How Cameras in Side-Scrollers Work

okay hello everybody today I'm going to talk about the theory and practice of cameras in side-scrollers and 2d games in general side-scrollers is a it's just a private case of what are you talking about there are a lot of parallels in to 3d games but we're going to focus on what's interesting to me and specifically for the game that I'm working on which is mushroom 11 I've been working on this game for the past three years and still not ready as you can see there are quite a lot of issues with a camera that needs to be tackled which kind of threw me into this journey of trying to figure out how to do proper camera systems and the result is what

you see here and I'll talk a little bit about that later on so scrolling we have when we have more to show than what the screen supports will obviously need to leave some things out of the screen but when we start moving the screen there are some neural aspects that we need to to figure out so first of all let's talk a little bit about the neural aspect of squalene we review the image through our eyeball and the image is presented on a set of cones and receptors in the back of our eyeball called the fovea centralis the powerful and very folia the centralist is in charge of the center of the image gives a very sharp image but kind of low frames per second so to speak there's a secondary belt and then the third and biggest belt with the the most cones is the parry phobia

responsible for the peripheral vision it has a very low density and but great depth of color but in order to get all this information from such a huge set of receptor it actually compresses the images into patterns and the patterns are visual textural patterns as well as motion patterns and changes in velocity that information is fed like other other pieces of signal signals from the brain through our Thomas and then instead it's sent to the visual cortex as well as the amygdala the amygdala is the part of our brain that's in charge of emotional reactions and inhibition of those reactions of fear and anxiety what we call the fight-or-flight by teaching in our brain to ignore those signals eventually our amygdala amygdala knows

to to ignore those those signals and just let the visual cortex take charge and when we were talking about visual patterns in our peripheral vision it could be something that just reminds us of something scary that we've seen and the corner of our eye could be something that moves really fast or it can also be the entire image which is the background of the screen moving really fast on top of that we have what we call the vestibular system which is our internal spatial orientation balancing system allowing us to basically stand straight and then when this thing is coupled with the vision it basically gives us a full orientation of where we are how to walk and what and what not now the amount of information is given to our ascent to our brain we try to minimize that

information and in fact even if you try to move your head really fast you'll see that your eyes will use what we call the vestibular ocular reflex trying to minimize the amount of changes in our in our view we are not wired to see the image move really fast in fact when we when we have any any conflicting sensory signals between our eyes and our vestibular systems system then it may lead to nausea and discomfort of course this everybody knows it when they well some people are very sensitive to reading in a moving car and this is even more dramatic when when playing first-person games or virtual reality obviously this is a much more dramatic so how do we properly properly scroll I split it into three

main groups tension interaction comfort attention is making sure that all the necessary gameplay elements are in the frame interaction how do we make the cycle that we take to teach the amygdala to know that those changes are predictable and tightly bound to the controls and those changes as they are made and they have to be made they need to be easy on the eyes those of course are all part of the same equation they all fit into each other also throughout this presentation I'm going to go through my own personal nostalgia of games all the way from the 80s that how they start working with scrolling as well as the indie games of today where all the innovation and and the care and attention that you can see in only in indie games really so how do you scroll

first of all obviously you need to follow your character so if we go back to what's considered the first real scroller it's a game called rally-x from 1980 it did the only thing that you could really do which is follow the character I call it position locking the camera means with a character nothing interesting here except with the fact that scrolling at the time was really a you know a difficult feat to achieve there were issues with the fact that the solution was really low the CPU was very slow there was segmentation issues you couldn't display more than a few sprites per line bunch of issues so this technically is a huge breakthrough by the way you'll see that throughout this presentation and I'll present my own terminology just from my own reference and I'll definitely post it later moving on the game called kung fu master use

the same idea only horizontally you just lock to the character as the level ends you'd snap to the end of the level obviously this is so obvious that I'm not going to really talk about edge snapping I just wanted to you know just mention it quickly the just a quick shout out to Takashi nishiyama the guy who made this game is he made this game between making moon patrol two years before and Street Fighter one three years after so this this technology still works very well today some games really benefit from this kind of the behavior especially when you have a smaller character it doesn't move really fast you need to see everything left and right up and down terraria is the perfect this is this position locking the perfect technologies here so how do we curb camera motion in order to eliminate those unpredictable bumps if

