're busy people too and they're also working

with hiring managers that are also busy sometimes poking them helps them uh uh poke them you get it um so uh everybody's juggling a lot on with the show back to cinematics um there's a lot of what we do with cinematics that is a like like other cinematic pipelines I'm going to focus on the things that are different about what we do um we still have story we still go into preis for some of our scenes uh and we iterate until it looks uh it looks great and it's we're happy with it so you guys can be happy with it um the format of this talk we're going to we're there'll be a couple Cycles uh it'll take you a while to to catch on maybe um but we're going to talk about what was the initial Vision when we started what did we think it would take to make cinematics um what were the problems that we discovered with those initial ideas and we'll go through a couple of those and then we'll move to a new vision how do we adapt how do we recover from those bumps along the road and and what were the new wrinkles

that were introduced from our new cool clever ideas and then some resolutions uh uh one that'll be for the kind of the project and our a general approach to cinematics and uh kind of a personal note for myself start on a personal note and on a personal note you get a personal sandwich uh the initial Vision We There were kind of four main things when I'm going I went back to look at all my documents from 2019 to figure out like okay what what where was my head at when before this whole thing like shook the world right uh flexible cinematics there's I I describe cinematics uh to a lot of people which was less helpful in Europe it turns out as peanut brittle because I don't like it over there very much it's toffee um it tastes great but it shatters into a million pieces when you try to bend it into a different shape uh so if you take a line that's good here and move it over here suddenly if you just jammed it into your timeline tool you've jammed it into Maya your whole sequence is broken and you kind of have to start over from scratch sometimes so how could we avoid that um

we didn't want to make the game smaller Divinity original sin 2 and that's a mouthful so I'm going to call that Doss 2 for the rest of the presentation had about uh uh 54,000 lines of dialogue that were delivered in what would be considered um uh uh what cinematics would consider an interactive dialogue for bg3 uh there much more writing in the game but like kind of that was how we were determining what did we actually have to support and we didn't want that number to go down I didn't want to be the guy that came on and r ruin larryan games nobody would um Muppet acting so this is I I talk a lot about a lot of weird stuff so Muppet acting there's this weird uh kind of story from I think it's Chuck Jones uh talking about watching The Muppet Show for the first time and he was just like uh angry at how easy it was for Jim Henson to connect with this audience through just uh uh syncopated motions bouncing up and down and flapping the gums at the right times along with the um along with the words and that created uh uh like

infinitely relatable characters of Kermit and fozy the bear Miss Piggy and all of them that we the reason that I'm mentioning that about a video game is we knew that we were going to map our um uh our entire vo pipeline to hide the NPC and this was going to amount to the the largest uh uh pile of Animation that was going to go into our game and if we agonized over art Direct in all of that animation we were going to it was just going to stop us in our tracks so we had to basically figure out like what are we actually looking for what we're looking for is a simulacrum of life we wanted to capture the ID idiosyncrasies of the actors themselves rather than and just worry about do they are they working on the level of a muppet are they a believable person rather than are they the exact right person to deliver this quest line all the time we had to pick and choose our battles and then grounded tone we went wanted to adjust a little

bit of the story book we had still had a narrator um uh that we uh brought over from from dos 2 but it had to work on a personal and cinematic level so this is kind of where the loop starts guys uh flexible cinematics this is our initial victory was this is we came up with scene templates that would stamp out cinematics anywhere and I got this really good advice one time when I was looking at getting into watercolors um uh uh by a guy named Derek Straton amazing artist amazing uh art director he said paint with the broadest brush possible which is NOT saying always paint with a broad brush but all paint with the broadest brush possible whatever the part of the the the job that you're doing use the broadest tool so what was that for cinematics and for us it was basic staging for basic conversations there's only so many ways that two people can uh uh uh stand and talk and look at each other and figuring out how to turn that into an infinite

snowflake of possibilities was not going to be great so we figured out how to templa our basic conversation um uh uh setups this allowed us to also standardize where our cameras were going to be and how they could respond and kind of uh deal with all the height variations if they just had to deal with the height variations of being a half orc or a half Ling um we didn't have to worry about like finding them in the horizontal axis as well and this also established a an inherited authorship we had something where we could change a few template files and then we could watch those changes rippled across the entire game so I I'll explain how this falls into the pipeline we have our dialogue file we have writers they're working on story authorship uh they're writing the lines of dialogue and the Order of the the lines of dialogue and all of the intricate uh uh spiderweb paths through those dialogues and then that we have these scene templates I want to skip over that for a second because this is

