Animating the Spy Fantasy in Splinter Cell: Blacklist

[Music] so I hope you've all had a good lunch because I'm gonna give you some animation pudding a first joka today okay okay so I wanna give you little bit of background on Who I am starting in history about 14 years ago worked on a bunch of games some probably a lot of them you won't know punctuated by assassin's creed in 2005 moved back to England worked on another couple of games that you might or might not know and then ended up moving back to Canada working at Ubisoft Toronto to be the animation director on Splinter Cell blacklist gratuitous baby picture day okay so I want to give you a brief overview what I'm going to talk about today so I'm gonna start with the evolution of the animation on blacklist giving you some context into some of the

decisions that we made then I want to talk a little about the involvement of our tactical coordinator and where he helped us out which will lead me into an overview of some of the methods we used and some of the tools from there I want to talk about the concepts that we had of everything in motion and how that defines some of Sam's movements which should help define and develop our animation contract system and how that helped would play a choice and then you can't talk about Splinter Cell without talking about the AI so I'll go into some of the things that we did there okay right so this is the sixth game in this one so franchise it was already established but even though even though that was the case we still had to go ahead and define what we were going to do for blacklist and this is the first game at Ubisoft Toronto and they they wanted to prove something and so we want to make a strong first impression to headquarters so what I'm gonna do is I

want to show you give you some context into where we started from this is the common video the has only ever been seen internally at Ubisoft and it's one of the ways that Ubisoft will greenlight a project now with Ubisoft Toronto being a brand new studio we didn't want to just create a pre-rendered footage we decided to go in and create a playable demo using the previous Splinter Cell as a base so I'm just going to go ahead and show that [Applause] [Music]

okay so that was it was a pretty good technical achievement and the basically from a bunch of people that hadn't really worked together we managed to get a playable demo up and running but and and that course has passed us to go to

the next stage but the thing is is we didn't feel like it was a Splinter Cell we wanted to make it wasn't hitting the right tone we wanted to like return it return Splinter Cell to his stealth roots and think more kind of hurt locker than mission impossible so we wanted to take some of the action out there the studio supported us wholeheartedly but with that came like a new animation mandate so there was renewed focus on the fluidity of animation we wanted to blur the lines between the systems as much as possible making the gameplays responsive as possible of course as I said that we wanted to return the game to its stealth roots and you know like really focus on that Splinter Cell has always been known for its aesthetic gameplay as well so player choice was going to be a huge thing and then we knew that there's going to be a lot of animation that was gonna have to support but ultimately at its core spun selves about manipulating the AI and the team wanted to excel in blacklist and push this further than any other Splinter Cell had done before and all those behaviors we're going to need an

absolute ton of animation support as well okay so this was my first time as the animation director on Splinter Cell so I wanted to define a couple of key pillars that I could help with the team so I define that as basically it is Panther and realism I find it useful sometimes to connect the character to an animal and the meaning behind Panther was to make Sam was to make blacklist Sam as fluid as possible kind of very cat-like this was an idea that was used in conviction but I just wanted to take it one step further and with realism I wanted to make sure that the world was aware of everything and vice versa you know grounding the player and each other like tactile moments and some cool stuff like that I split these pillows into three parts the first one being proud I wanted to set the concept of the panther through to navigation and giving Sam intent in all his moves you know like kind of a hunter like stalking its

prey that sort of thing so there's a very kind of cat-like theme that I use cat references a lot by the way making sure that all the moves flow together surgical strike now this would refer to Sam's combat style I wanted all his moves to have precision and I wanted them to everything to have a reason you know I didn't want to just have random strikes and just wasteful stuff I wanted everything so you know like if you hit someone in the stomach on it to go down so you could actually do something with it and the last one was response and response was important because with all these potential animation improvements and like slinky animation I wanted to do it didn't matter if it didn't respond it's like you dude when you push the input you want that to feel good and you want it to feel like you're Sam Fisher that's going to really embody the panther okay and with realism it was about Sam and the enemies having an awareness of the world around them as I said I wanted everything to feel very tactile I think I'd just seen children of men and there was lots of like touching a walls and all this sort stuff and I was obsessed with like grounding the character to that you know it's like