we go back to 1981 there's a game called jump bug that was still considered the first platformer it introduced a system that I call camera window which allows the player to move within that window without moving the camera without scrolling again scrolling was a really problematic thing to do and with the CPU and and the resolution so you would move freely but as you touch the edge of the window it will start pulling the camera with it so this system is a pretty problematic you think about the fact that you're moving away from that window which leaves you very even little room to see what's ahead and what's up and you know the exit of the level is up one way you could do it with is like they did with the Rosslyn saga one of my favorite games of the 80s they just placed their window in off-center

allowing for 50% view forward and downward this is great but the problem is that we need to start backtracking or if you jump and end up in the middle of the window you'll have even less room to see what's coming ahead or above well this system is still used in Fez it's used perfectly in fact this system is maintained the horizontal window is maintained even throughout dimension shifts and we're going to talk a little more about this game and the camera system here so we talked about when you showed Russ on saga we showed that the character could end up in the middle of the window how do you snap it so you Center the player back into the center of the window my favorite game of all time if I may say is shinobi from 1987 it used a horizontal locking system but

vertically it had a really long window because it fit the game perfectly the the player is pretty big on the screen the jumps are really high so they try to keep the window as high as possible the problem is that the player would potentially be off-center really quickly so what they would do is every frame when the player was off-center they would start slowly to focus back on to the player keeping keeping you always in check and we're going to talk a lot about the work of Shigeru Miyamoto but going forward a little bit a Super Mario World it was the first one that I noticed that introduced a platform snapping and in this game they really tried very hard to maintain the minimal amount of camera movement and in the case the window is the entire screen the

camera window is the entire screen it will only move when push the edge of the screen unless you land on a platform where it would quickly snap I think this is a obviously a brilliant idea some find landing and in the snapping a little abrupt but that's the first one that I noticed it used it you can see you can see a more regulated version of this in Raymond and throughout the Raymond franchise you still see this today awesome in outs is you if it from 2012 used the same technique obviously there's always a problem when you have characters that don't land on platforms so like the jetpack character would be pre snapped basically keeping it position locked both both horizontally and vertically but the motion that we have to do either horizontally or vertically or when we touch the edge of the window how do we smooth that that

motion and may I remind you that in the 80s we we had a very low resolution we're talking about 256 by 240 at the NES so the pixels there were huge today when we're talking about pixel art we can we need to remember that the pixels that we use actually contain a lot of screen pixels and that's a luxury that people that the 80s didn't really have so for example if you look at pac-land from 84 which this is actually considered as the first real platformer it didn't really do anything special with regard to smoothing the window it just locked to the position of the player but the player actually created some sort of pseudo physics and that that physics was a smoothing the camera on its own when we're talking about pseudo physics obviously even today everything the suit of physics but in this case we didn't have you know they

didn't have damping and and the friction so we're talking about accelerated from one pixel per frame to pixel three pics and so on dan slowing it down back again now I'm gonna do a quick shout out to Ichigo Miyamoto and remind - is first early experiments in scrolling in 1984 he made a couple of games that introduced screwing in some way one of them was a Excitebike which the scrolling was kind of the background of the game you didn't really have a lot of the game was not about to scroll and you didn't really have a lot of control that's going devil world from the same year had everything to do with scrolling in fact the game the devil was controlling the scroll in this really weird pac-man clone and this game was I don't know if it's considered one of the best work of Miyamoto in fact due to

some weird religious iconography this was denied release in North America and the reason why I'm reminding this really quickly is that I don't know more in a philosophical way if you look at some of the early works of the geniuses of today seeing some of their more mediocre work is kind of encouraging to everybody and and and I do that a lot sometimes when I feel down maybe then what may be the the the biggest accomplishment and this is completing my own conspiracy theory is that I believe that there's some cross reference to bubble bubble from this but I'm not sure this it's not official alright so moving on to 1985 Miyamoto actually did his first true experiment in scrolling and in Super Mario World the character was moving really really fast and if you think about it this is a