where a lot of this is kind of what's different about us the timeline file this is where a lot of people get into uh Maya or directly into uh you know U some sort of in tool cinematic uh thing if you start if you go directly from the dialogue file into that timeline level the blue level everything is brittle everything is bespoke every last moment is custom so what we did is we introduced the scene template took care of character staging and took care of camera setup and it was it gave us a safety net of authorship that we could Bank on uh and it also allowed us to to have a reliable line of action so that we could uh uh organize our actors and how they were standing when they were recording in the booth the timeline file this is where we would start to we kind of have our cinematic team set up in two different teams and it it does take pretty much the entire company to make cinematics but like two main teams we have cinematic artists and they are focused on the time in acting

pacing custom work and implementing uh assets from V effects sound effects and all this kind of stuff uh and bringing in especially the animations from our wonderful cinematic animation team so it's a kind of an Implement implementation timeline and then we have the game itself which places these timeline files into specific areas of FY run that make up the story of Boulders gate initial problems with flexible cinematics top down games don't have horizon line folks uh so our cinematics could play everywhere but not everywhere was designed for viewing at eye level we had short draw distances at first we had one-sided textures odd decoration placement floating scenery and visual scripting asylums where somebody just like said we'll put these over here no one will ever see these over here well if you look at the bottom right we did see them and so we had this game where you could uh talk to all of your companions you could click on them anywhere in the world and suddenly top down game turned into a third person

game we had to figure out what this meant for our art pipelines and our level design uh uh best practices oh I forgot a bullet point ceilings ceilings are not a thing in a top down game that became very important to Boulders Gate 3 uh initial victories back to victories I like talking about victories um confirmation by a negative it seemed that we were moving pretty fast uh but we didn't have any writing estimates Spen um and uh we didn't know if we were moving fast enough uh we had about a a year to achieve like we achieved like shipping level results but the in-game Cinematic tools were irreducibly complex if you're starting up a cinematic program you kind of need your entire feature set working before you know if you have anything before that it's just like you need like acting to know what you're shooting and then you need all of the stuff kind of working at once so it took a while to get all that stuff online initial problems about don't making uh

not making games smaller um there were no small scenes at first uh for sure casual conversations uh uh would have complications uh we have this Bard alfir in the game and her conversation took more effort than her song of singing this magical song to The Woodland creatures um and it's I'll show you on the next slide kind of why that was but to a certain degree that was by design it keeps the players on their toes when they can't tell whether or not they're on the golden path and it encourages exploration if every nook and cranny of this world can have crazy things where you can just talk to a squirrel and then kick it into a tree and the guy who storyboarded that is right over there so you would think that we spend all our energy working on the on this and making sure all the like we're making real cords with the the uh the the fingers and all that kind of stuff and we spent a lot of time on that

because we cared about it and we knew players would too but this is really what we spent our time on making sure that where there weren't continuity mistakes because we had this really wonderful dialogue tool that would allow you to take these nonlinear paths through this uh uh uh getting to know this person getting their baggage as an artist which seems really personal now while I'm on stage um and but making sure that her guitar was always in the right position no matter what path you played through the dialogue file so we had to figure out how to communicate this to a writing team uh and scripting team and and figure out how we could create a shed language for we'd rather work on this but right now we're spending a lot of energy on this initial victories Muppet acting um we got so much characterization from our actors uh we initially got like kind of some paddle hands and then we adapted to stretch sense thank you Ricardo for finding them um and then we started getting a lot of finger gestures that you're seeing Gail do here and then the Cinematic team actually uh authored a

lot of the we did body capture but we didn't do facial capture so the Cinematic team you can see place these cute little emojis around and time that out so we could basically go from different intensities of of different emotions and that and that would interface with the the face effects and lip sync and what we found is that when we were placing all this these intended contextual timings of these emotion Keys we didn't need some of the extra bells and whistles with that really deep tool set so we have uh Happy a b and c happy A and C here everybody loves Gail um happy AB B and C the you'll notice these are closed lipped uh and that they are also um uh they kind of go up in intensity so happy a is the least amount of happy that somebody could be happy Z is the most amount of happy but you'll notice x y and z are all open faced so a what's the difference here when we didn't have the entire alphabet that would be too many choices to make for every frame of