I wanted the AI to communicate visually I didn't just want them to have audio box I wanted them to actually be signalling to each other whether the player were gonna see that I didn't really care it was more of I just wanted them to be communicating and with all that awareness it wouldn't be much fun if everything wasn't reactive so salmony I would need to be visually visually reacting different to each other you know want people to react to explosions I wanted to be able to fire a gun in the air and then you know that would actually cause chaos and some of these subtle responses were going to be the front in how we played okay mood videos were created support all those those sub pillars but I what I wanted to do is to show the team like a concept of of all those pillars together so we didn't have the time or the resources to create again a fully rendered video to kind of show anything so I was given two days to basically create this from scratch so I'm going to show you this this is the first concept

for blacklist that I created and it's to try and show some of the things I've been talking about where when he was coming out from cover it was touching the wall and I was gonna behind someone take him down with a knife and using the gun and also I wanted to keep everything nice and fluid so I wanted I wanted all the transitions to be nice and smooth so you can see so sound nice and efficiently taking out a bunch of guys there okay so the good thing about that is the team was on board with where I wanted to go so they wanted to see where I went next the thing is is game play needed to try and cast their own Sam Fisher we needed reference for ourselves as well and I'm a huge believer in gathering as much real-world knowledge and reference as possible I was really keen on finding some of them back that embodied our Sam Fisher but the problem is is well where'd you find the world's most elite spy well it sounds like he lives in Montreal this is Kevin succour

he was introduced to me by one of our designers James Everett and he'd been training with him when he was living in Montreal and he was like hey you should probably speak to this guy he's got some of the things that you talking about and as you can see from Kevin CV he he's definitely got a lot of combat experience but I wasn't quite sure whether he was gonna fit the bill so I sent him an email and I had no idea what was getting myself into so this is the video or part of the video that Kevin sent me back and as you can see it proves to me that he's got a very interesting way of moving this guy to me is the real deal he he's not someone playing at things he'd never done a mocap shoot before and this is he kind of lives this so so this is like loads of footage from him just beating people up in his dojo for me so see her anyway so I was convinced I had to meet him sent me an

email I went to meet him in Montreal and you know it turns out is a huge Splinter Cell fan so we were kind of like sending e-mails back and forth quite a lot we redacted lines in there people can't see so but then the cool thing is is every time I sent him an email he's actually an instructor as well so he would always send me back a little video so so his his students must absolutely hate me now you can see a bunch of stuff on here you've got like I think the bottom right-hand corner he was talking about how to beat someone up with a shield and then he's there's lots of weird stuff on there there's basically it's every time I asked him something he'd sent me about one of these so he's pretty he was that's approved to be a huge inspiration to the team not only for me but to design as well so so this was going to be a great fine for us but I'd

mentioned briefly to Kevin that we wanted to bring back the knife from previous splinter cell's if we splinter cell's they've used just a normal straight edge blade but I wanted to create something a little bit more iconic so I mentioned to him that I was looking for this he's Kevin's eyes just kind of lit up and it's just like I've got something for you okay so he took me to one side and he basically pulled out the Karambit this ended up being kind of like an epiphany moment for me it's like the Karambit is more of a tool it's more of a curved edge blade and it can be used for kind of ripping pulling tearing rending that sort of thing and you could actually be used to pull people into cover which is really cool and the thing is when I went to his dojo he created this this chart on the side which would show you how you could suppress someone which is really scary that he thought this deep into it but he'd he decided to like just show me all these things and this was going to change exactly how we animated these moves and and one of the

moves that was there I will go into a little bit more detail oh and bonus points the fact that I've been talking about the cats and everything is its nickname is the Panthers claw that's just that that's pretty cool that works so okay oh and there you go it connects to my first pillar the surgical strike connections it's good okay so with that I was really convinced that we found our Sam so much so that I sent a not so cryptic Facebook message to the team which was very public at the time considering no one knew I was working on Splinter Cell that's funny didn't get into trouble for us this ok ok ok so we blacklist massive scope and we had this man for internally of the biggest Splinter Cell today it would where it would require a huge investment in our tools on our pipeline now small disclaimer none of these tools and techniques are particularly magic Bullock or groundbreaking it's more of the well-known to the industry it's more

of how we use them together that I feel is really powerful and without them many of the choices and decisions that we made wouldn't have been possible ok so I spoke with our creative director to see where his head was that he basically said more of everything I was like great fantastic the problem with more of everything is something's gotta give right we knew that it was near impossible to keep the massive scope ship on time and the quality level we're aiming for usually you have to focus on two of these choices I asked him which ones are more important to him all three brilliant so so this was max throwing then go on flip down I was fair enough it was early on the project he didn't see any need to cut anything and he's just like just see what we can do so it was just a but you know of course it was essential that everything was done on time we didn't want to compromise on quality and he wasn't really willing to let get rid of the scope so go ok so this is just a

