0:00

Blasphemous and Blasphemous 2 are

0:02

Metroidvania action-adventure games developed by the

0:04

Spanish studio The Game Kitchen. The

0:08

games have a stunning, distinctive pixel

0:10

art style and atmospheric world which

0:12

is inspired by Spanish folklore and

0:14

religious themes. They are known

0:16

for their challenging combat and intricate level design.

0:19

David Arroza is the lead producer

0:21

and Danny Marquez is a senior

0:24

programmer on Blasphemous David

0:26

and Danny join the show today

0:28

to talk about designing the game

0:30

systems, the game's development framework, engineering

0:33

character movement, console optimizations, and more.

0:35

Joe Nash is a developer,

0:37

educator, and award-winning community builder

0:40

who has worked at companies

0:42

including GitHub, Twilio, Unity, and

0:44

PayPal. Joe got his start

0:46

in software development by creating mods and running

0:48

servers for Gary's mod. And

0:50

game development remains his favorite way to experience

0:52

and explore new technologies and concepts. Welcome

1:07

to Software Engineering Daily. I'm your host for

1:09

today's episode, Joe Nash, and today I'm joined

1:11

by David Arroza and Daniel Marquez of

1:13

The Game Kitchen, creators of Blasphemous and its

1:16

recent sequel, Blasphemous 2. David

1:18

and Daniel, welcome to the show. Thank you for joining

1:20

me today. Hi. Thank

1:22

you. David. I

1:26

am the producer for Blasphemous 2.

1:29

With us is Danny. Danny.

1:32

Yeah. I'm Daniel. Danny.

1:35

I'm one of the programmers of the game. Awesome. Thank

1:37

you for putting out your job titles as well. Very helpful and

1:39

very excited to talk to you today about Blasphemous 2, which I've

1:42

been playing quite a lot of in the lead up to this.

1:44

So I guess to kick us off, you know, you mentioned

1:47

your roles there. We've got a producer and a programmer. Can

1:49

you talk a little bit about what it is you do

1:51

at The Game Kitchen and also on Blasphemous? Well,

1:54

I always joke by describing my

1:56

job as I Try

1:59

to prevent people. The ball from.

2:01

You. Know. Tearing. Each.

2:03

Other about. The

2:06

police or Raleigh's trying to

2:08

get in sync every of

2:10

the different aspects that are

2:12

involved in game development so

2:14

you have to know a

2:16

little bit about programming. A

2:18

little bit about our tiny

2:21

may show audio to the

2:23

very non off how all

2:25

these different areas are related.

2:27

during the development of a game. And

2:30

the dependencies that are. Involved.

2:33

And to get to agreements

2:36

with our publisher in order

2:38

to keep everyone happy. About

2:40

the progress of the game. So.

2:43

Let's. Say that it's my

2:45

fault. better angle differ. Everything

2:48

that is sense really working.

2:50

That's mine. Responsibility.

2:53

Survey. Relevancy with Les mis your the

2:55

tennis and once had a whole seat

2:58

while I was on the premises for

3:00

a blessing was one that was Modi

3:02

see Oh please the court and Ceo

3:04

and director of to to the of

3:06

I did the troll for blessings to.

3:09

Awesome! And. Danny urls the

3:11

program yeah well as a programmer

3:14

race a wide variety else. Stuff

3:16

but mainly in this game.

3:19

I was responsible for Sistema.

3:22

Let's say architecture. Yeah we

3:24

know we also rendering the

3:26

stuff too, but mainly systems

3:29

and architects of. Also

3:31

has launched into that sucks you haven't

3:33

played for the games, can introduce bus

3:35

miss to toss and I guess the

3:37

wider world of blasphemous as I would

3:39

be so. Little. Blossom is.

3:42

the second game is so

3:44

much we've been yeah game

3:46

the first one was more

3:48

on actual performance some people

3:50

call it ah souls like

3:52

game that wasn't really the

3:54

original intention but you know

3:56

people gave it's that description

3:58

but blossoms do pure

4:01

metroidvania game that

4:03

is based on a

4:05

lot of this Spanish culture.

4:08

It's a dark

4:10

fantasy inspired by Spanish

4:13

painting, architecture, or

4:16

religion. In spare, we

4:18

try to avoid direct, you know,

4:21

direct references to any kind

4:23

of religion, but the inspiration

4:25

is in there. Awesome. Yeah, I

4:27

have to confess, I didn't know about a

4:29

lot of the associated iconography before I played

4:31

the game. And when you know, while I

4:33

was like looking at other interviews about the

4:35

game, and Levi saw you comment on Spanish

4:37

Holy Week and another one, and I looked

4:39

at the images in Spanish Holy Week, I

4:41

was just like, Oh, okay, I get it

4:44

now. I understand. I understand the aesthetic vibe.

4:46

I'd highly friendly listeners haven't seen it. Definitely

4:48

look it up. Definitely.

4:50

So I guess on to the implementation of the game. What

4:53

is Thomas to built in what we talking in terms

4:55

of engine and technologies? Well,

4:58

it's built upon Unity engine.

5:01

We are using pretty okay,

5:03

some plugins here and there, but

5:05

almost everything is with the bias

5:08

from scratch in unity. Yeah,

5:11

that's we told

5:13

from zero. Okay, we didn't

5:15

reuse anything from the first part. I

5:18

had to say I was in on the first

5:20

last month. So we

5:22

build this from the ground up upon

5:24

unity. Awesome. Yeah, I mean, that was

5:26

one of the areas I was really interested in that as

5:28

you approach a sequel, you know,

5:31

especially when there's a lot of shared aesthetic,

5:33

it's still you know, they're to decide to call the

5:35

platform and etc. How you think about code reuse? How

5:37

you think about, hey, are we building on the previous

5:39

game? Is it starting scratch? What was that process like

5:41

with going from last month's one to two? What

5:44

I have to say, like, just because I

5:47

wasn't there on the first one, I

5:50

can answer really with much,

5:52

you know, I

5:55

don't know the truth or the correct answer,

5:57

but this one from we

5:59

are making I'm sure

6:01

this one is reusable. Okay.