dialogue so we had a b and c as a set and and then uh X Y and Z so the A and C would all work with phones XY and Z wouldn't or XY and Zed if you're in the UK uh and then you can kind of see the the bottom row we have uh a bunch of other emojis that we could just look at a glance and kind of roughly tell uh uh on our timeline tool what uh emotions somebody was in and where we could change our authorship initial problems with MPP acting look at and map so when you map every line of dialogue in your game you're baking in their whole performance and you don't the the the vo team the direction team as informed as they are and all the great information that we put into our dialogues they don't always know if you're talking to an ogre or if you're talking to a mouse so they can't and in some cases it could be either um so they can't direct the actor all the time or if the actor they're essentially we moved them from move directing their

uh their attention to uh an iPad on a music stand next to them they were they had a headmic and they were looking at a teleprompter and they were treating that as their acting partner but what if they were talking to multiple people they can't look away from the teleprompter because then they don't know what to say so this we needed more flexibility and we needed to give our artists control over what the head was doing so we but we we initially when we started making this look at tool it nuked all of the wonderful uh head bobbing and muppet acting that we were trying to get so we had to in incorporate this on an additive layer problem solved we actually did fix it eventually but uh you YouTube you know what you did uh initial victories uh grounded tone uh our CGI trailers served as excellent tonal pieces working with unit

image was the highlight of the project for me personally they are brilliant collaborators you should work with them unless I'm working with them and then you wait your turn uh their uh uh PBR style really provided some aspirational targets for us uh as we were building out our own art style and our own world and there was a lot of collaboration on what we thought we could do and and the the assets they would provide I think there was even some sharing of assets back and forth um this also cemented body horror as a tool everybody who learned something about themselves when that tadpole went into L L's eyes you're welcome um you know later on when Orin we were introducing this um uh Shape Shifter character we knew that well this could just be a normal like Mystique Xmen kind of uh uh uh UV scroll or we could do something a bit more interesting and it also let us know with our narrative tone we knew how dark we could go here's some side-by-side comparisons of what unit gave us and uh some similar

things where we could test the lighting we could test the rendering we could test the uh materials and all that sort of stuff camera angles and figure out how we could kind of chase after the excellent images that they gave us and introduced our IP to you guys with and we started with that wonderful sarhosis uh in in fast speed uh uh teaser and then we have Orin later on with this kind of body cracking uh uh motion and uh we we we could use these cuz rhymed and it felt right uh initial problem with grounded tone there were some old habits scenes were always not written with visuals in mind back in 2019 so aerion's initial recruitment scene was him actually in one of those pods because he was a vampire and he was afraid to come out in the daylight so you're trying to talk to him whether or not the sun was out and this is this makes a lot of sense and it it offers really good roleplay but if you look at that lower image you're just looking at a screen you're not looking at a person there's nothing to connect

with uh there's not any emotion that you're getting from that pod um even romance scenes were some I remember having a conversation with one of the writers early on I was like is this uh is this romance scene is this they find a Hy stack somewhere is is there a hotel nearby where exactly is this happening and it kind of dawned on us at that point like there was information that never had to be thought about before so we added a um cinematic context field in our dialogue editor for talking about what was narrative information that the player needed to know but the developers definitely needed to know first what was nonverbal information that was telling the story uh and they took to this fantastically so we got all of that and we hit our very first impass and it goes back to our core principle of flexible cinematics we wanted scripted motion we wanted scripted movement for our our cinematics so they could be one to one so that our cinematics were happening

exactly where the gameplay characters were standing movement would be driven by go-to commands and minor staging changes would be handled by gameplay Locomotion the results were interesting I'm going to talk over this because it's the only way I can feel comfortable um you can see so like actually laughter helps too um we're actually like there's not even like an auto continue on this stuff so there's somebody who's like clicking continue every time we get to the end of one of these shots um I remember this was like two weeks of work standing this thing up and I went back home that day and I like I might have made a huge mistake I don't know about this um the Lookouts folks uh but what you see a floating waterfall right there like almost everything that we have talked about is like is on display right here which is why I'm putting you all through my

trauma so this uh uh but you can see a lot of things that are working there's a lot of the the the art assets that are being used you can see the beginning of the material system we would improve the lighting uh and all that sort of stuff this next shot coming up is interesting I think maybe it's two shots from now probably after we get an air Message wait for the air Message hold there we go that's the sweet stuff uh this uh this custom movement right here we didn't actually know if the Goblins would be on the screen because their AI could send them anywhere uh and it was uh it was pretty difficult guys it was time to find a plan B so what did we do we implemented cinematic dummies not like that uh cinematic dummies for us were timeline entities that would replicate their in-game counterparts and hide them so