quick overview on some of the things that we wanted to add from conviction on blacklist I just want to give you some of the examples of the things that we're reading so we were going to add like a different movement speed I think conviction had only had the two movies we were going to add like a jog in the middle and we wanted to make the run a lot faster but also be at we felt strongly that we wanted to have like a lethal and non-lethal option really kind of playing it's a player choice we didn't know what we were going to do with at the time but we knew we wanted to have it all so there was a bunch of new abilities I don't know how well you can see on there but this thing's like like ladder climbing alleged jumping and some of these things some made it in the game some didn't also yeah we wanted to really focus on how crouch navigation was working a goal because we knew that players were going to be using that a lot and finally there was a bunch of new AI archetypes that we were going to add so all of this was going to need

animation support and this was all on top of things that were already there okay so just to give you a quick idea convictions animation budget was about 40 Meg so I don't know how much that means to anyone but that's that's a decent amount you can normally get quite a lot done with that so I was like great considering the scope if we get that for blacklist then we'll be fine but we got 23 and so how are we going to do that okay so what I wanted to do is I knew that we needed mocap just due to the sheer amount of animation we're going to need we knew that we were going to invest in a solid mode mocap pipeline mocap will allow us to keep as much the integrity of Kevin's moves as possible when it wouldn't be possible to mocap we need strong animators to help prototype systems because mocap well it wasn't always going to be sort of it wasn't always going to be the solution you know sometimes they need to animate some of the more difficult systems some of the things that we didn't want to go to mocap width that we might end up getting sued over also and you would need to

improve the animation tools so the the tools that were already existing and were functional for you know for all the previous games but we knew that to reach the level of quality and to do everything on time and just to get everything in we knew we were going to have to work on like increased animator control so animators could play around with things and really try and experiment and we also knew we were gonna have to somehow find massive memory savings okay and of course the scope didn't just apply to the animation team is every team had this mandate so every team we're going to have to add more so they all had to go through the same sort of problems okay Ubisoft we were really lucky the headquarters sought to invest in in us as like a center for performance capture and we actually asked for like a smaller system but then Montreal was getting so booked up with all their mocap shoots that they said well just build your studio I mean that's great but and we didn't wanna we didn't want to waste it I mean gameplay issues for me have have

different objectives from cinematics the focus shifts between the aesthetic how it might feel to play and the technical constraints that are involved our moves need to look good from all angles at all times so what I like to do is I like to like focus I don't like to have too many people on set but basically for me to run a gameplay shoot I like to keep things nice and simple so over the past 14 years there's been some simple guidelines that I'd like to follow okay as I said I like to limit the number people on on set at one time I find that as if a set is too busy you get distracted keeping the numbers down helps keep everyone focused everyone knows their role usually that would mean myself directing we'd have our team lead basically keeping hold of like the excel sheet making sure that we got the shots and she was always offer a really valuable second opinion of course she got the actors usually - and we'd usually invite like one or two animators onto the stage as well just if they were going to be working with those moves just so they could understand where the

intention and everything was coming from because if we just gave them the moves then they might lose some of the aesthetic it seems obvious but everyone on set needs to know who's directing and you know I'd like to be the Alpha so it's like I'll be on the set the actors need to know as well it helps keep the vision clear they know that they can go to you if they have any issues and you know you don't really want people kind of chiming in with opinions all the time because again things get really confusing so it's always good to kind of filter things through one person somethink I found that weren't really really well with Kevin was collaboration to me that's the key to success with working with mocap especially for gameplay suits is I like to work with goals in mind not necessarily a direction sometimes I'll have a very very clear idea of what I want and then we'll work with that but more often than not if I have like a goal like okay you start here you end there do something in the middle and make it look badass then

Kevin more often not came up with something that worked okay but that said you need to be respectful of your talent I mean there's no point in let's see excuse me it's no point in in having your actors working on like Crouch moves all morning where it's gonna kill their legs and then you go oh yeah by the way in the afternoon we're gonna have you beat someone up then that's not going to work it's gonna unless you want them to look tired when they do it then it's not going to work to me if a particular move is giving you trouble move it around you know move on to the next one it's it's gonna get your actors frustrated is going to get the people onset frustrated is that you can always come back to it later and you never know it's like you know it's like it's like almost like you're sleeping on it you'll have a better idea of what you want to do and personally I like to go to mocap with an almost impossible list of moves that I'm looking for so I like if I manage to capture everything I need I let the actors kind of play with some of the things that I need to do in in the afternoon and we actually got some

