Welcome back to On the Level - 
a series in which I play awesome

video game stages, alongside 
the designers who made them.

This time I’m playing one of my favourite games 
of 2021: the endlessly creative co-op adventure,

It Takes Two. This is a game about a bickering 
married couple - Cody and May - who get turned

into toys and then must work together to 
travel through madcap miniature worlds.

And that includes the level featured in 
today’s video: a tree in Cody and May’s garden.

Which is actually full of squirrel soldiers, 
killer wasps, and a giant robot bee. To fight

back, the couple have a pair of handy weapons: 
Cody has a gun that shoots big globs of orange

sap - and May has a crossbow that fires 
lit matches. And when the two collide -

*Explosion*.

To find out how this level was made, I talked 
to Oliver Granlund, a designer at Hazelight - who was

one of the designers responsible for 
this level’s layout and game mechanics.

Now, usually on this series I will play the level, 
and the designer watches my screen over Zoom.

But as this is a co-op game, I actually got to 
play with Oliver - I picked Cody and he chose May.

And so, without further 
ado, here’s our conversation

as we played through It Takes 
Two’s second chapter: The Tree.

CODY: Ah, this way!

MAY: What?! You're never gonna make that jump!

CODY: Yeah? Watch me!
Argh!

MARK: So I always like to start by asking: 
how did this level begin development?

OLIVER: Originally, art were just 
exploring settings. You know. We

know we're going to be shrunk, 
what kind of places could be fun?

And tree was one of them that was explored by 
art very early as a very interesting location.

And design right now is just doing 
prototyping, just going crazy with

tons of different mechanics - and we found 
the sap and match gun. For the very beginning

they were just loose prototypes but at some point 
they were paired together with the tree. They

went through each and every mechanic and, like, 
paired it to a setting that art had prototyped.

And then we essentially gave each, like, 
designer or a pair of designers a level

and then told them to, you know, make it, and 
come up with mechanics and everything in between.

MARK: So you and designer Robert Johansson were

responsible for the tree and also the 
snow globe, is that is that right?

OLIVER: Yes that is correct. So I thought 
it would be interesting to show this level

because a lot of people worked on this. This 
part was designed by a level designer named

Henrik Sandin. This one was actually made quite 
late. From your perspective it might seem like

of course this came first, but this actually came 
very late. And when the rest of the tree was made

there were no swinging for example, and we 
had to go back and add that in everywhere.

I think this part turned out really 
well, especially like a very calm

piece in between the storm, kind of. And also 
very much establishes everything around it.

There are some cool smart things that are actually 
done here with like ants moving in the direction

you're supposed to go. Like here for example. I 
think this was originally accidental: you know,

art added them in and then people started 
following the ants and then, you know, the ants

led you down like a hill and people jumped off and 
died. Henrik used that knowledge to his advantage.

MAY: We haven't got time for 
this! Is this a shortcut or not?

SQUIRREL: The wasps have invaded our tree, wiped 
out most of our tribe. You two must kill them.

MARK: So how did the sap and match gun develop

from a prototype to the final 
mechanic we see in the game?

OLIVER: In the original version of it, 
it was only you sapped and you exploded.

*Explosion*

Which was very cool, especially for combat, but it 
didn't really work if we wanted to do puzzles. So

that was one of the first things we 
did when taking them from a prototype

stage to the next level is that we 
looked into how can we make them

into more than just a fire and forget kind of 
thing. And that's where we started pushing,

for example, the weight mechanic 
and the spinner mechanic.

An interesting thing here is that this is one of 
the few places where we actually have a separate

tutorial for the players because, like, 
if you see a shooting sign (like on May’s

screen) both players are gonna shoot it. And if 
you see yellow… you know, the common ‘use this’

language that's been established by games: both 
just shot at both things and were super confused.

MARK: So you had to make sure each player, 
like, understood their separate roles?

OLIVER: Exactly. One thing we noticed, since 
we added weight as a property to, for example,

the sap gun is that we had to be very grounded 
for people to be able to understand this.

For example, if you weigh something down and 
something goes up they need to be connected.