6:04

So we didn't make the same mistake again.

6:07

I know the reason because I

6:09

was there during the development

6:11

of the three DLCs. When

6:14

we made the decision to

6:16

start the sequel, one

6:18

of the discussions was, of course, what do

6:20

we do? Do we reuse

6:22

the old framework that we

6:24

built for the first game? Or

6:28

should we start from scratch?

6:30

And this is a really

6:32

tricky decision because, of

6:34

course, on one hand, you have

6:37

the whole game already built. Most

6:40

people will think that this is just a

6:42

matter of, let's say, switching assets. But

6:45

the harsh truth is that we

6:47

wanted to explore many design

6:49

paths that would

6:52

be maybe not impossible but

6:55

really difficult to implement

6:57

using the old framework. For

7:00

example, we only had one weapon in

7:02

Blood famous one, and we have three

7:04

of those in the sequel. And that's

7:06

something that I find I

7:09

can even start imagining how would

7:11

that be implemented. Also,

7:14

there were many really

7:16

important things for the

7:18

gameplay feeling, mainly the

7:21

controller of the character, that

7:23

were not great in Blood famous

7:25

one. When the game was

7:27

released, we are aware that it could

7:30

be really improved and the controller was

7:32

improved for the DLCs. But

7:35

we knew we could do

7:37

that better. But the

7:39

only way to do it was to start

7:41

from scratch. This is

7:43

something that we are asked about quite

7:46

frequently. Why did you start

7:48

a new project from scratch instead

7:50

of reusing Blood famous code? It's

7:53

something that we can't really recommend

7:55

to everyone. It's a

7:58

risky decision. And

8:01

we can guarantee that

8:03

everyone would success in

8:06

making this. So it's not something that we

8:08

would recommend. We took the

8:10

risk and it turned out okay,

8:14

but it could not

8:16

have been this way. Yeah, it's very interesting. I

8:18

think a lot of folks would

8:21

be surprised to hear that like adding extra weapons is such

8:23

a huge constraint. But I guess if you've got a design

8:25

of a game that's, you know, not assuming there's any

8:27

reason to switch weapons from the start, it's ready by

8:29

10. As

8:36

a listener of Software Engineering Daily, you

8:39

understand the impact of generative AI. On

8:42

the podcast, we've covered many exciting aspects

8:44

of gen AI technologies, as well as

8:46

the new vulnerabilities and risks they bring.

8:49

HackerOne's AI Red teaming addresses the

8:51

novel challenges of AI safety and

8:53

security for businesses launching new AI

8:56

deployments. Their approach involves stress

8:58

testing AI models and deployments to make

9:00

sure they can't be tricked into providing

9:03

information beyond their intended use, and that

9:05

security flaws can't be exploited to access

9:07

confidential data or systems. Within

9:09

the HackerOne community, over 750

9:12

active hackers specialize in prompt hacking

9:14

and other AI security and safety

9:16

testing. In a single recent engagement,

9:18

a team of 18 HackerOne hackers

9:21

quickly identified 26 valid

9:23

findings within the initial 24 hours, and

9:26

accumulated over 100 valid findings

9:29

in the two-week engagement. HackerOne

9:31

offers strategic flexibility, rapid deployment,

9:34

and a hybrid talent strategy.

9:36

Learn more at hackerone.com/AI. That's

9:39

hackerone.com/AI.

9:50

I want to follow up on something that

9:52

Danny said, and to your last point there

9:54

that you wouldn't recommend starting Code from Scratch.

9:57

Danny, you mentioned that Blasphemous 2 has been

9:59

built to... be reusable into the future. Can

10:01

you talk a little bit about some of the ways that

10:03

that's happened? Okay, let's say

10:05

we build the features around,

10:09

how can I say, let's use

10:11

the term that Unity make

10:13

it trendy, like the data

10:15

oriented dots technology, something like

10:17

that, which you

10:20

said in our framework, it's

10:22

okay, don't take me like

10:25

expert, but it's like an array of

10:28

stuff. Okay, and we built

10:30

our engine around

10:32

modules. Okay, let's say the managers of

10:34

system, but just manage

10:36

stuff. Okay, and you let's

10:39

say you just have to add or remove

10:41

modules in order to

10:43

implement a feature this way, for example, let's

10:45

say some basic example.

10:49

And say you want platforms that

10:51

when the player stand on it on

10:53

them, they dissolve,

10:55

they collapse, we build this

10:58

module, okay, that's the component

11:00

of that

11:02

when the player stand still, that

11:04

will disappear. Okay, so we want

11:06

to remove it, just we

11:09

don't our, let's say config table,

11:11

we remove that module. And

11:14

nothing is executed. So

11:16

that's our first stepping

11:19

stone. Let's say like, this

11:21

way, with that in

11:24

mind, build all

11:26

the features, okay, so we can

11:28

build features that in the

11:30

end, they are easy to remove. Okay,

11:34

maybe we made a little harder to implement.

11:36

But then it's just a

11:38

tool, okay, I don't want this feature

11:40

anymore. So remove it from the core

11:42

of the game. So

11:44

that's one of the basis

11:47

on how we build the

11:50

framework. Okay, awesome. Yeah, we

11:52

could say that plus one was

11:54

like monolithic kernel show, plus two

11:56

is more modular. Okay, yeah, I'm

11:58

more like I know. that's

12:01

oriented like, okay, you got

12:03

a system, you got components, and the

12:06

system will be internally implemented

12:10

how hard that you want, but it's

12:12

like an array of components. So then

12:14

you have the game is nothing more,

12:16

our logic engine, our logic

12:19

game is nothing more than a for loop with

12:22

a deterministic order of modules,

12:24

okay, that platforms that

12:27

disappear, okay, player movements, input,

12:30

reading input, et cetera, et cetera, awesome.