really cool stuff from that as well so and some ideas that may end up just ending in the game that you didn't actually think of okay so this stage became our playground with Kevin becoming kind of the one constant on set but we found to get the most out of Kevin we had to employ a stunt guy that would work originally we had Kevin come in with one of his guys from the dojo but the problem is is he knew Kevin so well that he was anticipating everything and wouldn't show any pain so just the dates looked terrible so we got a guy in that had no idea who Kevin was um that was a very interesting first meeting it's especially considering Kevin's never been on set before he doesn't know what half-speed is Kevin disco is boom acid so if these guys didn't know how to take a hit then we could have been in trouble so I want to show you a quick kind of compilation of one of the shoots that we did

so that's Kevin sicko I had to deal with that guy every day for three years anyway so the cool thing is is we're starting

to get into the meter things now is we get this data what do we do with it now the things were Kevin again with it being the first time him being on set he didn't really know how to sell an action he didn't really know what it was gonna look like in game so as you can see on the left it's the raw mocap it looks pretty cool you can see what's going on think it's like a kidney attack or something but on the right is what we did to it and you can see that we've kind of accentuates some of the lines we've made things a little bit more aggressive we've cleaned up the feet we've made things look a little bit got a lot more impact I mean we we keep everything on the move as well and this was kind of typical to what we do to some of the animations the thing is is this particular move is Kevin did it at such a speed that we hardly have to do anything to except for just accentuates it so the timing was there so again it's kind of crazy so I'm gonna give you kind of a breakdown on one of the particular

moves that we decided to ride for blacklist we called them abductions they were seen as cover takedowns and they fitted really really well with the the stealth roots of Splinter Cell so some of the choices we made for abductions we wanted to be able to pull people in to cover or dispatch them really quickly hide when they were in cover as well so here is the raw data from one of the the moves and it's pretty cool but the thing is is we can't put that in game over well because we what we do is we actually we do put that in game but then we feel how slow it is and how it slows down a gameplay and it doesn't really work to me it lacks a bit of impact as well I mean Kevin was good but he couldn't get that the sort of impact that we needed because otherwise he would have just we would have gone through stunt guys like candy see to me there's some broken contacts in there as well the hands don't connect properly Messi foot positions it's a little bit too slow and

I knew I wanted to amp it up so then we give it to an animator and then we usually after about a week they come back with this and again there's a there's some more aggression on there if they were really smashing their head down you know pulling into cover with a lot more aggression dropping the weapon so this this was starts to work so what we'd also do is we'd fix shoulder and neck positions because we find that usually from mocap is the shoulders would always be really high or the neck so before would and also we wanted the move to kind of fit within five seconds at the maximum and even that was pushing it so once we did that we'd add one extra layer and one of the things that I really wanted to do is I wanted Sam to be aware of everyone all the time so now all the animators had gone through in animated this you know really cool animation I decided I didn't want to show it so we animated the cameras to focus on other things so it's kind of

you can have all this animation in the background but then the camera will be looking somewhere else so that also connected to our surgical strike response and awareness so again I'm connecting everything background this is really cool so as you can see is they well if you can't sleep it's quite dark you got a guy passing in the background you've got all the animation happening and the thing is is the camera control could be interrupted at any point so that was where the animator work was he seen this kind of idea stuck with something he called one step ahead so we wanted Sam to always be looking his next target once he had something in his hands there and this was actually something that came from Kevin Kevin never really looked at what he was doing he was always looking at the next thing so this would take the animator about a week and I think that by the time towards the end of the project he managed to be able to do one of these at about a day okay so

one of the other things that we did is we had this thing that we called the variety node and it basically allow us to create non-standardized Loops is like one big long sequence of animation that we could use as a base and it meant the Sam could be Harry's this was in one of the levels in London where he was poisoned so spoilers if you haven't played the game and we wanted him to stagger around I had all these ideas for him kind of leaning against walls and vomiting and you know is they I really wanted him to throw up for some reason so so yeah this was this was only possible with the variety no but the cool thing about this is that would be used over an existing set of animation so we could just have our base movements set and just replace like six or seven animations and all we'd have to do is mark them up as ABCD etc etc and then the engine would see them and then play them in sequence it also meant that we could we were clever we could play around with that sequence and create like any number of random ish