If you look at a lot of other levels they'll 
have just straight up, you know, Super Mario

floating platforms. So what we did is for 
each level we set up our own rule set and

our own internal logic. It's basically like 
teaching a player a new game in an hour.

MARK: And what challenges 
did you face in doing that?

OLIVER: Well it's interesting because the sap 
gun is a very systemic weapon that has a lot of

freedom. You can shoot this anywhere. And it was 
a decision we made very early on where you can

you can go two routes: you 
can either go the the very

restricted route where you'll have a white 
spot in a black room and that's the only

place you can shoot. Or you can do the other 
way around, and that's the route we went with.

This could create problems where, you know, this 
cage… I have progress videos for you on. It's a

very interesting progress of it. So this is one 
of the earliest prototypes I ever made. It might

look super different, this big wasp nest, but the 
actual gameplay is very similar. But the problem

was that you could sap on the outside, where May 
couldn't reach. And here we have version two:

yeah this also hurts my soul. So Cody saps the 
roof instantly, so May has to look straight up…

oh no. And, you know, it's just not 
nice. A lot of the work we did with these

mechanics was eliminating places where you 
can do things that just doesn't feel nice.

And we still manage to keep the puzzle intact, 
with just tons of iteration and simplification.

Design by subtraction. And I think that was 
really helpful here. Like some things came out

kind of fully formed, but most things always 
needs, like, at least one or two passes.

MARK: Okay, so this is the first co-op game

on the show. How did you go 
about designing for two players?

OLIVER: I think we can break it down to three 
types of co-op. You have what we call parallel

co-op, and that is essentially just you're playing 
your own single player game with another person.

And then you have step-by-step co-op, 
which is essentially like you do something,

that allows me to do something, it goes back 
and forth kind of. And then we have the third

one which is like simultaneous co-op, which is 
when we're both having to act at the same time.

We wanted a variety of all of these 
and I think pacing those moments

in co-op is a big challenge to to make right. 
And that's something we care about a lot.

MARK: This lab section is great, 
what can you tell me about it?

OLIVER: Here is one of the super early 
blockouts. Oh this is also, as you see,

this is cringe-worthy to to watch. You know, 
it hurts. The general idea here… you know,

the side-scroller camera, which made 
everything way way better, it's not here yet.

MARK: And so when did that come in?

OLIVER: I think it was still relatively early 
that we started experimenting with that.

But I think a lot of these things, like 
that you see here, they are obvious but

it's more like I've built it, it doesn't work, 
okay… what's the next step to make this work?

And a lot of the iteration was like 
that, you know, initial idea, oh cool

they're gonna spin, and then not realising 
that there is gonna be camera problems,

for example. And then you know, oh sh*t it's gonna 
work great if we just add a side-scrolling camera.

CODY: Hey, let me get out of here!

OLIVER: Oh yes, have you played this one?

MARK: No, I didn't see this the 
first time I played, I’m scared now.

OLIVER: Yeah you should be.

BOTH: *laughter*.

MARK: How did these side interactions come about?

OLIVER: After each big milestone, we basically 
played through the game and we played through

with some play testers and then we looked at what 
problems there were. And one thing we reacted

to was that there was too little to do, and so 
we went on a spree of creating both mini games

and the side interactions. And there 
actually ended up quite a lot of them.

And how this one came about was, essentially, it 
started with a very simple ‘wouldn't it be fun

if you could trap your friend’ and then ridicule 
them. And then ridicule turned into torture and,

yeah, it went downhill from there.

This kind of became our collectible in 
a way. We had a discussion very early,

because we're a platformer, you know, 
of course we're gonna have collectibles.

Right? Question mark. And we tried that and we 
didn't find a good purpose for it. We wanted to go

another route, like we want to give meaning to the 
player or something they'd want. And what ended up

happening was we created so many minigames and 
then we ended up hiding them a bit because they

were sometimes a bit distracting, so that actually 
became our kind of collectible in the end.

MAY: Something’s coming!

CODY: Killer wasps!

MARK: This is our first combat 
encounter. What was that like to design?