12:40

Let's face it, no one likes writing

12:42

or maintaining documentation, but when you start

12:44

a technical project or pick up a

12:46

new task, missing information can cost you

12:48

valuable time. Gitbook is a technical knowledge

12:50

platform that fills that information gap, making

12:52

it easy for your team to capture,

12:54

maintain, and find information from a single

12:56

source of truth. Here's a couple of

12:58

examples. With Git Sync, you can set

13:00

up a two-way sync between your repo

13:02

and Gitbook, so you can turn markdown

13:04

files into awesome user-friendly docs. And if

13:06

you make a change in your code

13:08

base or a technical writer updates the

13:10

docs in Gitbook, the edits sync between the

13:12

two automatically. Or what about when you need

13:14

to find something in your knowledge base? Forget

13:16

about searching, just ask Gitbook AI, you'll get

13:19

a neat summarized answer sourced directly from your

13:21

docs. Those are just a few examples of

13:23

what Gitbook can do, but it has plenty

13:25

more tricks up its sleeve. So why not

13:27

give it a try? Head to gitbook.com to

13:30

find out more. So

13:39

it's been a little while, I don't think I've touched

13:41

Unity properly since 2016, but

13:43

it sounds to me, the Unity.sync sounds to

13:45

me like an entity component system, is that

13:47

correct? This is everything from the VCS? Awesome,

13:50

perfect. I understand, awesome. That's great. So

13:53

with that in mind, I guess like when you're thinking

13:56

about we want this game to be reusable, and this might be

13:58

a question that I'm not allowed to ask, but is that... just

14:00

for future DLCs or are you talking like

14:02

other games entirely outside the blasphemous universe? Is

14:04

that or is it like you know we

14:06

just want to make sure we do another

14:08

blasphemous we're not writing from scratch without committing

14:11

to doing another blasphemous. That would be the

14:13

ideal situation. But

14:15

theoretically in other games too in

14:17

fact some modules are are

14:20

being reused in our prototypes. Awesome

14:23

very cool. So during that David

14:25

you mentioned the movement which was one of the

14:27

things that like when you look at you know

14:29

blasphemous players talking about blasphemous 2 certain people are

14:31

talking about the control and the movement being like

14:33

super improved. You said that was you know a

14:35

bit of a complicated process. Can you talk a

14:37

little bit about what the problems were with like

14:39

the control in blasphemous 1 and how you overcome

14:42

those for 2? I'm here

14:44

I about the same situation that

14:47

Danny I wasn't doing the development of

14:49

blasphemous 1. I was one of the backers

14:52

of the Kickstarter and I already knew that. That

14:54

is awesome that is a journey. Yeah

14:56

no we were friends we knew each

14:59

other I already knew Danny as well

15:01

on that time but I've heard about

15:04

that and I had to deal with

15:06

the code during the DLCs. So

15:09

the reason is that blasphemous

15:11

was the first platformer

15:14

game that was made in the

15:16

game kitchen. The previous game the last

15:18

door is a point and

15:20

click adventure game which has nothing to do

15:23

with this kind of games. So

15:25

this code was written by

15:27

Paco who I hope doesn't

15:30

listen to this but

15:32

he's the one that tells this story.

15:35

So it was made you

15:38

know like they could it's not something

15:40

that they had experience with it was

15:42

the first game that Paco made so

15:44

that was the best effort that

15:47

could be put into the

15:49

game with the resources that

15:51

they had which were not

15:53

many you know. I'm money

15:57

even having a Kickstarter that was

15:59

successful. It's not as

16:02

much money as people think. Money

16:04

flies when you talk about

16:06

game development. It's

16:09

something that we really

16:11

wanted to improve in the sense that,

16:13

hey, now we know what we want

16:16

to do with the controller. And

16:18

I think that we could, Danny,

16:21

correct me with this, but I think

16:23

that we got even a little bit

16:25

obsessed with the controller or the character.

16:28

I can do it a little trick

16:31

if I hope I try my best

16:33

again with the language where the character

16:35

controller can, let's say, something of

16:37

its basic implementation is

16:39

the character with the controller, obviously.

16:42

The character is like

16:44

the physical appearance, the

16:46

locomotion and the animation support, that's

16:48

the character. Then you've got the

16:51

controller, that is the

16:53

inputs, and the logic

16:55

you do with that input and with

16:58

the queries from the locomotion and from

17:00

the chart. So we

17:03

wanted to improve

17:05

this a little. As I said,

17:08

let's go

17:10

in a little deeper and say, okay, character

17:13

controller, not the character,

17:15

not the character, just this,

17:17

another module. Let's say locomotion,

17:20

say animation, inputs, say

17:23

the logic you do with that thing. And

17:26

then that logic we

17:28

divided with something we

17:30

call abilities and

17:32

these abilities are nothing more than

17:34

states that can

17:37

be executed concurrently or

17:39

exclusively. So we got

17:41

what I call, I

17:44

think I didn't invent

17:46

it, like hierarchy, finite,

17:48

concurrent state machine. Okay.