this allowed us to to show uh the the players a cool cinematic without actually disrupting the game State at all um this kept the Cinematic team or the the Cinematic tool as an artist tool instead of a scripting tool it cut down on animation Blends because not everything had to work from 360 degrees and it allowed us to cheat things we as a cinematic uh uh Storyteller if you want to you know somebody walk from one end of the to the other end of the hall you don't want to show that whole journey if you're using scripted motion you have to do that uh the results would take uh fairly quick for Tech um I I think uh uh uh uh they they anticipated the Need For Change um and then pipeline authorship and all that kind of stuff would take months uh results would get us here open the bloody gate nobody gets in Z's

orders the pack of goblins will be on you see like aridan like we changed him but there's a lot of these shots that are very similar if you've played This Moment In the game there's kind of some blinking out that you'll see that some of the characters doing at the end of shots that was a problem with our animation export we slowly got to fix that you can already see the materials look way better than they did a few months ago the sound team really brought it we got music in here it was really important for me that we nailed this scene even though it's just kind of like this transitionary beat in the story itself we spent a lot of time on this you'll see this uh wheel kind of screw around yeah that arrow is really tiny because like the animation was changing the render scale but we felt like we were there we felt like we had impact we felt like we were connecting with this [Music] moment form of line for everybody who says that form they

say formal line and then two people walk off leaving the third person behind I'll point out that the you only need two points in a line this was great we were all set nothing could stop us now one month later we were all sent home and locked away and working remotely I'm not going to turn this into a CO talk I will say that I got to know all the neighborhood kitties and played a lot of D and with my friends uh on the internet what this did give us was time to think and uh recognize that we actually accomplished our our kind of initial goal of flexible cinematics uh not making the game smaller Muppet acting and grounded tone and then Co slowed down the writing pipeline a little bit um and that gave us time to evolve the team and pivot and uh uh our our team uh was very very Junior heavy uh lots of folks that was their first some some uh first job out of college in

general and very many their first job in games they but they weren't the New Kids on the Block anymore they were they were growing into developers instead of just timeline jockies so what was the evolving Vision so we we we needed to start uh targeting final quality what we had gotten to with that plea of the gate sequence that I just showed you we had to spread that out a little get get more of those in the game we wouldn't have EA to hide behind forever uh we had to maintain the floor most of our game cinematics most of that 174,000 million hours of cinematics was going to be conversational it was going to be uh um it was going to be a a conveyance of information to the player about what effect they can have on the world what effect they have had on the world and where they can affect the world next that had to feel good we couldn't take those scenes for granted uh we had to collaborate to shape we had to provide a new team identity for our cinematic team

to and model how their role at the studio could change and how they could evolve outside of just their everchanging uh timeline tool we had to focus fire on the ceiling the hardest part about that was figuring out where the ceiling was uh we'll get to more on that later and then we had to solve lighting majority of the game we only had one or two lighting artists internally and there were a lot of cinematics victories uh maintaining the floor we had time uh uh with the the the production slowdown to revisit all of our scene templates and kind of drag those characters closer together the party makeup makes is is really important to players it's an expression of uh who what characters they're Val you know they value in combat uh and who they're Romancing and getting them in the background of player decision and question shots was really important for us we uh tighter staging allowed for more visible Companions and also left a

smaller footprints for gameplay cinematics so they can play Closer to the the that that game state that we were chasing initially it also established a baseline we had had we had these scene templates but when we started looking at it we had a lot more than we thought we did because people were just making a bunch of custom ones to do uh a lot of like slight variations and so we figured out okay well what are the best decisions and then how do we homogenize those and then set them up into a proper hierarchy that would then get all of that inheritance actually working for us involving problems with maintaining the floor some scenes didn't fit our templates uh wild shape uh would double as a speak with animal spell and we couldn't support all of the cute animals uh uh in our in our cinematics because you could be a giant Badger or you could be a giant spider or you could be a giant ox or I mean the cat was small but most of them felt pretty giant to me um Mid development we kind of opted for just keeping the player off camera during these uh you guys didn't seem to mind much thank you so much uh uh uh we

