Super Powering the Saints: Animation Direction of Saints Row IV's Super Jumping

[Music] all right let's do this so I want to thank everyone for coming out today big thanks to GDC it's been a career goal of mine to talk here for a while and it's a privilege to be here today so my name is Zach Lowry I was the project animation director on Saints Row 4 and I've been in the industry for about nine years this past February and there we go so over the course of those nine years I've had the luxury of working on the Saints Row franchise every iteration every box title every extended content and for those that don't know Saints Row is an open-world third-person action driving combat game with a fully customizable protagonist Saints Row can potentially be an animators dream come true over the course of the franchise it's gotten more over-the-top and almost anything goes so it's really fun just to kind of set the tone for some of those over-the-top experiences that you might

have in a saint's row game we've had cyber space Tron inspired levels paramilitary Luchadors skydiving gun battles nine foot tall brute clones furry mascots who want to beat you within an inch of your life driving a car through traffic light tiger mauls you from the backseat missions on Mars and a gun that shoots a freaking shark now this is only the over-the-top stuff that happened in Saints Row 3 so when Saints Row 4 was being concepted the entire team had was tasked with one upping the over-the-topness of Saints Row 3 so of course our only response was holy [ __ ] so the internet often jokes where can they take the Saints Road next Saints Row 1 started and you were an up-and-comer on a street gang and you join that gang Saints Row 2 you went on to run that gang Saints Row 3 you taking that gang to pop culture icon status and everyone in the city loved you no matter if you were murdering or driving over pedestrians so where does it go from there it went to the only logical place

that it could go and that's that you became president of the United States aliens kidnapped you because you were the only thing standing in their way from world domination and then you were trapped in a matrix inspired simulation where you get super powers so this talk is going to focus on the unique challenges that the volition animation team encountered in Saints Row 4 when trying to build our super jumping system there were a lot of super powers in the game there was telekinesis super sprinting freeze blast all kinds of stuff but we're gonna focus on super jumping here because it's something that a proposed the most problems for the entire animation team across the project and be when all is said and donnelly and overcome our challenges something that we're all really proud of so a couple of takeaways and none of these are going to be mind-blowing by any means but hopefully you guys can see how all this kind of work together to get to the end goal so start with what you know I'm going to touch on that here in a moment kind of an unconventional source or use of referenced footage kind of a reverse engineering creation of pillars

for a feature and then the biggest one is failure when learn from when learn from which is the important part can be an invaluable resource for understanding and executing your final product so the talks gonna go through a couple steps we're gonna cover the animation direction overview we go through the challenges that we faced talk about the initial approaches that we took spoiler those did not work we're gonna go back to the drawing board and kind of rethink our inks and then we're gonna show what we applied and show off some final results so animation direction overview what was the animation direction overview due to an extreme time crunch between Saints Row 3 and 4 and th cues need to kind of push out another product it didn't change from Saints Row 3 to 4 which is kind of good because it allowed for some consistency between Saints Row 2 and 3 everything had changed and there wasn't any consistency so this was something that at first we were kind of bummed out because we wanted to change some things but we knew that it was for

the best so Saints Row 3 is animation direction overview was built around two very simple words and that was comfortable which meant that our player character nothing was newer surprising for the player character he had done everything multiple times and he had gotten better and better through experience and the second word was confident our player leaped before he looked he knew that he was the best in any situation no matter if he had never really that situation before he was more blue-collar and more street experience and he wasn't tactical and trained hence the pose right there with a guy just swinging a m60 down by his leg Will Smith's character Mike Lowrey who was in no relation to me was the epitome of our inspiration so much so that the entire animation team began using the mantra of what would Michael our II do when we were looking for animation inspiration and character performance so we face two major challenges right off the bat the first was winning superpowers with the previously established conflicting

animation direction it was going to be really hard to be comfortable when you were just given superpowers and this is actually the first time in the franchise where we were actually telling the person holding the controller that you have never done this before it was always kind of assumed that you would shot someone or drove over people on the streets before and then confident now confident and superhero there's a very thin line between those two and that may not seem like it's important right off the bat but it actually is because we weren't making a saint excuse me we weren't making a superhero game we're making a saint's row game where you were given super powers and there's a big difference there so superhero posing speaks to who that superhero is if you look at spider-man the way they bends his fingers the way that he flares his arms out even how he gracefully flies to the air he embodies those god-awful demon spawns known as arachnids Superman in the comic books he's always drawn with a smile on his face broad chest arms flared out he's a manly man he's very confident or the character Nightwing the way that