17:52

The controller itself, what

17:54

it does is manage

17:58

these abilities. Like,

18:01

for example, jump,

18:03

when you jump, you cannot

18:05

crouch. When

18:08

the ability of jumping

18:10

is executed, it blocks

18:13

the ability to jump. But,

18:16

for example, you can

18:18

attack, so that can be

18:20

executed concurrently along with

18:22

the jump. This

18:24

way, you've got kind

18:26

of flexibility. You can

18:29

scale and match the complexity of the

18:31

controls, because then

18:34

inside each ability, on

18:36

each stage, you can implement it as

18:39

a stable chain to say so. That's why

18:42

it's hierarchical. But inside that

18:44

ability, you can do more or less

18:47

whatever you want in order to do

18:50

the behavior or make the character

18:52

do what you want. So

18:55

we've got the inputs,

18:57

the locomotion, the animation part,

19:00

and the abilities. And just

19:03

our control, just manage,

19:06

then you

19:08

can execute right

19:11

now, but you can interrupt this

19:13

one with this ability, and

19:15

things like that. So

19:18

with that approach, I think we

19:20

made something that

19:23

works pretty well. Yeah. Much

19:26

better. In Blasphemy 1, for

19:28

example, I remember the ladder climbing

19:31

ability. For trying to find

19:33

a book on the ladder

19:35

climbing, you had to check on

19:37

three or four different places, including

19:39

the animator state matching, which

19:43

was a nightmare, because I

19:45

think that Paco did, but

19:48

it was almost impossible for

19:50

anyone that was in Paco to try

19:52

to deal with the controller issues. Now,

19:56

with the architecture that Danny

19:58

and the rest of the team made. It's

20:01

way more easy. I

20:04

was going to say to fix issues, but

20:07

there are no issues. WorkOS

20:17

is a modern identity platform built for

20:20

B2B SaaS. It provides

20:22

seamless APIs for authentication, user identity,

20:24

and complex enterprise features like SSO

20:26

and SCIM provisioning. It's a drop-in

20:28

replacement for Auth0 and supports up

20:31

to 1 million monthly active users

20:33

for free. It's perfect for B2B

20:35

SaaS companies frustrated with high costs,

20:37

opaque pricing, and lack of enterprise

20:40

capabilities supported by legacy Auth vendors.

20:42

The APIs are flexible and easy

20:44

to use, designed to provide an

20:46

effortless experience from your first user

20:48

all the way to your largest enterprise

20:51

customer. Today, hundreds of high-growth scaleups are

20:53

already powered by WorkOS, including ones you

20:55

probably know, like Vercel, Webflow, and Loom.

20:58

Check out workos.com/SED to

21:00

learn more. Thank

21:58

you. that

22:01

game because it's really

22:03

cool how the interruptions of

22:06

each attack are done. And,

22:09

excuse me, I wanted that

22:11

kind of flexibility that, okay,

22:14

you can interrupt this ability. For example,

22:16

the dash, you can, from this frame

22:18

to this frame, you can interrupt that

22:20

with the jump. Or you

22:22

can interrupt this with this attack. So

22:25

you can do, I don't

22:27

know the, that I try

22:29

to do, you know, the, like

22:32

the, the lunch, lunch, lunch, and things

22:34

like that. We

22:37

made that, okay, we built a

22:40

bunch of tools around the frames,

22:43

like the heat frames or the

22:45

cancellation frames. That

22:48

was part of the main

22:51

objective. That's awesome. Yeah, I started

22:53

with the big mace, Verdicto. And

22:56

then the first weapon I unlocked was

22:58

the pair of rapiers. And it really suddenly

23:00

felt like I went from like, oh, you

23:02

know, I'm just going around smashing stuff. And

23:04

that's a relatively simple to like, oh, suddenly

23:07

I need to be caring about interrupts and,

23:09

you know, doing fancy tricks with when I

23:11

pull off different attacks and dashes and lunges.

23:13

It was a really like striking, striking gameplay

23:15

transition, which actually, I guess, do you want

23:18

to introduce the three weapons and how they

23:20

influence the gameplay, whichever one works? Something

23:23

that we knew for sure, we had

23:25

a lot of, you know,

23:27

doubts about what we want

23:29

to do with the sequel, but something

23:32

that we knew for sure was about

23:34

having multiplayer weapons. We

23:36

weren't sure how many of

23:38

those, but that is

23:40

something that we definitely wanted to

23:42

explore. And also

23:44

the community have asked about that

23:47

a few times. Hey, it would

23:49

be cool if we could use different weapons. And

23:52

it was really hard trying to do

23:54

something that could compare to

23:57

the weapon of the first game because

23:59

that's where. mea culpa was

24:01

kind of iconic to the game. So

24:04

we had to try to

24:07

find something that matches that

24:09

sense of wonder of the

24:12

first weapon. And that's

24:14

where the answer we,

24:16

you know, code names for the weapons

24:18

and the elements of the game are

24:21

stuck in our head. And yeah, I

24:23

have to think a lot about the

24:25

real, you know, the public names

24:27

of the weapons. Is that

24:29

the same thing? The single source

24:32

that's the pair

24:34

of rapiers. We

24:37

call rosary blade is the

24:39

internal name for the rosary

24:44

blade. Yeah. Awesome. We

24:46

have a lot of game development is full

24:48

of silly

24:52

names. Yeah. I don't

24:54

see this sometimes. So

24:57

that weapon is kind of

25:00

like that causing of the mea culpa

25:03

because they really

25:05

share the move set. So

25:08

there we had the, let's

25:10

say, mea culpa. But

25:13

then we wanted to have

25:15

more variety of gameplay.

25:17

So that means we need

25:20

something really powerful, but

25:22

slow. So you have to

25:24

really commit when you decide to attack

25:26

with it. That's the, ah,

25:29

the sensor. That's the

25:31

rosary blade. Okay. So the one big sword, the rosary blade,

25:46

is the replacement for mea

25:48

culpa. Okay. Yeah. Yeah. It's

25:51

really embarrassing when you come

25:54

to these interviews and you don't know

25:56

the real name of the stuff in

25:58

your game. or as

26:00

a big fan of name driven development, I'm

26:02

really enjoying hearing the internal code names and

26:04

picturing how they came about. But

26:06

sometimes people ask us about something and

26:08

we are like, I don't know what

26:10

you're talking about. So yeah,

26:14

I forgot the name again.