one-sided window you could backtrack for the entire half of the screen and then when you would go back and push against that window it would move from zero to maximum speed in one frame and that would feel very uncomfortable what he did is something I called speed up push zone so it would as it hit certain point on the screen it would start spinning it up so when Mario hits that anchor point the camera is already in full speed and this thing I actually didn't notice it for 30 years because it's so brilliant and so clean and I think this is a just genius solution another solution of the same idea but kind of the opposite is what they did with Metroid they had either horizontal or a vertical window depends on the level and it allowed the player to step

out of that window and then start catching up with the speed of that window so the character would always be a little bit couple of pixels depend on the speed depending on the speed out of that window but that work community well it didn't leave a lot of room to see what's ahead but that's the problem with camera window in the first place when you're talking about smoothing of course when there were more pixels to play with and you know use a floating-point became a little more easy you can start doing allure ping which is stands for linear interpolation the idea is that you would in every frame you would reduce the distance between the camera and your character by a certain factor they did it with Dante key Kong country that's the first game that I've noticed that they did it properly but they do it ever since you can see it in almost every game today Super Meat Boy uses it it's one of the best ways of easing the

the camera motion if you think about it LARPing is ease out there's another thing which is ease out if you know the term the idea is that you ease the camera into motion and then out of motion again by applying some actual velocity to the camera itself using for example smooth damp function it's used in the amazing never alone game from 2014 and many other games of today it works very well when you need to to have some drama and narrative and smoothness that some games really benefit from and if we go back to pixel art what they did with their doing with hyper light drifter coming out soon hopefully they they actually have true a pixel perfect art that they present on a canvas and that canvas is is moved on

the true screen pixels so you could you would still have the pixel perfect big game pixels but the scrolling is the only thing that breaks that pixel perfect behavior by applying the scrolling on the true scrolling on the canvas so framing how how do we make sure that the player sees everything they need to see so auto scrolling as you can see is not exactly a part of this presentation because I feel like well if you don't have control of scrolling it's a different kind of question but what they did was scramble which is also considered one of the the first scrollers it gives you a good sense of what the player would do if they had a true control of full control over where they want the player to be you have the entire screen you have things coming at you you want to be as far away from the

edge as possible but still leave a little bit of room behind you to make sure that nothing creeps up and that's more or less what they did with pac-man instead of locking it to the center like they did with with other game like kung fu master they put the the anchor point a little bit off-center because that's the principle direction of the of the player you would still be able to go back you don't do that a lot but you still have a little bit of room the idea in general is that the player would want to go to the center and you'll see that throughout the presentation in general you always want to aim towards the center another way going back to 1981 is what they did with defender they had what I call the dual Ford Focus you had Ford Focus that's based on where you're heading as as soon as you switch it will immediately start snapping toward the other anchor point

this game is way ahead of its time it took a lot of a lot of years until they actually figured out how to do this properly this worked really well even back then okay second technical difficulties sorry about that I'm running off Google Docs nightmare okay all right let's see if we can do this bear with me all right so um it's okay for now trying to connect okay so going

forward a decade or so Bonanza Brothers is another type of implementation of the same idea but unlike what they did with defender from it from 1981 they instead of having the switch abrupt and immediate they actually made the change based on your walking speed if you look at the character at the bottom you would see that the switch between the anchors would be based on the speed of your your character and you wouldn't even notice it when you play unless unless you start placing some lines over there so going back to Miyamoto again I'm going to talk a lot about a lot about this game this super super mario world it had a lot of interesting ideas this is life my favorite one it also had a dual threshold but their true genius here is

that it actually let the player backtrack a little bit before switching and to the other anchor and this is perfect for this kind of game because this game actually allowed kind of encouraged the player to to do all sorts of back-and-forth motion so instead of every time that you switch back and forth forth it would switch to the other or starts moving start moving to the other anchor in this case it actually lets you walk a significant amount before hitting switching to the other anchor a game that does this today not exactly this it's closer to what Bonanza brother actually did is cave story well it's not exactly today um you can see that this game has requires a lot of for forward view you you have a lot of things coming at you and you would really require a vast a clear view of what's coming ahead they