OLIVER: One thing that's very interesting about 
co-op combat is you have to think about it in a

whole different way than you would normal combat. 
Especially if you're going for the simultaneous

co-op where you're both acting. So, for example, 
in this case we actually have an asymmetrical

relationship where you're the instigator and I'm 
the detonator. Which means that I actually don't

have a lot to do until you sap one of them. 
So what we've done in the AI is make sure that

more of them are aggressive to May, making it 
more interesting for me until she can detonate.

The other one is how do you handle 
if you're playing with a partner who

who's not as skilled as you, or the other 
way around? And I think that that was one

of the harder issues we had to answer. I think 
Tree is actually the hardest level in the game

in terms of introducing a new player to it. It's 
the only level with combat and aiming for combat.

But as you can tell when you play it… like, the 
auto aim system is very lenient and, like, we

try to make the gameplay less about ‘can I hit the 
skill shot’ and more about making sure you dodge

the wasp. Making sure that that you're spraying 
the right thing. And like all of that stuff

instead. It's still one of the hardest mechanics 
for people who don't know how to use both sticks,

but it ended up being way more usable than most 
first or third person shooters were at that stage.

MAY: Interesting mechanic looks 
like some type of tug of war.

OLIVER: Oh you're good at this! Oh no!

MARK: Yeah! Oh no that was 
easy. We need a rematch!

OLIVER: I'll take a rematch on that 
one. This has to be cut from the show.

MARK: Ah!

OLIVER: Oh. Are you playing with a keyboard?

MARK: Nope, just an Xbox controller.

OLIVER: : Okay. I'm ashamed.

MARK: So why did you decide to 
add in competitive minigames?

OLIVER: So the whole game is about corporation, 
and it's very hard to break the pace of that.

And I think that a big reasoning behind 
us doing that was wanting to create,

you know, a variation in moments, essentially, for 
the couples. Also kind of enforcing the story of

them bickering. We had them in some places where, 
you know, the characters are bickering - now the

players will be bickering. And it’s kind of 
fun and funny when you see that working out.

And we tried to create a variety of minigames as 
well that would allow different kinds of players

to win. We even have a literal chessboard in 
there. So the idea was that not everything

would have to be technically skilled, because 
then the same person would win over and over.

MARK: I love this giant vault 
door, it has a really cool design.

OLIVER: This one was actually the 
original, where you would spray

on it to explode it. But it had to be so 
tiny to not make your camera go wide. So

we ended up switching for this, basically 
turning. So, instead, we iterated on it and

made it into this lid actually. So the 
prototypes were basically just moved around.

MARK: Just how important 
is that iteration process?

OLIVER: I would say iteration was immensely 
important. We didn't have a lot of time

to actually, like, make everything perfect. 
We aren’t triple A, we can't spend however

much time we want on something. But we need 
to make it so that it's good enough for the

player. Iteration was an immense part of that, I'd 
say. Even though a lot of the ideas were similar,

I'd say the actual execution of them 
were quite different - wildly different,

from beginning to end. Especially when you 
have so many mechanics that you're teaching.

Like half of it was, you know, working on the 
mechanics themselves. Maybe this isn't clear,

maybe this isn't good. Can we communicate this 
better? Another half of that was just iterating on

the puzzles themselves, making them smooth, making 
sure it's fun and intuitive. A lot of the things

that are fun here, they aren't fun first pass. 
Taking something from idea to fun gameplay - it

takes a lot more work than than you might realise. 
Because the way we worked, it's very similar to

how Respawn structure their action blocks: make 
something in a week, and throw it in there. The

one deceiving thing, though, is that even though 
it might take like a week to prototype and get

playable, it might take way longer to polish. And 
that's something we realised, like, painfully so.

MAY: Don't worry, it's just 
a squirrel inside a robot.

CODY: Yeah! A very big robot. It’s trying to kill 
us. You know, maybe we should just turn back…

OLIVER: This is the cutscene that was added in 
later in the game when we realised… holy sh*t,

this level is growing. So we had like a storyboard 
for the tree and then eventually we looked it over

and we saw that it was supposed to be one hour… 
now if you play the tree you know that it's

probably more than an hour. It's bordering 
on two hours. This actually was a problem

because we had diluted the story with more 
gameplay because we we had so many prototypes.

One place where we actually filled in with 
story what was these places where, like, okay,

here we're gonna need to put a cut scene.