26:16

The sensor is really heavy

26:18

and it punches with a

26:21

lot of power and it's really powerful

26:23

and it has this spinning

26:25

that acts that feel

26:28

really nicely, but it's

26:30

kind of slow. So you have to be careful

26:32

when you use it because once you

26:34

commit to attacking with it, you

26:37

are, we say that you are

26:39

sold to the enemy. You are

26:41

at his mercy. On

26:43

the other side, we have the pair of

26:46

repairs, which are which is

26:50

blazing fast, but deals

26:53

a little bit less damage.

26:56

And that's, you know, the three

26:58

approaches that we wanted to try

27:01

with the three weapons. We

27:03

find the three is a proper amount of

27:05

weapons to have in this kind of game.

27:08

We could talk during

27:10

hours about the weapon

27:13

switching feature. Yeah,

27:16

that was the first moment when I realized

27:18

how quick it wants to change weapon and

27:20

how you were expected to change in midair

27:22

to continue moving. For example, that was the

27:24

moment where I was like, oh, the skill

27:26

ceiling on this game is so much higher

27:28

than I had appreciated so far. Like it

27:30

suddenly feels very, yeah, like you have to

27:32

understand a lot. The mechanics for

27:34

changing the weapon, I think, correct

27:36

me, Danny, was it four

27:39

or five iterations of

27:41

the weapon exchange? I

27:44

don't remember, but I have to say that the difficult

27:47

part wasn't the technical,

27:50

it's like the feeling was

27:53

more like, okay, that doesn't feel

27:56

great. Let's try another thing. And

27:58

then another thing. and let's try

28:01

stopping the game for, you know,

28:04

half a second or let's slow

28:06

down time. Let's pop up this

28:10

wheel, this selection. I

28:13

guess just to give an example for listeners who

28:15

haven't played, so like a case where this might

28:18

come in, for example, the pair of rapiers, you

28:20

can use those to ride these like currents of

28:22

light. And then at the end of that, there

28:24

might be a bell you have to smash with

28:26

the sensor. So you have to, while you're mid-air,

28:29

change from the rapiers to the mace really quickly

28:31

after having done this jump. So I can see

28:33

why you'd be experimenting with, you know, delays or

28:35

wheels, et cetera. That's really interesting. Yeah, but

28:38

that introduce, you know, for

28:40

example, slowing down time introduce the problem

28:42

of, hey, people can

28:44

use these two-bit bosses way

28:47

easier that we intended because they

28:49

can slow down time and have

28:52

time to dodge projectiles or

28:54

to think about what to

28:56

do. So this

28:59

was maybe one

29:01

of the most challenging aspects

29:04

of the game, one of

29:07

the most not. We'll

29:09

come back to what the most was, but I

29:11

guess one, while we're on the weapons, one thing

29:14

that I think is really interesting, you know, you

29:16

described the games of Metroidvania. So, you know, this

29:18

idea of gaining equipment and abilities to unlock new

29:20

areas, but you do that through the weapons as

29:22

well. They're not just like, hey, I've got weapon

29:24

in my inventory. I have a new style of

29:26

tack. They're also your navigation ability, your obstacle clearing

29:29

ability. And then you can choose any of them

29:31

to start with. So like that just seems like

29:33

a whole heap of design problems, like the layout

29:35

of the world, which areas can people do

29:37

first? What order do they get the weapons in?

29:39

How did that decision to give the players that

29:42

much flexibility came up, come about? And how on

29:44

earth did you start dealing with that? That's

29:46

one of those situations in which

29:50

maybe you make the decision before knowing

29:53

the consequences. Truth

29:55

is that we knew what we

29:58

were assuming when we decided to

30:00

play. give players that selection

30:02

of weapons. It

30:05

was quite a challenge for the whole

30:07

team and design because

30:10

they had to take into account that

30:12

players could go three

30:14

or four different bosses when they

30:16

started playing the game. So you have to be

30:18

able to traverse the world with any

30:21

of the weapons. And you have

30:23

to be able to beat the

30:25

three first bosses with any of

30:27

the weapons. It was an interesting

30:29

challenge, but I think

30:31

that it was worth it. Yeah,

30:34

I would definitely say so. Yeah, it's one of those things where

30:36

as you're playing a game and you realize kind of what is

30:39

happening and how things

30:41

would play out if you restarted the game, it was

30:43

one of those elements where I was like, oh, this

30:46

is very interesting. So you mentioned that the weapons weren't

30:48

the hardest technical challenge you faced, which I have

30:50

to ask, what was the hardest challenge? Don't

30:53

get me wrong, but we are doing a

30:55

little indie game. We are not doing a

30:57

higher AA game. So just just take it

30:59

with a grain of salt. Maybe

31:02

the one of the things I found difficult

31:07

was we weren't in

31:09

charge of the port, but Switch

31:13

has some memory

31:15

limitations. Our

31:18

artists tend to use

31:21

really large textures.

31:23

So that was a

31:25

little challenge. And

31:28

also remember

31:31

that 5.6 weren't as smooth as we wanted, but

31:38

that was not in

31:40

Switch, that a little, but

31:43

it was just with simple optimizations and level

31:45

of detail with 5.6. That

31:47

was more than, okay, but in

31:51

PlayStation 4, that

31:54

was really interesting. Truth be

31:56

told, I didn't know how

31:58

much optimization to do.