Nightwing fights or flips through environments and moves around speaks specifically to the acrobatic family known as the flying Grayson's we weren't making this we had a fully customizable protagonist so we had to speak to the breath of superpowers we couldn't pigeonhole you down this route and say you're gonna move like spider-man or this or that people wanted to create the superhero that they wanted and the second biggest challenge is knowing where to begin and when it came to superpowers our super jumping we were a very inexperienced team there were no no one on the animation team that was like oh we worked at this studio over here and we kind of made something similar to that so we're all just you know trying to figure out where to even begin and this goes back to my first key takeaway and that's starting with what you know so it was a relatively simple problem to overcome we knew that we had a simple Saints Row 3 jumping system and it was specifically animation driven because we wanted to get away from the physics dreaming jumping system of

Saints Row 1 and 2 and animators wanted full control over what that jump looked like Saints Row 3 wasn't about jumping so we just wanted it to look good so we had a simple take off animation that would take you back to the ground and if you didn't contact the ground you would go into a fall animation and then based on your velocity we played several different lands like I said Saints Row 3 wasn't about jumping but Saints Row 4 was definitely going to be about jumping so we took what we knew and we did a quick and dirty hack for the design team so that they could prototype this out we basically took that jump system scaled up the animations and allowed you to go higher in the air to give you a quick sense of super jumping that's clearly not where we ended up but we didn't want to be a bottleneck for design we want to just give them something that they could get in move forward the proof of concept while we had a little buffer to figure out what we were gonna do so while design was busy with their hack jumping system we went on what was called the great reference footage hunt we were trying to gather super jumping

reference from any medium that we could find we were looking at comic books and YouTube and watching movies and this quickly became known as the not-so-great reference footage hunt one thing that was quickly brought to our attention was that the matrix was a huge inspiration for the entire game so maybe we should start there and we all remember sitting in the theater watching Morpheus fly to the or jump through the air and neo do the same thing and it was all really cool so we thought this might be a great starting point but we quickly found out is that none of us would actually really watch the matrix in several years as you can see here Morpheus pretty much holds a single pose and is pulled through wires through the air there's a green screen involved and really what he's doing doesn't really matter the most motion that you see here is from the cloth simulation on his cape so this wasn't going to be our silver bullet so we decided to go back and start looking at some other things we found out that their action excuse me accident actually isn't a whole lot of super-jumping reference out there we watched a lot of clips on

YouTube and things were mostly flying more than jumping footage wasn't long enough it wasn't clear enough or there was just strange results these are actually two results that come up when you goo when your YouTube super jumping and we definitely didn't want to do either and it wasn't you know super jumping something that we actually couldn't go out and videotape ourselves doing or it could we yes this is actually me laying on the ground line animator stands on a chair and tries to videotape me jumping now this proved one thing and I was pretty foolish but we did try and learn something from everything that had failed and one thing that we learned here was that you know my center of mass or my route never really moves so this kind of could embody the fact that this is just being composited onto the screen so something we wanted to keep in mind that maybe we need to play around with moving that on so we went back to the drawing board and then what we wanted to do at this point was we were kind of sitting at a fork in the road where we needed to look at the technical investigation because the

programmers wanted to know how we're gonna break all this stuff down the way I still had to figure out the art investigation which is what I'm talking about right now so we wanted to approach this from a different angle and kind of focus on that tech investigation because we knew that the art would eventually come once we understood what we had to figure out so we switched mediums we're going to look at video games that had done jumping systems in the past and kind of approach it from a different angle so three games that we settle on looking at we're crackdown infamous and prototype so we started with crackdown and the things that we noticed immediately with crackdown was that it was simple and clean it was basically three animations almost similar to the hack system that we had given earlier they had a take off they have a state in the air that no matter how fast or how long you're following his arms might flail a little faster and then it eventually loops back to going slow and then they had various lands but the things that we we didn't want to take away from this was it we felt the