we're we're we're we're still toying with how we can handle that better uh evolving victories collaborate to shape so the there are two columns on here we have execution and broader tasks um execution is kind of what I told people the job was going to be when I hired them like it's ah it's moving around cameras and characters and editing facial performances and all that kind of stuff the broader tasks were things that were emerging as we were Mak the game uh we were understanding the drama we were advocating for Fidelity if this was a gold scene we had this kind of tiered system you know bronze silver brass gold bronze brass silver gold there you go um owning kickoff meetings kind of collaborating bundling information and context for everybody to do their work uh uh advocating for asset reuse if we had somebody jumping off of a platform we didn't need to map that again and the Cinematic team would know this and they could communicate this they could reactor old work flows like those scenes and then that would have

broad effect on the entire game incorporating new workflows and new features that were coming on of the Tool uh all across production and they would escalate Technical and narrative issues when we were trying to get things uh into the game that just wouldn't work with our tool set so we chopped all of these things up and we figured out that these were actually pillars of uh expertise so we have analysis how well is the assignment understood that's understanding the drama ideation how strong are your visual goals strategy how clear is your plan of action execution we all know in love and revision how quickly does your work achieve polish we turned this into a bit of an HR grid performance and development all that kind of stuff but the important thing for us was that yes working in the timeline tools really important telling a story is very very important but collaborating and fitting into uh the rest of the company at large and spreading that context spreading that passion is a majority of what your identity is a team that worked very well there's actually no downside to that so

more victories uh Focus fire on the ceiling so this will be a long walk we'll get there just bear with me cinematic context we explained is that's visual information non-verbal information that the we need to communicate to developers so they can put it on screen later each line of dialogue is represented by a node in our dialogue editor I'll show you in a little bit um uh this added field uh for visual information was used also for making our map shot list it was very easy to do a quick export and kind of read and say it's like yes this context needs new animation no this context doesn't need new animation uh once writing got out to speed we suddenly had more work than we could review it felt like we could either read the dialogue files or we could make cinematics but we couldn't do both and that was really really daunting we needed a way to prioritize our most complicated scenes without having to read them this is where we came up with the idea of custom nodes we added a toggle so that we could could count the nodes with cinematic

context we uh uh then had this idea of dialogue Fidelity it's a gold dialogue file right um and we had this these context nodes and we married the two so one custom node or a node with cinematic context in a gold dialogue would become one gold node and so we could look at if we had 20 gold dialogue files and one of them had more gold nodes than the other ones we knew to read that one first because that either need scoping or that need to get into the pipeline as fast as possible and we could also and this is kind of oddly enough with the game being so big data was on our side there were Trends there was an aggregate truth most of the time we spend time agonizing over our juras and task flows and all this kind of stuff to figure out what is the specific truth but suddenly we discovered there was an aggregate truth we could find a ratio of how many custom nodes in a gold dialogue file would result in the ask of a new animation asset so we could predict not only what

was going to be most expensive but how many animation assets that might uh require and so we could then prioritize The Cinematic animation team and you know that that's that's uh I think uh moving on so we have the the dialogue tool we have a uh this is the same dialogue file you'll see on the top and in the middle there we also highlighted our custom nodes uh so you'll see there's a lot of these like little green nodes uh we made a joke that a bronze uh um a bronze dialog file would have no green nodes in it so you had to get in there and wipe the boogers if they were there um we have 99 nodes in this dialogue file 71 of them would uh be timeline nodes they would generate visuals in some way there was a line of dialogue or there was something that needed to happen so that meant there were 22 nodes on here that are either logical or there display options uh uh for for your your branching paths and all that kind of stuff but 24 nodes would be custom nodes and

this is where we actually had to spend our time they were more likely to take more time from our team as cinematic artists or more time from The Cinematic animators and we could focus on those and now when we open a dialogue file we knew that essentially I was going to spend all my time on the green nodes and I could let the scene template take care of the yellow nodes we knew where to read because not all not everything was disposed and not everything was custom evolving problems solving lighting so there's a big game two lighting artists uh we This is Not By Design we did try uh uh uh much recruiting it's a very high demand role and we found that adjacent Industries didn't translate as well in our efforts um uh TV and film like just don't have to deal with you could you could throw 50 lights onto a scene in TV and film because you're doing offline rendering and that just wasn't working in a game engine our proprietary uh workflow was

not Outsourcing friendly so we couldn't just go like you know hire a bunch of uh extra hands to to to chew through the work and we had an auto liting lighting solution for Early Access but it didn't handle movement people like the moment they moved they would step out of the light and they would be in darkness so we built this massive backlog of super timeconsuming work that needed to happen in the game I chose this example because you can see even though we have V effects that emit some amount of light you can really see the difference between uh uh an unlit scene and a beautifully lit scene so we came up with this idea of light linking so we initially for Early Access you can kind of see in that middle area there's like the green line and the red line those are binary keys we're toggling on and off the lights we had to do this for every light for every camera all across we had to turn it on for this shot and then turn it off for the next shot and it was maintenancing and God forbid you might actually have to change CH the length of that shot which happened a lot and then you would