sequences okay replacements was also another thing for us as well this scene from the very very beginning of the game was actually just replacing a couple of animations from Sam's base walk set with no weapon in his hand and it meant that all we had to do was again create a few animations and we got the rest for free the design team really wanted to show Sam kind of gradually get thing better as he was walking you know so we wanted to show him really injured the beginning and originally this is really like oh my god it's gonna be like a massive animation set we can't do it and we managed to do this call that story but this is this is me and a mocap suit is I got injured the day before and I don't think I told my calf muscle and I quickly ran into the moat well not quickly I got into the mocap I got into the mocap suit and then like let's go get it so then this came back and that's actually for all posterity is me injured so there you go so I was the very first

actor to play sound fisheries when cell blacklist fact okay so everything was going great and then Mike was a loose key our league a programmer and that's an offender picture by the way they came up to me and said okay we need to talk we're running out of space okay all right fine he's like so he but he basically threatened to start cutting things he's like we've got a budget and you're not sticking with it you're just adding stuff and yeah it looks great and everything but we're not gonna be able to ship like this so they were all the words I needed and he was basically saying I couldn't have all my toys and they see as like if you wanted utilities back you had to tidy your room so this was the animation this was basically an overview of where the animation budget was I think it was like August 2012 we had a budget of 23 and before I'd added a load of other stuff we were already at

34 so we knew we're gonna have to find savings somewhere so we worked in this tech basically animation bundles again this isn't any secret to the industry but they were essential to how we used animation in blacklist up until this point animation all animation have been loaded into memory at once systems so what we wanted to do is we would split systems into smaller packages and you know we can have these smaller packages such as like cover and Sam's navigation and like doors and moving body and these are things that we knew we were going to need at any time in the game but then what we did is we doll solo packages such as basically sir excuse me we could also load things that we would or wouldn't need for a certain map so we knew that maybe in one map we wouldn't need a climb system or we knew we wouldn't going to need ladder navigation so we could then strip that out and then put that on the disk and then it could

be loaded in as and when we needed it also we could load certain packages on the fly things like lethal non-lethal that was actually a toggle so you could play that in the game and you could you know it would essentially pause the game and because the loading was really quick it gave us an opportunity to load out the lethal and bring in the non-lethal we could also do things like riot shield as well because we didn't use that that much and fast rope again is that we didn't use that that much so we could load that in and that was absolutely fine overall the simple organization I think saved us around six Meg which is huge doesn't sound big but it is it's impressive it's that almost deserves a clap okay okay anyway okay so the cover system in Splinter Cell Conviction was really good was was actually was actually really efficient for what it was but was so complicated we wanted to reuse it but we found what we could do is we could just mirror half

of it so we stripped out I think the right-hand side and just press the mirror button which we added again there's a lot more complication to how that tech came about but as an animator it was mirror and that was it so the good thing about that is we could just replace using the replacements and everything we could just throw in an animation if we needed it say for example if you wanted to do something with his right hand and he's over on the other side it was really easy to put something else in so again this saved us around 4 Meg's already that's 10 Meg we've saved so Mike starts to smile again puts his gun down and also for us layers were huge I'm not going to go too in depth of what layers are because I'm gonna kind of assume that some people talk about it and there is all the talks today talking about layers so I don't want to teach a grandmother to suck eggs but for us we'd have our base animation this was our like one-handed animation and then what we could do is we could quite easily add just a weapon layer of

just some jiggling arms that look like he was kind of running and then we'd out that over and then we'd get that animation this basically meant that we could cut two-handed as a base animation set and we got it for free so this basically took me a weekend to do and then it's like and we got rid of I think it was six make six Meg in a weekend and it's like so already that is what 16 Meg that we've got rid of so again Mike's really really happy at this point and the cool thing is it kind of stuck with the whole idea of prowling and response as well and it basically meant that I didn't have to work on three animation sets I didn't have to work on 0-0 handed so when they haven't got a weapon in the hand I didn't have to work on one-handed or two-handed I only had to work on one of those sets and then use layers to kind of help us out so so using all these together would actually be really really useful okay so at the end of that

after a quick spring clean with the biggest savings I think we're in navigation juice the extensive use of layers the cover system because of the mirroring and we managed to make a dent into our AI behaviors again with some of the the river variety node and and the layers as well so we ended up getting a more reasonable number of about seventeen mega we were down to the bare bones now it's like but the cool thing is is I could now have my toys back so some like handing me the keys and they're all the numbers I'm just going to show you animation from now on there okay so one of the things I became a little obsessed with was I wanted to keep Sam on the move pretty much at all time I loved this kind of concept like I wanted to use it through everywhere so I'm going to come describe what it is the idea was to keep as much inertia going through animations as you know improving how fluid Sam was and blurring the line between the systems as I said locomotion traversal hand-to-hand