motions never evolved over time you were always seeing the same thing over and over and going back to me laying on the ground like a this characters route stays in the dead center of the screen the entire time but there's a big caveat with this crackdown is a great game and I love it dearly but crackdown came out in 2007 and we were coming out almost seven years after that so gamers were gonna expect more than this they wanted bigger and better and while this worked for them it wasn't gonna work for us we wanted to take our findings and build a prototype of this in our Saints Row engine and this is a super quick and dirty prototype by the way and it actually confirmed everything that we felt it was slow and basic it wasn't going to be competitive the motions weren't evolving enough over time and this was not going to be fun for hours upon hours so a little bit later in the video you can see that we started to play around with FOV and a sense of gravity and that was something that we also learned in here that again moving the character on the screen allows you to to feel like they're actually part of that environment instead of superimposed on top of it so

from there we went to look at infamous and we quickly learned that infamous does not have a super jumping system but what they had was really good and we wanted to keep watching it and learn from this we love the fact that no matter where the jump was done whether it was over a car off of a post from the top of a building it was the same jump it was simple there were no weird middle heights in there that didn't blend well and this is something that we wanted to jot down and take into our system and from there we looked at prototype and it's immediately noticeable that in prototype everything is very primal it fits the character and the supernatural back story that goes with prototype we love the sense of speed that you had on a scent and ascend deceleration and then we also pointed out that this is very superhero something that I touched on earlier and we couldn't do this it embodied their character but our character was kind of a whoever you wanted to be but we took

these findings and we built another prototype and we wanted to test out that superhero premise to see just how superhero he could be and it quickly proved not very much our characters moving around very similar to spider-man in here so it proved that point pretty quickly one thing that this also shows is that we understood the technical side of how to break down the anime while the visuals aren't what we wanted the tech side of how the animations were broken up is pretty much exactly what we ended up shipping with so after these tests we decided that we should try and incorporate everything that we learned and go into the concept of slide not working great creation of pillars now pillars are something that oops too soon creation of pillars is something that is normally done at a project level and we really wanted to bring that down to a feature level so we created four pillars motions must be grounded in reality we started to realize that the reason we couldn't find good reference footage was because we were looking for something that wasn't grounded in reality but our game was so looking for super-jumping

was always going to result in superhero stuff our motions had to evolve over time if people are gonna be doing this for hours upon hours we wanted to kind of mask the fact that you were seeing the same thing over and over the motions must look good and feel good at all Heights and speeds and emotions must fit the overall theme in animation direction now that last one's kind of beating a dead horse of this entire presentation so we took what we learned and we wanted to go back and look at stuff that was grounded in reality and I'm gonna fast-forward a little bit just jump to where we actually ended up and that was with Olympic long jumping we found out immediately after looking at a lot of Olympic long jumping reference that if fulfilled pillars one the motions were grounded in reality pillar two the motions changed drastically over time his body's probably going through three or four motion changes as he reaches his apex and then comes back down and pillar forward this could fit our art direction it was grounded in reality but we could push it to become a little bit more heroic and we could show a little bit of

his confidence in there so from here we were failing pillar 3 which was that it didn't look good at all Heights because what we took was our art direction that we merged with our tech investigation and we built a fully charged jump now in Saints Row 4 if you played it you know that you can hold down the button and as long as you hold it that determines how high you go so we decided to start with building the highest jump that you could possibly do and these are senic some snapshots here kind of squashed to fit on the screen but this failed pillar 3 because when you tap the button and you only jumped like you know two meters off the ground or whatever it looked like he was exerting himself way too much so we didn't want a brute force this but because we were also running out of time and the project needed to move forward we actually did exactly that we brute forced it so we built the exact same motion path just at a lower height - and changed up some of the posing so it didn't feel like you were exerting too much energy so it was

nothing amazing and it was just part of game development and then we took these to various height jumps that we had built and we kind of merge them together in our animation Network and depending on the heights in between the the minim and the max we would merge those two together and this is kind of what the simple version of the super jumping system looks like emotions are grounded in reality they're evolving over time and they look and feel good at all Heights and speeds and emotions also fit the overall theme in animation direction you can kind of squint your eyes picture any customizable character in there feeling feeling satisfied by the amount of freedom that we've given you here so I want to touch on a brief bit of the tech side just to show why we started will be new and how we kept it simple now what I just showed you was a very basic form of the jumping system so this is kind of a sample of our motion graph so we had a jump state down at the bottom left there and this was