MARK: It must be really difficult to pace 
a story when the level keeps changing…

OLIVER: Yeah, totally. And I think 
it’s… especially for this type of game,

knowing how long can you go without having 
story, and also how long can you watch a

cutscene without any gameplay. It's a very 
fine balance to not bore the players, but also

not having them lose interest in their larger 
purpose in the story. It's interesting because

it's something you don't really see until the game 
is done. A big reason why a lot of this doesn't

land for a lot of games is because you can't 
see it as a whole until very late. You don't

know if a segment is gonna get extended or 
shrunk, and it's very hard to judge that

without the whole context. You're gonna have to 
do a lot of imagining to see the whole there.

MARK: And when you're balancing the story 
and gameplay, does one ever win out?

OLIVER: The previous game we made was 
A Way Out which was very story focused.

In this game I think that gameplay won 
out, but we try to keep them very balanced.

One of our mottos is to try to marry gameplay and 
story - keep them together. We want some mechanics

to have a multi-purpose kind of… you know, 
they’re filling story. For example, like, the

magnets are about attraction. Clockwork is about 
time. You know, we call them emotional levels.

MARK: So one of the things I love about this 
game is just the sheer variety of gameplay.

Where did that come from?

OLIVER: That was one of our core pillars 
all the way through. We don't want to

repeat things. And so we we questioned 
the very classical Nintendo's game design,

you know the four step level design 
rule. That's a great way to do it

and we do at some places, but we were more 
interested in cramming out as much variety

as we could. In another game the whole boat 
thing could have been like an entire game,

but here it's a five minute segment. We could 
have definitely extended and had some ideas for

how to make it 30 minutes long, but I think 
that would have ultimately made a worse game.

MARK: How much depth do you give a mechanic if 
it's only going to be there for a few minutes?

OLIVER: There is a lesson there 
that depth doesn't mean it's good

if you don't have time to use it. And I 
think that's something where we talk about

‘finding good enough’ which is one of our 
kind of pillars as well. Like ‘variety’

and ‘finding good enough’. And they go hand in 
hand. How deep does it have to be? It doesn't

have to be like a deep 20 hour mechanic if you're 
just gonna do it for five minutes. And I think

that's a very important thing: knowing when to 
stop. Because the players might not appreciate

anything beyond that point. What we did now is 
we try to take all the best things that we found

and put them in the level instead of, you know, 
finding all the things and dragging them out.

CODY: Whoa, I think they're 
bringing out the big guns now!

OLIVER: This thing took it took a while to get 
working: the swarm technology, the one with the

hammer, the whole system for this. And for it to 
go like seamlessly from one fight into this slide.

This was quite a lot of work but I 
think it was it was worth it in the end.

CODY: Hey hey! Don't do that!

OLIVER:

Sometimes as a designer you're kind of like 
I'm doing this thing, it's going to be very

very expensive, I really hope it's gonna be cool 
so I'm not wasting everyone's time, you know.

MARK: Yeah, as a designer like 
you're giving lots of work to

programmers and artists. So how do 
you make that relationship work?

OLIVER: I do believe it's 
very important to think of,

you know, the final look before you actually 
commit to anything. We give art, you know,

like an abstract piece of gameplay and it's 
like ‘what could this be?’ and, you know,

art sits there, and they're looking at it, and 
they're like… it's not fun for anyone.

MARK: Are there times when an artist made 
something that like really surprised you?

OLIVER: So, so many things. Like the the whole 
lab section, you know the big wheels, this one

was kind of like, I’d say, one of the prototypes 
that didn't really have a good context. You know.

Why do you have big spinning wheels? And they 
turned it into this very elaborate lab that ended

up influencing parts of the story. And I think 
that a lot of the times both art and programming

as well would just build things 
that surprised us and I think that,

you know, it's very true that game 
development is, you know, it's a team sport.

CODY: Aww… now what!

MAY: I think we need to head that way…

CODY: Come on May, we can't swim in this.

MAY: We're going by boat! Help me out!