I didn't want him to stop I didn't want I don't want to have to go up to something push a button and then it happened I wanted you to be able to think on the fly and like analyse by and like Kevin does I don't need to be able to act not stop and just carry on so I put this video together you know it's things like you know when we're landing you can see how out of control years and you can see that they'll take the landing through the run and you know like bond here was a great reference for me and you know just from the way that he was always kind of just continuing to move and always you know I loved the whole idea of having his hand down when you know when you're jumping down to things I didn't want him to be all parkour II you know I was looking for something a little bit more or a bit more cat-like so then we went to mocap and we basically got another guy a guy called Stefan julienne and he's done a bunch of mocap for a bunch of other Ubisoft projects but we wanted to just play around in the space you know we set up a bunch of obstacles and said right I

want you to go from here to there as quick as you possibly can use whatever you need to do to do it and we'll work with it afterwards so at this point we didn't really know what we were trying to capture we just knew that we wanted to capture something nice and fluid and the whole idea is we were going to try and put it all together so then very quickly we managed to start to see some of the animation together this was very early again it's like this was before we cleaned up some of the animation we just broken it down into raw form and you can start to see that some of the lines kind of keep nice and clean he's always on the move whereas before we would have gone up to these positions stopped stepped up stopped again and then carried on we all keep need to do is we wanted to make sure that this work for walk jog and run sets we also made sure that we wanted to connect everything to Crouch so you could be any position at any time and then just dynamically change the Crouch and then it would absorb more of the impact you know so there you can see he's putting his hand down and it starts to work really really

well so we're really happy with this okay and again that was connected to the whole Panther idea of prowl it was really responsive and he felt really tactile and again it was you know it was reactive to the world so we wanted to maintain the momentum through the hand-to-hand so we first created a moving version of all the Sam's hand-to-hand moves so as you can see you're just kind of you know take the guy down minimal movement put them down nice and nice and quietly it looks really cool the problem with this is it lacked precision so we went and created a static version of these moves as well so the thing is is this is effectively the same move because it's an attack from behind but depending on the proximity of the character you were or depending on how close the stick was or how much you were pushing the stick would play one of these I think if you're a walk speed we'd play the static one and if you were anything above that then it would take the inertia through with the moving so here you can see the

difference so and again connections to prowl and to surgical strike because this was you know nice and nice and precise the way everything's working and again it was really responsive so we proved this worked really well for hand-to-hand and it's clear we wanted to take it even further so we developed a few systems that would encourage the place to play stealthy we wanted to show Sam's awareness of his surroundings and in turn the AIS perception of Sam it was extremely important that we showed Sam actively acting stealthy and Kevin was really cool basically knowing about his surroundings he used to talk about using his hands like whiskers again the cat references cake coming back but he'd like he's always having hand now and he'd be like touching the wall he never really been looking at anything and it worked really really well he used to do things worried like cover like his tracks I with his hand behind them as well which worked really nice but we

didn't put that and again so anyway this resulted in our animation contrast system and it would allow our animation to change dynamically depending on the situation Sam found himself in we wanted to focus this the attention on three key areas so again Hansa hand navigation and the AI awareness this would involve creating unique animations for all of Sam and the AI whenever their awareness level would change and this would change dynamically as we were playing so again we're going to have to throw in a load more animation and but it could very quickly explode if we didn't control it okay so we wanted to add combat and stealth variants of everything and hands a hand was the easiest way to prove the concept and decided to start without cover takedown mechanics as I said the deductions so here we've got a combat version of a move and what would happen is to me this was about threat suppression what would happen is if an AI was aware of you were an AI was aware

of you then they would play this animation and then we had stealth which was more to me about sound suppression it was if the air I wasn't aware it's about bring you into cover and then just hiding you so here you can see these two moves this again it's essentially the same move it's all the same criteria the only thing that changes the AI awareness state these moves would be called automatically and it starts to work really really well for us and again connecting a lot of the connections so it's like you've got the awareness she got surgical-strike and you've got reactive and everything's pretty tactile so there's lots of touching okay but the criteria for this again to kind of give you a context inserter how many animations were involved in this is we choose the cover type so the cover type would be whether it is like a side cover or whether is an over cover so that'd be two animations then then on top of that it would be the