controlled by controller input based on your input this determined whether you were trying to do a standing jump a moving jump moving high jump double jump and then from there once we understood that we would fire off a take off animation there were a few different take off animations but they were basically just can transitions and then depending on the the amount of charge or the velocity that you had when you were moving we would go into a falling or jumping state as we called it this was all velocity driven it was multiple transition animations that were dependent on your velocity or the jump that was executed and we would ramp into different parts of this or even scale the timing within there based on those inputs and then similar to the same throw three system that we started with based on your velocity we played a different land animation so in a nutshell this was the super most basic version of super jumping that we had and it was good that we started with a simple version because as we all know something that starts off simple always ends up most complicated system known to

man because quickly after this amount of freedom was put into the game designer egg comes into the room and he says you know it's great that I can jump around through the city but I can't quite land where I want to land so it'd be great if I could dash through the air so then dash was added and then the designer egg comes back the next day's like you've you've given me - but now I kind of need something else and I know I know scope wise we said we weren't going to do flying so I'm not gonna ask you for flying but if I could kind of throw my arms out in the air and glide for a little bit that would be great so then we give him gliding and then combat designer comes in the room and he's like you know all our combat takes place on the ground but you're jumping all over the place and I can't you know throw a nice blast at anyone or use telekinesis so you think we could do that in the air so of course that gets added and then you're come your designer egg comes back and he's like I'm dashing and I'm gliding and I'm jumping I keep hitting walls and I can't jump on the wall so you think we could add wall jumping so

while jumping gets added and then he comes back the next day and this one's gonna make every animator in the room throw up and he says you know our takeoff animations have about four frames of anticipation and I know that's been a battle what I really want to do is hold the charge button down lime in the air so when I land on the ground if I let go of it I have no anticipation and I rock it back up into the air so maybe we could do something with zero frames of anticipation so we ended up going with that and allowed double jumping and then combat designer comes back and he says all this is great but I really want to be able to hold a button down pause everything that's happening in this network hold my fist in the air be able to aim 360 degrees and then fire myself off like a rocket so death from above was added so as you can see something that started out so simple became the most complex network outside of our player locomotion that existed in the prod

and this would have been a terrible thing to try and noodle between if we loan it out from the get-go each line here is a transition between multiple state nodes that contain blend trees other complex state machines inside them and the parameters such as velocity character heading desired direction we're all controlling the shift between these states this is this kind of what it all looks like when it comes together mmm that Network also didn't include wall running which is seen here which was handled in a different method and came online about two months before we shipped but I single-handedly believe that if we didn't add wall running all the comments such as you know the most enjoyable part of Saints Row 4 just moving through the world those probably wouldn't have been said so it was something that the animation team actually proposed to the designers and they really didn't want to do it but we did it anyway and then they were very thankful that we did it so in summary

the challenges that we faced were blending superpowers would previously establish animation direction that was conflicting finding reference footage for motions that were not humanly possible in experience with this type of system and some of the solutions were to start what we knew develop a set of pillars for the feature explore reference from all mediums and different approaches to how we use that reference and then the biggest takeaway again is that failure when learn from can be an invaluable resource for understanding and executing your final product I want to give a big shout out to the entire militia and Deep Silver Koch media without Saints Row wouldn't be here today and I want to thank everyone again for coming out that's all I got and there's about four minutes for questions if we have any

Bueller Bueller No hi did you guys did you guys do any explorations with physically based or procedural animation no we didn't I go saying we were do kind of in a big time crunch trying to get same syrup for it originally started off as like an expansion product and then THQ wants to turn it into a full product so we had scrapped a whole lot of what we had and we're rebuilding it in a very short amount of time superpowers are something that everyone wanted to keep so there was really no time and we actually didn't have the tech up and running to do something like that can you talk briefly about the way you blended between animations and engine yes so we have a proprietary like toolset for building our state machines and stuff like that is something that we've built internally I kind of touched

on it briefly it was the designs big request that they wanted was that you know controller input determine how high that you went and we we tried a few different ways to where it was like as soon as I press the button I take off and as long as I hold it I keep going up and that was just not very fun and it was too hard to particularly you're going so we basically built the you know super high jump and we built the the super low jump if you will and we just had parameters controlling the blending that was basically like what is what is the input coming from my controller and there it really wasn't anything magical it was really just like a linear blend between there's oh all right well my time is about up so if anybody once they ask me more questions that they didn't think of right now my Twitter's on there love to talk about it

and thanks again for everybody coming out appreciate it and have a good GDC big thanks to all the CAS and GDC again thank you [Applause]