have to move all those keys and it created a lot of bugs the maintening thing was kind of nuts and then we finally kind of looked on the same screen up and then there's that little block area above there with all those little kind of uh buttons right next to the time code and that's the the shot activation track that was this camera is turning on now and then it changes over to this camera and it's like wouldn't it be great if we just put light activation in with the camera activation and we this concept of light linking was born it was very intuitive and it saved massive amounts of time and it suddenly kind of uh uh led us to to realize that we had two buckets of lighting work we had bulk light linking and we had custom work remember those scene templates the scene templates that every dialogue in the game every cinematic in the game used and referenced what if we could throw our lights in there and this was like this was a very like

we kind of got lucky because what this allowed us to do is we could take all of these scene templates and we could light them for every different environment in the game and it was a lot of work but it was a finite amount of work we could put our standard lights do all the linking for all the standard cameras that were in those template files and do our color authorship there and then on the timeline layer we only had to do the custom work where people were moving where they were stepping out of their lights and it was like we figured out a way to infect the game with lighting authorship rather than just doing it the way that you would normally do it and then we had the game which was essentially selecting what colors the lights needed to be here's an example for Valentine's Day we released the more kissing everybody loves more kissing um and so and we we put that in our like in party dialogue so you can uh talk to one of your um uh companion characters and get little peck on the cheek or more um and this would the the lights would respond on to wherever you were even though it was the same timeline file the

scene file could be swapped out on the back end and we could infect the game with very very cool lights that felt that it fit the mood of the environment around you so project takeaways what worked uh inherited authorship uh the those scene templates really saved our bacon not just with like getting cinematics done but getting lighting done um and it gave us a lot of standard cameras and all that sort of stuff uh the Voom moap pipeline uh just Ed a tremendous amount of characterization um uh and really helped I think uh the players connect with the amazing actors uh both uh uh in the game and on Twitter um and then uh scaling the team's expectations to just like you're not just an artist toiling away on your timeline file you are participating in informing the processes that are happening around you and then using data prioritized vetting the one thing that is never out of date is P4 I mean unless somebody leaves a file checked out over you know the weekend

that never happens though but uh when you go into like jira and you spend so much time fretting about these tracking algorithms and this kind of stuff uh uh it's you can get out of sync with some of your your your systems but if you're just looking at the raw data if you can just look at this dialogue file has three lines of dialogue that don't have a wave file connected to them and use that for your tracking it's constantly telling you you what work it needs you to do what didn't work uh golden path scoping like this is typically what you do is ah this is important to the game so uh we won't spend any money on it well that wasn't going to work that that de incentivized the exploration that we wanted to achieve and then one one cinematics is still stings but uh uh we had we adopted our cinematic dummies and uh the game looks good but I I I I still want another crack at this uh personal I'm going to get personal for a second um personal takeaway uh kill your darling so this is a phrase that's used for storytellers

you take eliminate all the parts of the story that you love but you cannot justify and what remains are the stories or the moments that actually serve your story this over the course of balers Gate 3 is what I did with my job I loved storyboarding and I got storyboard artists i l working with the team and making their shots better and I took a lot of that and handed that off to my leads um I love doing map shoots and working with actors now hir performance director who's now our head of animation and I hope you saw his talk earlier because he did a fantastic job earlier and what remained were director tasks what remained was unblocking these people so they could do their jobs and I'll I'll read off one of these quotes uh at the bottom of the screen from Lazelle go ahead go ahead thank you uh I'll try I'll try a laelle impersonation here it's like it must be

said because of you I still breathe because of you I still battle I am not one to show gratitude but a hem thank you I had a new vision for vision I thought being a cinematic director was about me delivering the story and that is what I had worked my career toward was trying to chase this Ur ISM which at this point sounds more like a disease to me and instead it was removing obstacles so my team could deliver the story and that's the bottomup approach my raw material wasn't the story my raw material was ambiguity and my goal was processing this into Clarity and the clarity meant that my team could focus on storytelling uh thank you so much you've all been a wonderful

audience uh I also want to thank uh Ash Jameson who did all the wonderful little uh lazel uh head shots that I put those funny little quotes that you can find on the Vault later uh I'm sure I said a lot of weird confusing things so let's do Q&A uh hi great presentation thank you uh how did you solve the ceiling problem this oh the ceiling problem that that would be a GDC talk of its own uh it was the the the art team um and uh yeah I mean they they they basically just bailed me out on that one I I I but I'm not going to claim to know what they did all right thanks they made the ceilings show up [Laughter] sorry hello wonderful talk thank you for that um at one point you mentioned that your team was a lot like Junior heavy uh so how did you guys approach uh making the product but also like teaching the