cover direction so it'd be whether you were facing left or right then it would be that we take into account the AI orientation but it was facing forward away from you and towards you then it'd be the lethal a non-lethal choice and on top of that will be the combat of stealth so add it all together thirty-two animations unique doesn't sound like a lot but when you think each one of those animations takes a day to a couple of days to animate then that's just the tip of the iceberg yes so that's this so unique in emotions but we didn't want to stop there so we wanted to apply this to open combat as well so we had the lethal and non-lethal choice the lethal would be using the Karambit and the non-lethal be using hands so be like choke outs and everything things like that lethal was usually quicker so again the player could choose to do that and it would kind of be more for the Panther or

assault player which is some of the play styles we talked about and like non-lethal form or the stealthy player so you can see them together and again essentially this is the same move it's just it's it's the criteria the we would choose is like so we chose to do the not only for version of this so all of these had to be dynamically loaded in game and again it was really kind of going back to the player choice so all of these would have the rest' attic version or a moving version the direction attack will be taken into account where there is your in front at the left or the right or behind them again the lethal non-lethal choice or the combat of stealth so it's 32 animations there add that to the other 64 so already we started to explode but we're not done so there's also the the aerial takedowns that we

called death from above the contextual takedown so they could be anything from like pulling people out of a Ledge hitting them against the wall various other things and also we decided to add a hand-to-hand finisher to the mark and execute system so there you go altogether 98 unique animations that could be called at any point at any time and we had to fit them into memory if anyone's seen if there's a video everywhere held there about all 98 takedown they'll be very interested to see ok but by far the the animations seen the most would be Sam's navigation so it's clear we need to apply the same attention to the basic navigation controls and early on in development we questioned the use of basically a manual Crouch players were staying in Crouch all the time so we kind of we looked at removing the actual option to not so we played around with the idea with the

automated Crouch and it kind of works you know and it gave the again cat references it gave the idea of Sam being stealthy and it was a great way to show Sam's awareness of surroundings you know again it kind of embodied some of the stuff that Kevin was doing and it would really make it look like Sam was aware of his surroundings but problem is is by removing the ability to manually crouch players would get spotted and they wouldn't be able to control it and to be honest really it just made him look like he was scared all the time because as soon as you got near to an enemy just they'd just be he just look weird but I really liked how it worked you know there was something about this system that I thought was really useful so we kept the idea we still kept the stand in the crouch but then we just kept the criteria so we just all we do is we we did this first test and you can see where this first test is all we did is as soon as you got into cover then he just reach and put his hand back on the knife that's all he did so you know if

there wasn't an AI there he wouldn't do that but then if he was he just reached his hand back on the knife in his life okay that works that's really cool we felt that was pretty successful but you know we just didn't want to do whenever he was static it's my whole obsession with everything in motion we wanted to make him aware and reactive at all times so basically that turned into what we called closer than ever so here you've got like the base animation set and this would be if no one was around Sam would just be crouching all the time and it would kind of look cool and again there'd be a stand bearing of this but then what we wanted to do is dynamically as you got closer to the enemy Sam would start to look a bit more stealthy and this would blend in over time so the closer you got the more stealthy he'd look and then also as you got really close he'd reach back put his hand on the knife and finish so each one

of these was starting to show Sam's awareness level to AI and everything and we've again we really really liked how this was working and this was actually an old system from a previous splinter cell that had been commented out and we decided to kind of bring it back and like freshen it up and and do this too so that was everything all put together and you can start to see that all the ideas that we're looking at start to come together so yeah so that's going to bring me to our AI behavior and to me one of the very first things I said to our leader I programmer was to me splinter cell was about I'm not gonna say the actual word I used but to me it's still about manipulating the AI and I was obsessed with you know making the a I'd do whatever you wanted to do and

kind of you know like showing into this over here and then you go and over there and I thought it worked really really well for stealth and to me AI communications and interactions were going to be a strong focus reactions to everything happening in the world we're going to be huge and it was important to show that the AI were going to be going around their everyday lives it's something that Jalil was talking about earlier was everything happening before and after an event and I wanted to show people going about the middle part so this is some of the things that we that we put in there and I had this this thing where I didn't want people to you know for stealth games everyone always has their back to you and I'm like ah that doesn't work for me so I wanted them to have a reason to have their back to you so I did things like washing the face nice and simple you know like loading boxes again it seems pretty obvious peeing was always a good one and the thing is is something that I found that I used to do when I was over on the phone was I'd walk around and I wasn't