did more or less the same idea they had two anchors and as soon as you would switch it would start slowly walk and go to the other anchor based on based on your walking speed yes another thing that I notice is what what I call target focus and the reason why I call it target focus is because it actually Maps through the actual target of the player how do we know what the target is is where what where the player really wants to go what they did with a jazz jackrabbit - from epic back in the day they actually they move the camera based on what the player and what the player did so if the player pushed the stick left they would move the camera left if they did it right they move slowly right until a certain distance this is pretty brilliant because it really Maps perfectly to what the player wants to do and I haven't seen this done

a lot the good thing about this idea is that it actually works up and down as well because in some cases you want to see what's what's going underneath you you would do the same exact thing you would press the down cursor key and it would react exactly in the same manner so and we're talking about target focus of course it doesn't is not limited to cursor keys or controllers some games that introduce Mouse action like a snapshot which allows you to capture elements from one side and move them to another and side so if you already have the mouse the mouse is a good representation of your actual view you know where you're looking at so why not put the camera and this sort of average between the character and the cursor that is very clear that that's where you want to go in the same way as they did

with the the swapper an amazing game that has a bunch of great great camera behaviors so if you have a gun or in this case a cloning device that aims to some place this is where the player is looking at this is where our attention is directed in fact you would sometimes want to look to some direction you will go to the other one and you would want the camera to to follow the looking direction and not the walking direction so this is a perfect solution it also obviously requires either in a digital some sort of additional control either a mouse or an additional stick projected focus doesn't work necessarily very well for platformers it could work for horizontally but vertically you have differential that's very abrupt if you think about it about jumping jumping is changes the speed that your vertical

speed from 0 to 2 X in one frame that doesn't really work very well and try to project the speed but in all sorts of flying elements like they did with secrets of red econ from last year this works really really well you just aim for where the player is going to be in in a few milliseconds a great representation of very intelligent system it can be found in the luftrausers I guess I spelled that wrong I'm sorry it is kind of weird so yeah they have a really interesting combination of target focus and projected focus projected focus as we said it looks to looks at your actual velocity target focuses it looks at where you actually aiming so they found this nice combination of those two always giving you a good sense of where you're actually going sometimes you were

aiming to some direction but you're thrusting to get to something to someplace else so they found a good averaging of that of those behaviors on top of that they have other elements that we'll talk about in in a sec all right so so far we've been talking about what the player needs to see but what about what we want as the directors of our own games we want the player to see we have some choice more in the narrative aspect one of my favorite games of the 80s I'm not that old and just bear with me is the Wonder Boy that game actually introduced something that this was very modern at the time and even technically it's not very easy if you think about how to do this it would move the camera along a path that was not parallel like other games of the time in fact it would not only not always Center it it would sometimes just

move the camera up for example and keeping the boy at at the bottom kind of foreshadowing something that's coming up ahead that gives us as level designers much more freedom something that we can do a lot with going forward in almost a decade in the birth of the fifth generation of consoles this is the PlayStation it cloning off from the PlayStation they started having two and a half D or three D in general and two and a half D in in particular so the camera path that as we've seen is with wonderbook could be much more versatile it doesn't have to be necessary position its position and zooming and tilting and whatnot you can use it in any any way you want you could start moving towards certain elements that you want the player to notice or or zoom out to give some more interesting dramatic effect

and works very well for that grain this is super important and not a lot of game did that at the time donkey kong country has a bunch of standard elements like lerp smoothing we talked about that a vertical position locking dual forward view horizontally but the interesting thing that they did is that every reach every level has multiple regions and those regions would would each one of those region would set different anchors some regions would be more linear go from left to right other ones would be more exploratory those would actually place those anchors in the center kind of giving you an idea where you need to go and that was very clever and you could say when you play it you can see the camera would shift dramatically between different regions another thing that you could do back at the time when the technology kind of started allowing that is using zoom to