Juniors to become better professionals but the product wouldn't also you know kind of suffer in quality because well people don't know what they're doing so we focus very heavily we have we have a a pretty tough skills test um and what I would look for especially because I knew we were going to take this template approach with those scene templates that I talked a lot about um there were only there was it it minimized where uh uh people that are learning the pipeline and learning the the Cinematic rules could actually damage you know the the the the the vision um and this left me open to just looking for people who could take notes well I often um gave uh uh feedback on the skills test and I wouldn't ask them to redo the whole thing but I had asked for a section to be redone to see that they could take Direction well um and uh it really allowed me to to just focus on can't do can they learn and do does does their work get better with notes and if on on on that rock I can actually build

the you know the foundation from so sweet thank you no problem hey Jason thank you for your talk um yeah uh I guess I just had a question a clarification you said you mentioned cinematic dummies right and I just just to make sure I understand what you were putting down there um obviously the top Domin perspective of you know all the Lian games that presents a problem when you kind of need to transition into to the kind of like Yeah The Cinematic perspective and so the solution then was creating a dummy which is essentially a replication of your scene without all the problematic art assets that you could then use for the cut scene itself is that right uh no just not the problematic art assets just the characters themselves so so we could have you you could have the gameplay character standing here and then you could have the Cinematic character actually like talking to the audience and that sort of stuff and so we we created these like little micro and we tried to keep them as close as possible but it allowed the Cinematic team to

have a character they were controlling and then the scripting and gameplay team to have a character they were controlling and then when you were in a cinematic you were seeing the Cinematic dummy but not the gameplay character and vice versa okay so it's less to do with the scene more to do with the characters itself yes correct oh that makes so much more sense okay thank you no problem uh hello I have a much less serious question oh okay I think it was uh update two or three where they added the extra leaning down for romantic Partners to kiss the player character their gnomes or halflings my question is one why wasn't that in the main game and two what was the reaction when you realized you didn't put it in there uh so there there's like there technical reasons and and and that sort of stuff uh um there was uh there there's some of that where we actually did the authorship but the uh the way the ground finding worked like actually kept it from looking right and so we we fixed that um uh and then there

was just mostly what we did in the patch that I think you're mentioning is we added more of them so we wanted them in the main game and we we thought we did well thank you for clarifying no problem you mentioned that one of your core principle was not making the game smaller and do you still live by that should I like quck my job for the next game or something like that and at what at what point enough is enough at what point is enough enough uh that's a great question uh I'm not going to uh speculate on that that is like something that I ask Sven often um but it is always as always his question to answer sorry hey uh I am Yan and thank you for that very entertaining talk Jason uh I had a question uh regarding more production and team so You' shown very vividly the victories you had the

challenges you faced throughout the process what I was wondering is especially during Co how did you and your team uh contact burnout pronouns no burnout burnouts burnouts um so there's uh there there was a little bit of uh uh working in the EU there's there's actually limitations on like how many days in a row people can work and that sort of stuff that certainly helps um we this was one of the things where like I was talking about like all of our custom nodes and all of our tracking this kind of thing this eventually led to I didn't want it to become a spreadsheet kind of demonstration because I'm not a producer but I do look at a lot of spreadsheets uh we were able to predict how long we thought it would take a cinematic artists to chew through their work and if they were working at a good Pace we would make sure that we would kind of we we sent them home sometimes uh it was something where I wanted um I I had some long hours when I was coming up in the industry and uh I

wanted better for my team so we tried to to look after them in that way by really having a clear idea on how long it should take them to get things and then if it were taking too long we would we would sometimes they would protest but we would re um distribute the work to other team members who were moving ahead so that's that's how we we took it very personally very insightful thank you thank you hey Jason uh so super informative talks thanks so much for sharing the information so I too work for RPG games uh RPG single player games so I'm wondering uh you mentioned a lot about the staging and I really like the staging for balers skate so I'm just wondering who are responsible for those staging like are designers or writers do they also like contribute to how sensor stage light lit or like how how they compose implement it so our our our director of um uh uh World build buing farang like once said there you can look at no part of balers Gate 3 and and

