How To Make Indie Games In 2026 w/ Jonathan Blow — Full Time Game Dev Podcast Ep. 061

Okay, imagine creating one of the most beloved indie games of all time, Braid.

And it brings in millions of dollars of revenue. That's like the indie game

revenue. That's like the indie game dream. That is so cool. Okay, so you

dream. That is so cool. Okay, so you continue to make another game. You want

to make a sprawling puzzle game, and you make The Witness, which again brings in millions of dollars of gross revenue and then you move on to your third puzzle game. Okay, now sure, you want to

puzzle game. Okay, now sure, you want to make it maybe a little bit bigger than your previous games, but you want to keep it simple. And yeah, you want to create your own custom engine, of course. And well, yeah, let's go ahead

course. And well, yeah, let's go ahead and just create our own custom coding language. Why not? And then day after

language. Why not? And then day after day, year after year, the project just keeps getting bigger and bigger and bigger. And suddenly, it's 2025, and you

bigger. And suddenly, it's 2025, and you find yourself waking up one morning realizing it's been 10 years working on a single puzzle game. That's the story of my guest, Jonathan Blow, who recently

revealed this 10-year project at the Game Awards. It's called Order of the

Game Awards. It's called Order of the Sinking Star. And there are a few words

Sinking Star. And there are a few words that keep running through my head when I look at this game. Beautiful, intricate

complex, deep, massive, and maybe even a little insane. But don't take my word

little insane. But don't take my word for it. Let's talk with Jonathan Blow

for it. Let's talk with Jonathan Blow about this 10-year journey and how so much of the gamedev world has changed since he last dropped a commercial game in 2014. We talk about marketing indie

in 2014. We talk about marketing indie games in 2026. How to price your game in a very tumultuous season for the game industry. Which platforms are better for

industry. Which platforms are better for you as an indie? Steam versus console.

Why publishers matter in 2026. Yes, they

actually still matter and Jonathan Blow actually went with a publisher. How the

gorgeous art in Order of the Siky Star was made and why it was so expensive.

Why Jonathan created a custom coding language and a custom engine. And

finally, at the end of the podcast, we jump into his custom engine and he shows how Order of the Syncing Star is actually made. It's fascinating. Uh

actually made. It's fascinating. Uh

this conversation is really um, it's kind of wild. It's really interesting.

It's honest. It's raw. Yeah. Let's jump

into this conversation. And by the way if you want to make your own indie games and get paid to do it in 2026, I have a massive gamedev program and it's called Full-Time GameDev and it's 50% off for New Year's. This is a big one. And yes

New Year's. This is a big one. And yes

these are actual reviews from real students on the front page of my website., Many, of these, students, actually

website., Many, of these, students, actually went full-time with the help of the course. That is so cool. See, in the

course. That is so cool. See, in the program, you're not only going to learn how to make a game in 2D or 3D, and you can pick whichever engine you want to learn, Unity, Unreal, or GDAU, but you're also going to learn how I secured

funding from publishers and crowdfunding, or maybe even hit the Steamfront page multiple times. And I'm

going to show you as best as I can how I did it, and perhaps how you can do it too. Yes, I've gone fulltime on three

too. Yes, I've gone fulltime on three separate occasions with just a demo. But

don't take it from me. Click the link below and check out the reviews for the program. It's 50% off again for New

program. It's 50% off again for New Year's. And you're also going to get my

Year's. And you're also going to get my 2D and 3D art programs completely free but again, only for this sale. It

doesn't last forever. And there's just 100 seats available. Let's make 2026 the year that you master gamedev. And who

knows, maybe you go full-time. All

right, click the link below and let's jump into the conversation.

Do you think the marketing of Order of the Seeking Star is just a different playing field than it was for Witness or Braid?

>> Super, different., I, mean,, the, internet, is really different now. Um, I mean, The Witness came out 10 years ago, which is a really long time in internet time.

It's like infinitely long, right, in internet time. And so, there's a lot

internet time. And so, there's a lot more games now, first of all. There's a

lot more places for people's attention to go. Like social media obviously was

to go. Like social media obviously was around then, it's around now, but it's sort of blown up to take much more of people's attention now. And so games are kind of competing with things that are not games.

>> I, know.

>> And, that, makes, it, harder., I, think, I, saw some stat that like spending on games actually went down.

>> I, think, it, was, like, 2024,, some, year recently it actually went down. Like the

the overall spending >> also, prices,, I, don't, know, if, you, saw this article, but prices of games is going down as well. Well, I mean, they tried to raise them, right? They they

tried like 80 and $100 games.

>> And, it's, a, weird, thing, because, again, you know, when I was a kid on a personal computer, I think we would pay $40 back then or something.

>> Yeah., I, mean,, what, was, that, in, the, 80s or the '90s?

>> That, was, in, in, the, late, ' 80s, maybe., So,

that would be like over $100 now, right?

It's like $120 or something, right?

>> Y >> um, and, so, by, that, standpoint,, you, could justify $100 games, but like games are a lot more common now. They are easier to

make. They are better even though a lot

make. They are better even though a lot of games are are pretty bad. Um they are they are better now but as we've said there's just also a lot more things that you could spend your time and attention on.