OLIVER: I wanted to show you the iteration of 
the boat because you asked me a bit about how we

prototyped and I actually have the setup here. 
The initial idea was that you'd shoot and the

boat would propel in the other direction. This 
is a terrible idea: you don't see where you're

going. Absolutely terrible. And, you know, it's 
no worries - we made it in like a very short time.

So then we took the the second question: 
okay how can we make this in a way that

feels fun and easy to control? We landed on 
this one which was basically you controlled

it with a rudder. This was very fun to control 
but it didn't match the kind of rules we had set

for ourselves. As I said before, each kind of 
designer of a level set rules and Robert and I

had decided that our things were gonna be kinetic. 
We were gonna avoid as much UI as possible and we

would like to use the weapons. Instead of having a 
lever you press Y on, you would blow something up.

So that's when we came up with essentially 
taking the same kind of function but tying

it to the actual mechanic. So here you 
press the fire button to just start going.

I think we're great at being 
reactive: we make something,

play test it, talk about it, and then we change 
it. I think that that's one of our strengths.

It's definitely like one of the things that 
I think made this game great in the long run.

MAY: Wow! They are beautiful!

CODY: Yeah… deadly beauty. Jellyfish 
can be lethal, we better get moving.

MARK: So at this point the level 
starts getting really weird:

were there rules for how 
strange the game could get?

OLIVER: No, no. There definitely wasn’t. 
In meetings the feedback was ‘could we make

it a bit crazier’ from Josef… I think, 
like, think Avatar on LSD or something.

So we did. It's definitely out there: this 
is one of the moments where people playing,

you know, they’re like ‘are we still in the tree’?

CODY: You know, I am getting 
really tired of falling.

MAY: Well be grateful you're in good shape.

CODY: Oh yeah? Thanks May!

MAY: I meant good shape in this world.

CODY: Oh.

MARK: What was your approach to pacing the level?

OLIVER: So I think this is something we probably 
should have paid a bit more attention to because

we were just focused on creating cool sh*t and 
variety. A lot of games have to care about pacing

in another way than we do: what they pace is 
the repetition of content. Here you're gonna have

platforming, here you're gonna have a combat 
encounter, here you're gonna have platforming,

here you're gonna have combat again. And to 
not make that boring you have to make sure

that you're pacing it correctly with enough 
variety. we didn't have that problem as our

game was just full of variety by the baseline. 
We just created new mechanics all the time.

But we still have to make sure we pace 
intensity so that people don't get exhausted or

people don't get bored. You still have to make 
sure to pull the intensity curve up and down.

CODY: Oh, whoa…

MAY: What is this place?

CODY: Uh, I don't know but they look pretty angry.

MAY: Yeah they do.

OLIVER: This one is one of the most intense 
fights in the game actually.

MARK: What do you remember about designing it?

OLIVER: We actually wanted the beetle to be 
like a mini enemy. And at one point we said,

yeah, we don't need that variety - should we 
just make it into a boss? Which is not really

what we intended because there 
is a boss right after this.

But after the passes it became that 
way - I missed it, that was my fault…

MARK: No worries!

OLIVER: And I remember iterating on this so 
many times to make it more difficult for, like,

platformer and reducing the skill it needed from 
a mechanical or an aiming level. For example the

grates were both a switch up from what you'd done 
earlier but also reduce the amount of, you know,

how much aiming you need to do. In an earlier 
version we had it so that you had to like shoot it

on the butt, but it required like a lot of 
mechanical skill and communication on our

side to make that happen, you know. To go in 
behind, aim, you know, it was a lot of stuff…

CODY: Well, uh, you know the queen? She has all 
the nectar. So if you help us take her down,

you can have as much nectar as you want!

BEETLE: Oh yummy! What are we 
waiting for? Come on, jump on!

MAY: Jump!

QUEEN WASP: Intruders!

MARK: Okay so now we're facing the 
final boss of the level and I was

just curious about the design 
of the game's health system.