32:02

Luckily, one of the guys

32:05

from the 14 team got an answer

32:07

from Unity, which I

32:10

don't know how it works

32:14

because it was like, use

32:16

that instead, move the 5.6 thread

32:20

on a different thread. Because

32:22

they were facing with the issue of having

32:25

zones of the game which

32:27

were running at 60FPS

32:30

on the Switch that

32:32

dropped to 40 on the PS4. Wow,

32:35

that's not the order round I would have expected, yeah. That

32:39

was our reaction. We were, you

32:41

know, the Xbox One and PS4 version

32:43

were released a couple of

32:46

months after the release of the game and

32:48

we were pretty confident about

32:50

the performance of the game on

32:52

those platforms. But when the

32:55

producer of the console port told

32:57

me, hey, we are hitting a severe

33:00

raindrop on many zones

33:03

of the game on PS4, I thought that

33:06

he was joking, you know. Ah,

33:09

this is a console programmer

33:11

joke that he's making fun of

33:13

me. But no, he showed

33:15

me the benchmarks of the different zones.

33:17

It was true. There was a huge

33:19

drop in performance on PS4. And

33:22

when you don't know the

33:25

reason, it's something that when I

33:27

read it, I was like, I

33:30

have no idea. I don't know how

33:32

much or how, I

33:34

don't know what to do in order to help them. It

33:37

turned out to be, as you said, something

33:39

from the internals of

33:42

Unity, how it managed different threads or

33:44

something like that. I

33:46

can't do any less retest or

33:48

I can't do any

33:50

less use of the physics.

33:54

We implemented like, okay, it was very basic,

33:56

but it was a level of detail. Okay.

33:58

When the enemies are not on

34:01

screen. Like they likely

34:04

pop it just either or they use a

34:07

little, for example, is that five

34:10

or three rate as just one

34:12

and it's just

34:14

the bare minimum we can do anything

34:17

more. Because if we do less,

34:19

it's like the leaf NPC

34:21

or the enemy just ill

34:23

and the player will notice it. The

34:27

world freezes when you're off when you're not on that screen,

34:29

right? Yeah. Fascinating. Yeah. Console

34:31

porting has been widely discussed on

34:33

social media and stuff recently because

34:35

of Godot and Godot not having,

34:38

not coming with like facilities for

34:40

compiling the console like Unity does.

34:42

And it just seems so

34:44

many obscure complexities that you need to ask

34:47

the console developers or porting houses to help

34:49

with, right? That seems like

34:51

an impenetrable issue. Fascinating. So

34:54

earlier you mentioned the bosses and I wanted to

34:56

ask a little bit about that because in this

34:58

type of game, with the combat system and I

35:00

guess like the, as you said, players have kind

35:03

of given it a soul's like label on the

35:05

first one. Obviously players are very sensitive to boss

35:07

design and building a satisfying boss as a whole

35:09

thing. Like how did you approach the boss design

35:12

for this game? Did you have a particular ethos

35:14

on what you wanted them to feel like? This

35:17

is kind of hard to answer

35:19

first because it's not our direct

35:22

area, but I can

35:24

say that basically

35:26

the whole team is

35:28

always involved in nearly every aspect

35:31

of the game. Art

35:33

enemies, how the game feels. Bosses

35:36

is one of those places

35:38

in which everyone has something to say,

35:41

mainly from the creative

35:43

direction of the game where

35:45

we strive for the bosses to be related

35:48

to the song. They are appearing trying

35:50

to be meaningful for

35:53

the whole story, the

35:55

game for the lore. So we try

35:57

to approach those in a way. that

36:00

they add to the game. They are

36:03

not in there just for the

36:05

sake of being a boss in

36:07

the middle of a palace or in the

36:09

middle of a basement. They have

36:11

a reason to be there and they

36:13

contribute to the lore of the whole

36:15

game. So the design

36:18

has to be something that is

36:20

split between creative direction

36:22

and game design. And they work

36:25

together into defining what's

36:27

going to be the aspect of this boss,

36:29

what role does it play on

36:31

the story that we are telling. From that

36:33

point, they keep working together to

36:36

define how it should be. Should this

36:39

be a humanoid, a human sized

36:41

boss? Should it be huge? Does

36:44

it fly? Does it have multiple? I

36:47

think we don't have flying. Well, yes,

36:49

we have one flying boss, which

36:52

is, I don't know the real name,

36:54

but it's Susone. Maybe it's Susone,

36:57

the real name. I think

36:59

another one was the... I

37:02

was going to say the

37:04

girl in the elevators. I

37:07

don't remember the name. It's not a girl,

37:09

but a ghost. Ghost girl, okay.

37:12

Yeah, I guess it also depends on

37:14

your definition of flying. The embroidery, I would

37:16

just does a flying thing at points, right?

37:18

Yeah, that's a really popular

37:20

boss. And I like her a

37:23

lot. We have our

37:25

favorites as well in the studio. As

37:29

an interior feature, I remember

37:31

with the bosses and enemies,

37:33

for example, in the game, you can

37:35

change because when you build

37:37

a boss, maybe you can try some

37:41

behaviors in order to see if

37:43

the animations are the incorrect leader or something

37:46

like that. The game has a

37:48

hidden feature that you can change the

37:51

controls to that character

37:54

and play a little, test

37:56

it a little. Also, it wasn't

37:58

really useful. It

38:00

was there and I used maybe

38:02

twice, but it was there. Awesome.

38:07

Just to say, another

38:09

feature we had in

38:11

order to test the bosses. That is great to

38:13

hear. There are so many, you know, you spoke about

38:16

the iterations for things like the movement and the controls

38:18

and the weapons and things like,

38:21

you know, bosses and platforming

38:24

segments or another area where I imagine there

38:26

must be so much testing and

38:28

iteration for feel. As like, you know,

38:30

as you mentioned, as like an indie studio, as a

38:32

smaller studio, what does that testing look like? Like, I

38:34

think people tend to imagine focus groups and, you know,

38:37

huge open betas and that kind of thing. But what

38:39

does it look like for a studio of your size?