>> Yeah.

>> And, so, games, are, competing, with, all that. And so I you know I was listening

that. And so I you know I was listening to some podcasts last night and they were talking about $80 games and I was like yeah even for me you know I mean I've been in the games

industry for a while. Um, I don't sweat decentsized purchases, but like $80 does sound expensive for a video game.

>> It, does,, doesn't, it?

>> You, know,, and, so, I, don't, know, how, the industry is going to approach this because if we just do race to the bottom on pricing >> that, won't, be, that, surprising., I, mean, maybe that's already happening. It won't

be that surprising if it all goes that way. But that's kind of what happened on

way. But that's kind of what happened on iOS, for example. And like I don't know any gamer who says iOS is their platform of choice for sophisticated games

right? And so we might be just driving

right? And so we might be just driving ourselves off a cliff. Um because you need I mean you know if you drive the price down super low two things happen.

One is if margins become slim which is just like the amount you could expect to make from a game is only barely more.

Then that lends itself to industrial approaches right? So like margins on

approaches right? So like margins on farming are really slim. You don't make that much money farming. And so we've got all these mega factory farms that make the money off economy of scale

right? And that's kind of what happens

right? And that's kind of what happens on iOS as well is just like, you know it's factory games in some sense.

>> And, so, if, you, want, to, see, creative independent games, that's probably doesn't help. Although there's a point

doesn't help. Although there's a point to be made for like then maybe the only people making games individually will be the people who super duper care about it.

>> Yeah.

>> Which, might, also, be, true.

>> But, it's, just, I, don't, know., It's, it's hard. Like I think I think if something

hard. Like I think I think if something if somebody makes something really decent and they put a hard work into it we should all be willing to pay that person for their if if I appreciate

their effort and want to play their game, I should be willing to pay them a decent price. But also, is that $100? I

decent price. But also, is that $100? I

don't I don't know that it needs to be $100 right?

>> Well,, what, creates, value, for, a, game?

Let's, say, Order, of the, Sinky, Star, is, 800 hours of gameplay. Maybe it is.

>> It's, not, that, many.

>> Okay.

>> Uh, it's, like, 500., Maybe

>> 500., Okay.

>> We're, not, That's, not, a, promise.

>> Different, amount, of, play, for, everybody.

Asterisk.

>> Yeah, >> it's, just, a, really, big, game, >> you, know,, if, it, But, let's, pretend, that it's 500 hours. who's going to play 500 hours, you know, you can't base the

value on the maximum hours, right?

>> Um,, you, can, base, the, the, value, on, a minimum like like or like a low number.

So, if it was like an hour, then yeah you could you could create some kind of valuation and say this is a $5 game. The

reason I say that is it's it's hard for me to figure out. I I've been going through pricing issues, too. Like, cuz

I'm about to finish Twisted Tower next year, early next year.

>> Yeah.

>> And, um, that, game, is, probably, five,, six hours. It's linear, you know.

hours. It's linear, you know.

>> Yeah., Yeah., Yeah.

>> Um, and, I'm, like, I, was, I, was, doing, play testing and all my play testers were like 30 bucks, 25 bucks. That's that's

the price.

>> But, I, know, I, think, you, have, a, publisher for Word of the Sigy Star.

>> Yeah., Um,, but, I, was, just, having, a conversation with my publisher and we're just like, we don't know if that's going to fly, you know, um, anymore. I think

it's going to have to be 15 bucks, you know.

>> It, depends., I, don't, know., I, haven't, had those equivalent conversations yet.

>> Yeah.

>> Um,, we, should, start, having, them, soon.

Uh, you know, given that we just announced the game's coming out next year. Um, but I'm honestly not sure. So

year. Um, but I'm honestly not sure. So

back when The Witness came out, we charged $40.4 $40 American dollars which at that time was outrageous enough for an indie game that it made some news and some controversy because it was

like, "Oh my god, how dare." Cuz there were games coming out at $30, right? And

stuff but there may have been some other games at 40, but it was pretty rare. But I was like, look, I don't know if people are going to buy this game. To the extent that people are, I figure there's some

core of puzzle fans who would really want something like this.

>> Yeah., And, then, maybe, we, need, to, make enough money to survive off that core group of fans, right? And so I priced it a little bit high. I mean, obviously I was like, look, this game, it's a really big game. It's really sophisticated. I

big game. It's really sophisticated. I

think it's easily worth $40. But of

course, everybody decides how much $40 is to them and and what's worth it for them. Um, but you know, it turned out in

them. Um, but you know, it turned out in retrospect that was not that big of a deal, right? uh people after playing the

deal, right? uh people after playing the game they felt oh this is actually a real game right it's not some indie like in the years since then we've had quite a few indie games what would I say a

higher and higher percentage of not even no not percentage because there's so many slo games that just come out but you could think of many examples of indie games that were like big and beefy and gave you everything that you wanted

out of them is what I would say now whereas back then there was less of a history of that there was more indie games were a little bit more like, well it's not quite a real game, but almost right? And and that's just been

right? And and that's just been something we've been gradually getting better at.