fit you have a certain element that you want to including the screen no matter what like the boss then you would zoom out to make sure that that element is always on the screen so if you're the player and you see that it starts zooming out you immediately know what you need to do that was Yoshi's story by the way now cue focus and I think this is one of the most important elements that we can learn from from other games and other games has done this before but insanely twisted shadow plant really made this into a almost an art form and in fact they had a blog post exactly about this technology and that's where I got this footage from what they do is they have what I what they call attractors that as you get in close enough to those attractors it would start slowly shift the camera towards that attractor based on the distance from the center I call it double ring attractors because you have this outer

ring once you're inside that ring it will start every doing some weighted average between the character and that cue clos and the closer you get to the the center of that cue it would get closer to that queue if you immediately know what you need to do because you see the camera will start as you are as you get into the center ring it will lock on that cue in this case it's the check point and the good good thing about that you can have a bunch of overlapping outer rings so the camera would average between all the different elements this game has a bunch of other elements that are very interesting such as them using the the gun that they have as a target focus representation of where you're actually looking at you should check out that blog post another way of doing Q focus in a very innovative way is what they what Edmund McMillen Tyler Gayle did with ether they they have as soon as you

get into the gravity zone of any planet they would start slowly shift or tilt the the screen so downward would lead to that gravity source which is very predictable very clean and gives you a good sense of drama and narrative and it's a perfect solution and something you don't see a lot and no presentation on camerawork would be complete without talking about limbo it's really a perfect combination of using different like very simple camera system which is a simple position locking if you think about it but then the anchor changes dramatically between regions in the same level even well it's all the same level really and and you have a bunch of cues and those cues are set on or off so all of a sudden you have like a boss coming

into the screen and that would take the attention or is used as a tractor to get the camera closer something else that I call Regent focus is in this case you can see that the camera is based on some sort of average between the center of the region there's only one region at the geometry wars really and the character so there are a bunch of different ways that geometry wars could have scrolled they chose this one they could have used camera window they could have used any type of target locking they use this one because it actually gives you a sense of where you need to go where is the center you always know where you are in relation to the center and that's really important and you're always on the move it's also kind of a dramatic effect the same kind of idea combined with other element is what they did with a vessel from 2012

this is a very heavy puzzle-platformer a lot of things happening at once for a further four in each puzzle you need to see a lot of different elements and each of those regions it's like like room puzzles that you would see in earlier games but instead of having that they actually have anchors that focus and zoom into or out of certain elements to make sure that everything that's important for that specific puzzle would be shown for would be shown to the player but on top of that what they would do is they wouldn't just freeze the position of the camera on the center of that region if you notice the motion of the player actually takes the moves the camera slightly so there is like a small extra slack that they allow the camera to move around their region and this is pretty brilliant it gives you

the sense of control you're actually moving the camera as well but also as you hit the next region or the edge of the region that you're in the camera is already in motion so there's never really unpredictable movement of the camera one other thing that it was done in the first days of PlayStation tomba from 1997 used something I call gesture cues which is elements that are based camera movement that is based on what you actually do there are a bunch of things in that game that changed the camera behavior one of them is when you start climbing the camera would shift behind you give you more sense of control and drama you know you're just looking ahead you can see what's behind you giving giving a lot of control feeling of control if you did this with as test in many many different elements this is a great fighting game that's supposed to come out sometime and

and now I have the same problem and a lot of the different elements cause different camera effects and that gives a lot of sense of drama control and also kind of lack of control when some of these things happen you said you feel almost powerless sometimes when we see these things happening especially in succession and I feel like this is one of those things that we're just scratching the surface with gesture cues moving the camera based on what you actually do now we already have a camera why not use that camera to actually tell a story I call this cinematic path paths so instead of having a cutscene telling something that that happened someplace else why not just use that camera it's also I also almost see this as a lazy solution but it's exactly the one thing that you need to do they did some sort of like is something like that