38:41

Well, it's

38:43

not until the game is mostly

38:46

done that we don't

38:48

start testing in the studio. I

38:50

think that the first play testing suggestion was a

38:53

year before the release. I think it was August

38:55

22. That

38:58

was the first time we had someone

39:00

from outside of the studio to come

39:02

and play the game. And

39:06

that was some point in which you

39:08

are supposed not to be

39:10

changing many things because

39:13

we had the content complete and

39:15

the gold version. The content complete

39:18

was in April, maybe

39:20

23. And what was the

39:22

total development time? It was 37 months.

39:27

Very producer answer. I

39:29

remember that people tend to say three years,

39:31

but it started in July.

39:34

So no, it was 37. So

39:37

it was in August and the

39:39

content complete was set

39:42

for February or April 23.

39:46

And the gold version for

39:48

I think it was May, 1st

39:51

of May, something like that. So

39:53

there wasn't really a lot

39:56

of time to make important

39:58

changes. So,

40:00

we did change things based

40:03

on that playtesting session. Well,

40:06

of course, a playtesting is not the

40:08

same as a QA session. The

40:10

publisher starts with

40:12

the QA of things

40:15

very early in the development

40:17

process, but we, you know,

40:19

that's technical testing.

40:21

We wanted to see the reaction from

40:24

players. Yeah. Making sure the

40:26

game doesn't break and that kind of thing, right?

40:28

Yeah. What we wanted to

40:30

test was the reaction of people

40:32

who hasn't been playing the game here

40:35

in front of their faces

40:37

for two years because

40:40

it's because you have been playing the game

40:42

forever. You lose perspective

40:44

of the game and

40:46

you fail to see the

40:48

mistakes or the things that could

40:51

be improved. So, there's this moment

40:53

in which you need fresh eyes

40:55

on the game. So, we brought

40:57

people to the studio. I

41:00

think it was like 12 people,

41:03

people that we knew. Most

41:05

of them we knew personally. But

41:08

when you make these kind of testing

41:10

sessions, you don't need a

41:12

fan who will be

41:14

happy with whatever they

41:16

play because they are excited to

41:19

try the sequel of the game so

41:21

everything will look great. Well,

41:23

we wanted to have people who

41:25

are really brutally

41:27

honest. I hate

41:30

her. Yeah. Excellent.

41:34

So, yeah, we contacted these

41:36

people who were, some of them were

41:38

friends of us, asked them

41:40

to be brutally honest with the game.

41:42

And from that session, we

41:44

got a lot of

41:47

good feedback that we implemented. As

41:50

long as the feedback that we got was related

41:52

to the weapon change system, I

41:55

remember we give them pen and paper

41:57

so they can write down their... thoughts

42:00

and their comments, then they

42:02

feel a form online with

42:04

their thoughts. And there

42:06

was something, one of the players left

42:09

the paper on the desk and

42:12

there was this comment. I'm trying

42:14

to translate it. Sorry for my friends in

42:16

here, but it said,

42:18

weapon changing, motherfucker.

42:21

Excellent. So

42:23

when we saw this, we were suspecting

42:25

that the system wasn't

42:27

really what people expected, but

42:31

that phrase, that comment, confirmed

42:34

it. That was in

42:36

August and then we have another

42:38

playtesting session in February. I

42:41

think it was February, barely a month

42:43

before the content complete, which

42:45

was really risky. It

42:49

was barely a month before

42:51

delivering the alpha of the

42:53

game. So there was not enough

42:55

time to make huge changes.

42:59

But being as we

43:01

are, we made big

43:03

changes in some systems

43:05

that we weren't really happy about

43:08

like some things that maybe

43:10

people take for granted like, when

43:12

did you save the game? But

43:15

if I have explored this

43:18

huge amount of the map and then I die,

43:21

do I have to go through the whole

43:23

map again or is this progress saved and

43:26

you keep every item that you already found? Because

43:29

this, which seems pretty obvious,

43:32

wasn't that obvious during development.

43:35

That's one of the things that we

43:37

changed with that playtesting session. I

43:40

think that we weren't saving the progress. After

43:42

dying, people got really angry

43:45

about that during the playtesting session

43:47

because you had to remember, ah,

43:49

if I went to this place

43:51

and I picked this item and

43:54

then I had, I don't know, money

43:56

in this place and this boss is

43:58

over here. We entered

44:00

in kind of a crisis

44:02

state in the team because

44:05

we needed to fix that, but

44:08

we weren't sure what

44:11

to do. At the end,

44:13

it's like with the weapon chains,

44:15

the solution, the final decision is

44:18

something that after it's made, it

44:21

looks pretty obvious, simple,

44:24

but it wasn't. One of

44:26

the versions of the When

44:28

Saving Game was the

44:31

classic Metroid or Castlevania

44:33

approach that if you

44:35

save on a broom, or

44:37

the play idea in our case,

44:40

and when you die, you appear there

44:42

like you were never there except the

44:45

guild. So it was

44:47

like this is a little weird. You

44:51

restore the state, but the guild is there. That

44:54

doesn't have much sense. Because the guild

44:56

is on a song of the map that you

44:58

haven't discovered yet because we were removing

45:00

the progress on the map. That

45:03

was one of the first versions. Yeah, I get it. Okay,

45:07

for listeners, so when you die in Lastmas

45:09

2, you gain guilt and that changes like

45:11

mechanics and different like what abilities you have

45:13

and how much mana, etc. Yeah.

45:16

Okay, that makes sense. Yeah. Yeah,

45:18

sorry. For a moment there, I was like,

45:20

I thought you were saying that the guild always appears somewhere you

45:23

haven't been and I was like, wait, did I just completely miss

45:25

that on my playthrough? No, I understand. No,

45:27

I mean, you roll back to that progress. Yeah,

45:30

okay. But it was like, okay, the idea

45:32

now sounds obvious, but a lot

45:34

of games implement like we did,

45:37

like we tried them and

45:39

it just works. But when

45:41

we tried, it didn't work for us. Yeah.