OLIVER: We actually had some different health 
systems we prototyped. We prototyped a Gears

of War type of health system where one player 
dies and you have to go and revive them. That

didn't work because of the sheer variety of our 
gameplay, you never know if you're dying during

a platforming section or if you're separated. 
So that didn't work. We tried a shared health

bar in the very middle of the screen and this idea 
seems great: you're sharing health, that's co-op,

but what happened is if you had one player that 
was really good and one that was really bad, the

bad player would just take damage over and over 
and then they would both die - instead of allowing

the good player to carry them. So what we ended 
up on is just like a timer that that goes down,

as long as you're both living. It becomes this 
kind of cheer on, like, ‘don't die, don't die,

I'm almost back!’ that ended up being actually 
quite effectful I'd say. We generally didn't want

players to have to replay parts either, that 
kind of goes against our motto of variation.

Our health system ties into that because if one 
of us dies that's fine and then we'll just spawn

in eventually, so it basically becomes like 
you have two lives. It's fun to die because

then you feel like I'm challenged and you 
get that stress, but it's not fun to redo it.

*Explosions*

OLIVER: It's funny because a lot of people 
reacted to the beetle dying here. Which,

for us, the beetle was kind of a throw-in, 
like, that was a one week prototype that we

threw in. They were like ‘oh no the 
beetle’ and we're like, ‘oh yeah,

the beetle, that's right’, and we didn't 
think of that at all. It's kind of funny.

MAY: Come on Cody, over here! You drive!

CODY: What, May, this is a plane 
- it's not a station wagon!

MAY: Just fly!

MARK: This plane section is really 
memorable, what can you tell me about it?

OLIVER: So this was made by Per Stenbeck. 
You had the storyboards already,

and we had the plane in there from like a 
story perspective, I think, before there

was a prototype. And I think that this could 
have easily just been like a flight section

that's that's kind of cool but it took some 
turns where it went even further than that.

And what happened was that they had this idea 
of the squirrels that are flying behind you

that they would like land on the plane and 
you'd have to like shoot them off-- oops.

MARK: Sorry!

OLIVER: They they were talking about that and 
they were like ‘oh what if you could jump out

of the turret and like kinda kick them off’ 
and then somebody said ‘like Street Fighter?’.

SQUIRREL: Time to say goodbye little doll. 
Can't wait to smash your wooden face!

MAY: I can't wait to kick your furry arse!

OLIVER: You know it's one of those things where 
in a lot of game companies those ‘what-ifs’, they

don't happen. But they do happen here and I think 
that's one amazing quality about Hazelight. And I

think kudos to everyone who worked on that because 
I think it turned out to be a great highlight.

ANNOUNCER: KO!

MAY: I’m here!

OLIVER: This part is like a homage to Brothers: 
A Tale of Two Sons where you have a section kinda

like this but with the brothers, but now in co-op. 
And, yeah, it worked excellently. And I think this

whole section is great because you get variation. 
You know we've gone through three stages of plane

but they all feel different. 
And I think that speaks to the,

you know, like integral to the way we 
worked was just variety and both from art,

programming, and design, everyone really 
got to add their own flair to everything

and have so much ownership that it 
just naturally became so different.

CODY: Oh! Oh!

MAY: That is the last time I'm flying 
by the seat of your pants.

CODY: Oh ha-ha, you're real 
comedian, you know that?

MAY: Ah, thank you.

MARK: And so there we have it.

It was fascinating to hear about how the sap and 
match gun evolved from a purely combat-focused

mechanic, to a puzzle-solving tool. And it was 
great to look at the step-by-step process of going

from prototype to polished game mechanic. Plus, 
this conversation really highlighted the very

collaborate nature of level design. And, as such, 
Oliver wanted to thank a handful of colleagues -

OLIVER: I would like to thank Robert, Alexander, 
and Tom, who helped with the video. and I would

like to thank Robert, Per, Henrik, and Filip 
who all worked on Tree and made it excellent.

As usual, a full level playthrough and 
conversation is available exclusively to

Patreon backers. And if you want 
to support Game Maker’s Toolkit,

please check out this quick YouTube ad before 
we get to the indie game recommendation.

My indie game recommendation this time is Umurangi 
Generation - a low-poly photography game with a

subversive sense of humour. In each level you're 
dropped into a 3D environment with a camera

and a list of things to snap. You'll need 
to find each target and figure out the right

lens to use and the right spot to stand on. 
If you agree with me that photography is an

excellent video game mechanic, play Umurangi 
Generation. It's out now on PC and Switch.