focusing on anything else I'd just be distracted and I thought that was a really good thing so you could see like there we'd have the range or everything so again I have like you know have the guys kind of like playing games when they and again I just wanted there at the the AI to look a little bit more friendly instead of these these kind of these guys that had no lives I wanted you to think oh maybe I shouldn't kill him he's am I have to go home to his wife or something but you know that didn't really happen okay so one of the very first things that we worked on was impact and it was we found though because we wanted to improve the gunplay in blacklist and we found that the final bullet impact wasn't quite working everyone will go to rag doll a little bit too early and it wouldn't quite fix it won't quite fit so the very first thing we did was remove the blend this straightaway again snap

so we started with basically the final bullet impact that was the very first thing I wanted to work and I wanted to end on the end sorry I want to start on the end because I felt that the further you don't kind of like Leon and the professional is like you know the further you get was it the closer you get the better you are I wanted to do the opposite I wanted to basically you know start with the death and then kind of work my way back and I felt the these are the things that the players were going to see when they were looking down the barrel of a gun or whether they were watching someone or whether they're observing adding snap was important we wanted that kind of violent reaction of a bullet and we really wanted bullets to mean something and you know basically I think we had the blends about like maybe two frames something like that but it really worked we also added a bunch of animations for directions so this whole idea of

everything in motion went through to the deaths as well so we start to take into account criteria such as where you shot the speed you're getting shot when you actually get shot the direction they go from and suffer and this was really simple it was like this was so maybe this was basic stuff but really started to make this was before we did anything else to the guns we just changed the AI reactions that all those started to make everything feel really really good okay so the thing is with reactions as well is I felt that there was like an untapped kind of area that we could really focus on is like I you know I was fine with our pre combat animations looking great they'd already look really good in conviction and I didn't want to break that and the combat works but I was obsessed with that transition in the

middle I used to like in it when you're playing with your sister there's like the fun part is when she's scared is not the is not the actual part where she goes and tells your mom so yeah so we had things like double takes we added all these other things and you know I want to show you this it's a problem working in a game that is all about stealth is you can't see anything it's always black but yeah the whole idea was that is I wanted to be able to distract someone from somewhere they reacted in a certain way and I wanted that reaction to be believable I didn't want that to be just going oh sure I wanted them to kind of just nice and soft ball go over and

investigate and then something happened so this was kind of building up an ecosystem and you know is basically I want to kind of finish off with like a chain of events so I'm just gonna go straight you could see that there but the the AI was banging on the door and then in the middle of the banging we blew out the light bulb they kind of react nice and softly go over communicate to another guy and they goes hey don't worry abouts just a light bulb go back to you job so the thing is we could play with that so what we did is distract them again

[Music] again I don't know whether you could see it up but he distracted him in the middle and because this was the second time the AI remembered the first time and then would react and come to ology we pulled him out the window which would then create suspicion [Music] [Music] so now we create suspicion you can see that the the air went to investigate we're going to start to create fear and again we can play with this deer the AO realized that was wrong and the reaction

is fear but maybe we can use this to our advantage [Applause] cool so now so now we now we've done it in money to scare him then the a I've started to see this they're just like to communicate with each other by creating more panic the force sort the ante come here and have to be searched so the cool thing is is now they've got the back naturally turned towards you and then we can do something with that [Music] so you go we could have completely

avoided those guys but the whole thing is is for us we really wanted to keep these subtle reactions if you want to hear more about the AI and everything there's Martin Walsh is talking here on Wednesday at 2:35 and I'm sure he'll go into a little bit more depth tonight bill so I want to close with basically my points it was to me it's really important to find your muse Kevin was invaluable to some of the decisions and everything we made and without him we wouldn't be able to do some of the things that we did it's like I always like to try and find a reference for as much as possible if you've got someone to bounce off then that's really really going to help it's gonna you know so I think I'd like to do to me embracing optimization early would have helped us more we managed to get away with it but then if we'd really optimized early on we could have maybe added even more and no matter how big your memory budget is with the next generation of consoles there's going to be a cap so you've always got to think smart and of course for me we're making games game players

King this was something said Mario's talk this morning is again you can have all the slinky cat like animation but if it doesn't respond there responds like a dog then it's not gonna feel like a cat so so yeah so that's my main thing and that's my talk I think we've run out of time I don't know yeah I'm getting another run out of time I'm gonna be at the Ubisoft lounge tomorrow morning at 10:00 leave people in last week or I'm going to be hanging around if people want ask me questions and there's my my details if people want to stalk me or send me weird emails and stuff on fine with that so yeah thank you very much [Applause]