45:44

So it changed enough for the Metroid formula

45:46

that you've broken a standard element of it.

45:49

That sounds obvious, but it's not so obvious

45:51

because if for Metroid

45:53

works, why it doesn't work for

45:55

us? Yeah. And you've ended

45:58

up with, I guess, a checkpoint system that I... I

46:00

guess is the only souls-like, well,

46:02

one of the most souls-like remnants where you kind of

46:04

have this like, you know, you save at a checkpoint

46:06

and that's where you come back and then you have

46:08

to go get your pick up your guilt in a

46:11

very kind of, you know, yeah. Awesome way. Perfect.

46:14

Wonderful. We are getting close to time. So I wanted

46:16

to ask a couple of questions to wrap up and

46:18

I guess the first one, do you keep an eye

46:21

on the speed running scene for the last most two?

46:23

You mentioned earlier that players, you know, being conscious that

46:25

players will break things. Do you pay attention to what

46:27

they're up to? Honestly, I

46:29

keep an eye on the community

46:32

in general, mostly Reddit. Most

46:34

of the people from the team are not

46:37

really, don't really interact with the

46:39

community, but there are people who

46:41

are constantly interacting with social media,

46:43

but most of us don't. We

46:45

see what people comment about the

46:47

game and what they want us

46:49

to do with the game. This

46:52

morning, I sent the lead programmer

46:54

thread from Reddit in which someone

46:56

was talking about the progress of

46:59

the rapier and dagger and

47:01

hey, check this and see how

47:03

much reason this person has. But

47:05

for the speed running scene, I

47:08

haven't seen anything just for Blastmas

47:11

2. Okay. We

47:13

did for Blastmas 1 and I

47:15

remember that the first speed

47:18

run that we saw as a team, I

47:20

was cooking on the kitchen and

47:22

I had my iPad on the side

47:24

and I was checking the speed run and

47:26

I was like, how is he

47:29

doing that? Okay. What

47:31

was that jump? Because there was a

47:33

book on the jumping,

47:36

let's say system because Blastmas 1

47:38

didn't even have proper systems that

47:41

allow players to jump in an

47:43

infinite amount of distance.

47:46

But I haven't seen that for Blastmas

47:48

2. It's not that it

47:51

doesn't exist. I'm sorry. I

47:53

don't know. Now I'm curious.

47:55

When we finish this, I will go

47:57

and check. this

48:00

run, there was a really fast run that

48:02

got submitted for games done quick. But I

48:04

believe the reason it got rejected was because

48:06

you patched it. And I don't think you

48:08

patched it intentionally, but a patch you released

48:11

fixed the run. So there is

48:13

some there's some quick runs. We made only

48:15

one update to the game. And

48:17

honestly, I don't remember. It

48:19

was a sub 40 minute run. And then

48:21

the first the patch fixed it, which is

48:24

Yeah, I remember it. It was something like

48:26

pass the game when the Ah,

48:28

the room switching. When you do

48:30

the launch attack with Rosary blade,

48:33

you could use or something like

48:35

that. Or with prayer that with the time.

48:38

Yeah, I remember that. And I

48:41

did that myself when I saw

48:43

it, which sends you off words.

48:45

Yeah. Yeah, that's when

48:47

we passed. Yeah. Yeah. Amazing.

48:50

So I guess inevitable question. And it's totally okay

48:52

if you can't. But can you share anything about

48:55

what's next to Blasphemous 2 or you know, what

48:57

else the game kitchen is working on? As

48:59

a company, I can say that the game

49:01

kitchen is working on multiplayer projects. Some

49:04

of those are public like the

49:06

VR platform that has been built.

49:09

But sadly, or likely,

49:12

most of the games are not most

49:14

of the projects are not public yet.

49:17

As for Blasphemous 2, after

49:19

the release of the game and the

49:21

update, we took a

49:23

really long holidays.

49:27

Most of the team, those who could afford, I'm

49:30

sorry for that because he's one of

49:32

the people who couldn't and I didn't

49:35

took as many days as

49:37

I suppose to

49:40

have. Blasphemous 2 started really soon after Blasphemous

49:42

1, right? There wasn't much gap between

49:44

the two of them. That's something that we

49:46

didn't talk about. But Blasphemous 2 started the

49:49

development where we were working on the

49:51

three DLCs. Yes,

49:53

you were long overdue holiday. So Blasphemous 2

49:55

started in July 2020. Yeah. The first DLC.

50:00

was released, you can't

50:02

remember. I made a presentation not long

50:04

ago and I had to check all

50:06

the dates because I totally forgot. I

50:09

think it was February 21. I think it was more

50:13

than a year or maybe September 2020, October.

50:16

I'm sorry, I can't remember. I had to

50:19

check Wikipedia the last time. So

50:21

yeah, we overlapped the

50:24

two developments. Yeah, very overjoyed. You

50:27

took a lot of time after last semester and

50:29

still thinking about that. When we came back from

50:31

holidays, then we started talking with

50:34

Mauricio and with our publisher about

50:36

the future of the game. We

50:38

haven't behind anything yet. If

50:41

we had, I couldn't say so. Absolutely. Had to try,

50:43

had to try. Yeah, I

50:48

know. I know. For sure, there's always

50:50

things that teams

50:53

want to add to a

50:55

game. I don't know who said

50:57

this, but games are not finished,

50:59

but abandoned. So we

51:01

will not abandon it

51:04

yet. That

51:06

feels truer than ever in the

51:08

age of games being patched online. Long

51:10

gone are the areas where you shipped a cartridge

51:12

and didn't have to think about it ever again.

51:14

Perfect. Well, thank you both so much. This has

51:17

been delightful. It's always nice to learn more about

51:19

how the games we play are built. So thank

51:21

you both for your time today. Thank you, Joe.

51:23

Thank you for having us. Thank you for

51:25

having us.