in Streets of Rage from 91 and the behemoth did this with Alien Hominid and Castle Crashers they do it all the time and it achieves a few goals one of them is obviously tell a story but also provide a direction of where the player needs to go so it's two in one and you you feel like you're playing the whole time there's no cutscene the game is running the whole time alright so what do we do when we have more than one thing to focus on I think we all played this game gauntlet from 1985 it's a pretty amazing game the technology is pretty well ahead of its time the scrolling is wonderful so many things running at once there is slight problem when you have more than one player and we all know this what happens if one of the players just slacks off or just tries to go to the other direction everyone is everybody's stuck I call

this position averaging that's pretty clear you have a lot of things at the same time you would try to average to go to the average position but that's an obvious problem this operatives problem could be seen also pretty clearly I mean this is I really love this game it's a pretty crazy game but what happens when you have all these players die and respawn or move from one side and then reappear on the other one also listen the average would just shift dramatically so it works really well for summer I gun is as a player but if you're trying if you're not a player and you try to watch someone else play it could be a little uneasy on the ice one solution is what they did with Street Fighter 1 and then throughout the franchise they created the camera window the good old camera window if one player pushes through pushes the edge of the the window it

would push the camera obviously but also push the other player so the only way you could actually get stuck if both players forcefully try to push two different directions it works for some games definitely works for one-on-one fighting games like this and obviously the best solution is what they did with Super Smash Brothers 1 and throughout the franchise you have you average the position add some additional smoothing and you zoom out to make sure that everything that's important is in the frame yeah one Rockets Rockets Rockets which came out last year it was also a great representation of what they do do is Super Smash Brothers and here's kind of the develop designer choice as well it will always try to zoom out the minimal amount of zooming so the player would not really see what's behind and

also if you get back far enough it's actually uncomfortable you don't really see very well so you have this implicit encouragement to always go back to make close encounters which is exactly what the designers wanted another solution is what they did is spelunky obviously I the solution is not even get to the not even address the problem of multiple characters you just give the camera to one if that person dies the camera would move to another player and that's super important for a game that if you go out of the screen you actually has some grave consequences and everything that pertains to multiplayer would would would fit multi character and you don't have to do a straight 5050 average you can do whatever forty and something else

manual control when would you provide additional control to what is implicit by the camera movement so I think this might be the last time I'm going to talk about Super Mario World and it's it's pretty interesting I feel like the fact that they introduced so many different systems in one game is some sort of a overcompensation for a really bad system that they had in the Super Mario Brothers 3 or something on top of everything they did they actually had an additional manual control the shoulder buttons would move the camera as well I don't think anybody really used it a lot I don't think it's really needed and it's not implicit it's not very clear the player wouldn't would even know that this thing existed this is obviously not a real I mean it's not that your classic side scroller but what they did with Osmos is using control that's very clear

and implicit to produce exactly the results that you would expect in this case the mouse wheel or or in tablets that pinch-to-zoom would do exactly what you would expect that's what they did with jazz jackrabbit II they use the left and right keys or stick to move the camera but the same exact behavior would would move the camera up and down which is very useful in some cases they did the same thing with spelunky you just expect that if you and you would probably do it do it if you didn't notice as you were climbing or crouching quickly I want to talk about camera shake camera shake is kind of the the opposite of what we're talking about right now it's providing some sort of lack of control the player has no control there's something happening on

the screen that's bigger then then the player could actually control so I try to look back and with help of friends to figure out when was the first implementation of camera shake this is a the farthest I got and of course it's Miyamoto the it's not exactly the camera shake that we know of today because it's it actually has a some part in the game it's not just an effect but still the camera shake close to what we know today the funny thing is that you would only see this behavior on the arcade you wouldn't see it on the NES they probably tried and couldn't get scrolling to work and I have this theory that that's when Miyamoto decided that scrolling is something that he wants to pursue so I built an entire story in my head so anyway if we're talking about screen shakes and screen freezes Vlambeer and

they know best they made it into an art form really and I had some talks about that if you look up gee devil JW's work it's it's a huge subject on itself alright so how do we put everything together and make our own custom-made camera so first of all we look at what we want to do what's what are the characteristic and significant important features of our game then we get inspired by other games as I try to do here as well and then we make it our own change what we need we take some tips we combine and make our own camera solution this game from 1985 legend of Kage doesn't have anything amazing going for it except for the fact that it's a probably the last game going from right to left for some reason except for mine what