point to something that 30 people didn't touch so they'll be like for the most part cinematics would read and be inspired by the work of writers and come up with the staging plan but sometimes the writers would have specific ideas or you would gather ideas from the level design itself uh as far as like the standard staging that was mostly the uh cinematic leadership because we we needed to make sure that we had alignment on what the standard uh uh setup was going to be and what the what the goals were and then the rest of the team got to implement and use those yeah that makes sense thanks so much and also thanks so much for bringing now such a beautiful experience thank you yeah hello uh first of all thank you second of all I have a very serious question so with the Valentine's uh day update uh not ascended aaran got less kiss animations that other characters so

uh how are you going to fix this this ground Injustice I mean this goes back to how much is enough right uh uh it it's something that uh uh uh yeah I we I have no good answer for this bye yeah thank you anyways thank you you said something so profound you said you as a lead don't worry about delivering the story but you unblock the rest of the team so that they can create the story and I wanted to know how that type of communication actually went between the Departments involved with that type of mentality geared towards the well so it it's funny so I I have a a uh I have a twin brother and uh he's

here at GDC and so the joke is is like now all my team get to meet him and figure out which one of us is the evil twin and and everybody at laran knows it's me uh because unblocking would take confrontation it would take uh some real talk uh with some of the other collaborators and we would do that on a director level um and that's how we arrive at consensus and something that would work for their team and would work for my team as well um um and uh yeah I think it it did take like some of the folks that that joined the Cinematic teams from from other cinematic pipelines that ship games before it kind of took them a while to get used to me like not having something to say about everything they were doing um because I wanted them to they they needed to grow and there's a there's a kind of a term this one of my weird terms of compatible mutation uh you you want everybody to uh uh invest a little of their own creativity in something and it's not it's not going in

the wrong direction it works if a plant is seeking the sun it won't die so you have to make sure that it's seeking the Sun and that's when you would choose whether or not I would intervene on something was it veering off course or was it working well and could like if it could be clearer that's where I would typically intervene but if it was just like not I didn't want the team to try and play the song that only I could hear cuz uh the game was too big for that and then how did you handle the more difficult conflicts like maybe you had somebody that was stubborn and was like no I want to tell this specific story and this is all I want to tell and you're like uh a lot of times it's the communicative goals like there's every like it goes back to analysis what is this scene about and then if we make the conversation about what the scene is about then we can arrive at a consensus on what the execution should be but you have to go back to core principles and that's something that that I pushed a lot with the team so there weren't many like there there was no like uh

loggerheads with uh with my team that I can remember maybe they disagree gotcha thank you sir thank you all right uh thank you for the genuinely most funny session this GDC has been amazing um and it's also always wonderful to get to see that it's truly an iterative process riddled with all these things um how did you get your you know player community to be so tolerant of of what's going on in the cinematics some of these funky things going on in the background until everything gets patched up uh I they were just with us like I mean like they they they they had us like they supported us like they're they're our community and and like I inherited a wonderfully trusting Community from uh the Divinity franchise and all the stuff that they had already done so like I just I got very lucky with that but I don't know what makes them so awesome but thank you for being [Laughter]

awesome awesome amazing talk has been a lot of fun and thank you for making balers G three um I was also impressed by the comment where your job was to unblock the team and allow their creativity to thrive I'm curious if you have advice for people who are not lead level um you know not managers for when their managers are overstepping or where they would like to try to involve or bring that culture to their own Studios I think think it is it depends right um there there there two truisms that I found about that that is applied to to my whole career and the first is the reward for good work is more work um and you can never be right at the wrong time and sometimes you're in a situation like that where you know in your bones this is right but the universe is like nobody's feeling that need yet nobody's hungry for for for for for what that truth is sensing whether or not that's the case like and I also know that expending a lot of energy trying to make

it the right time when it's the wrong time you you can burn a lot of political Capital that way so tread lightly with that if you can have a conversation about about those initial principles what are the communicative goals if you just feel that you're a good kickoff keeps notes from feeling punitive because if if you if you have an agreement on what the scene's about uh uh again it's just kind of talking about how clear are your artistic decisions making the point of that scene and that that like if you can if if you've lost your way in that conversation with your lead bringing it back to that and making sure that you're on board and aligned with what their goals are is typically going to be where you're going to find that collaboration amazing thank you cool all right I think that's it uh thank you everyone have a great GDC [Music]