they did is they created a camera window that's exactly the width of a tree because most fighting actually happened on trees so once you're on a tree the camera would stay still and again not a brilliant camera solution necessarily but the implementation is ideal for that specific game as we've seen before with shinobi if you have high jumps use a very high vertical window but remember to to snap it back to the center if you have a character that moves really really fast in all directions up and down maybe use a camera window that's very narrow that would work really well especially since most camera and most direction changes happen on the ground so you wouldn't really see the bumps from one side to another except when you've flying down into the bottom edge

of the screen which in this case is kind of abrupt and this is exactly what they did they solved this in so that was sonic once in Sonic 2 this is they solve the same problem by using what Super Mario Brothers one used that's the speed up push zone each one of the halves horizontally and vertically stops the motion of the camera and then starts easing it out back again so even when you go into a high-speed loop you wouldn't even notice that the fact that it actually reduced the speed in one side immediately increased it back and it feels very very smooth and it's a major change specifically on the camera system between the two games long distant view we talked about the concave story I think that's a perfect solution for games that require a lot of forward view I won't talk a little bit about Fez again for games for platformers we don't really see camera windows too much these days because you miss out we miss out a lot of information of

what's coming ahead but that's not fast Fez doesn't have a lot of things coming at you in fact you would I I think that most of the important elements since most most of the dimension shifts happen in corners of buildings so you could argue that most of everything that's important is actually behind you and that that is maintained throughout the dimension shift so this solution the camerawindow solution that works throughout dimensions it is the perfect solution for this kind of game if you have elaborate causal systems with a lot of things happening at the same time make make the regions as your the jurisdiction of where the camera needs to be but you can easily do what vessel did still allow a little bit of motion some control of the motion to the character giving them more sense of

control and if you're telling a story like we've seen with Alien Hominid the swapper actually made told the entire story based on in-game camera shifts and region changes and cinematic paths the entire story is told that way zooming in zooming out changing of course and lighting and whatnot if you want to tell a story always already we ordering in the game just use the camera to tell your story so what do I do in mushroom 11 so mushroom 11 has multiple regions per each level you can see that each of each of those regions is very very basic rectangular area in could be angular as well and and the camera can only move in

that specific region also each one of those region has a zoom factor a field of view factor that allows me in some cases where there are a lot of things happening I can zoom out give more attention to two things up ahead all these regions are numbered so the first region the first mushroom piece to go through the next region would take the camera with it there are two main region archetypes one is the center oriented which is basically the camera goes to the average position of all the different cells and the other archetype is the progression or oriented the regions where those those regions have a trajectory in this case it's just from

right going from right to left and I'm looking at all the different cells different mushroom pieces and the one that's most progressed on that line on that trajectory gets the camera and since I know that I'm only using this behavior this region in in cases where it's more linear motion not really puzzle I know that I can apply an additional static forward focus I'm moving the camera even further based on that trajectory in some cases there are additional cues like the tractors that we've seen other games in which case I would average the position based on a certain weight that I decide on between certain elements in the game I in this case the tip of the boss and the control mushroom and of course on top of everything there's an interesting

behavior for smoothing I'm using projected focus obviously there's no control here the only thing I know of each mushroom piece is its velocity really I can make it jump rings so I'm looking at its current trajectory and then apply a further physics smoothing on it the interesting thing that took me a long time to kind of figure out is the physics smoothing factor is based on the speed of that specific mushroom when I move really really fast either in tunnels or like we see in in this case we have a mushroom piece that moves really really fast between regions then the factor would be very very low I would basically try to aim for zero point five seconds of full smoothie but in case I'm just trimming and just exploring it could easily go to two-and-a-half three seconds smoothing and that works very well for me and when

we have a lot of different mushroom pieces enough different velocities and different regions this is how it all comes together so that's much from eleven and that's my talk thank you very much