>> You, use, the, word, sophisticated., I, I would say that your games are sophisticated. And in fact, I think I

sophisticated. And in fact, I think I think they're probably too sophisticated for me. Um

for me. Um >> Okay.

>> I, always, watched, my, brothers, play, your games. I never played them. I would just

games. I never played them. I would just sit and watch because I'm an art guy.

And I want to talk about Word of the Siky Star art because it's so good. It's

so detailed. Um, but anyway, I'm an art guy. Your games are very sophisticated

guy. Your games are very sophisticated for me. I wonder if the level of

for me. I wonder if the level of sophistication of your games allows for the higher pricing of a game like Witness. And the

reason why is because you don't want just, this sounds so bad, but you don't want just anybody coming in and playing your games, right? Cuz they're going to get frustrated.

>> You, want, people, who, are, like,, "No,, I'm going to sit this thing through." Cuz

like you, would you say you have to grind through your games? It's it's a it's it's a special kind of grind right?

>> Well,, it, depends., It, depends, how, much you like puzzles and then especially with the new game, we're trying to make it feel less grindy, right? It's

definitely something that I'm conscious of.

>> Yeah.

>> Oh,, to, finish, the, witness, story, before we go further into it. Yeah. Go ahead.

>> I, will, say, in, hindsight,, I, might, have made more money if I priced it at $30 instead of 40. um just because even though it was for for these various

reasons a decent number of people felt like paying $40 for the game, it still was like above the general price point at that time. And so I think a lot of people like if you don't know if you're

like, "Oh, I heard about this game.

Maybe maybe I want to play it, maybe I don't." Right? I'm kind of on the edge.

don't." Right? I'm kind of on the edge.

Then the the price you probably think about how much is that price, right?

>> Yeah.

>> On, a, on, a, total, unknown., And, so, since, it turned out the game ended up being more mass market play, I mean it's it's not like, you know, mass market like a Nintendo game or something, but it's

like for a weird indie puzzle art game it it had a much broader appeal than I thought. And so we might have done

thought. And so we might have done better with a lower price back then. I

don't know. But but this is always, you know, when you make these decisions, you have imperfect, incomplete information and you just have to do the best that you can at the time. And your would you say that the the majority of your

revenue don't have to answer this question? Uh the majority of your

question? Uh the majority of your revenue came from console for most of your work. That's something that has

your work. That's something that has varied over time.

I would say that for both Braid and The Witness, and you got to take this with a spoonful of salt because it's been a long time since I looked at the initial numbers for either game >> but, even, like, let's, take, The, Witness

since it's more recent. It was roughly 50/50 on PC and console at launch. But

then as the game ages, you know, you go a few years later, let's say, and it's an old game now. I think Steam is a lot better at keeping old games alive >> in, part, because, so, with, consoles,, you, of

course you have several things stacking up. You have, oh, the hardware is old

up. You have, oh, the hardware is old now and I've moved on to something.

Maybe I've even switched sides in the console war or whatever. I bought cuz cuz the releases are sometimes staggered from each other and like oh I used to be an Xbox guy, but Sony came out with a

new console first, so I upgraded to that. And then the old game, I'm not

that. And then the old game, I'm not going to buy that cuz it's on the old console that I don't plug in as much. So

that's one thing. And then I think I think just the nature of the console stores is that there's just less discoverability of old things just due to the way the interfaces work and stuff. You think it's an interface

stuff. You think it's an interface problem?

>> Yeah., I, mean,, maybe, it, just, has, to, do with >> the, gamepad, and, not, being, able, to, type stuff very easily.

>> I, think, that's, it., It's, not, a, quick, It's not a quick a quick um browsability. You

know, on Steam, you can just click around and add stuff to wish lists. 10

games.

>> Now,, I, will, say, like, like, Valve, has, done a lot >> to, help,, right?, Of, course., I, mean,, I'm not I don't want to say they haven't done work there and stuff. I'm just uh

like I wonder if for some reason on PC all you ever had was gamepad, you know the the things that they implemented on Steam to search for games might be pretty you could imagine they would be pretty different from what's there now

right? That's all. But yeah, so so the

right? That's all. But yeah, so so the the point of that is just um as games get older, what we see is it's a lot more PC than console >> um, in, the, tail,

>> but, in, the, beginning, it's, pretty, even.

And that again that probably depends on how many platforms did you release for and all these things.

>> So, am, I, correct, in, saying, that, your focus your focus for Order of the Sinky Star is Steam? Would that would that be your priority this this in this era of gamedev?

>> This, is, where, you, got, to, be, careful, what you say.

Yeah, I figured I figured. So, so when the trailer premiered at the Game Awards, >> um, you, can, watch, the, trailer, on, YouTube or whatever.

>> There, there'll, be, a, link, in, the description.

>> So,, so, at, the, end, we,, you, know,, we definitely ask people to wish list it on Steam.

>> Mhm.

>> I've, heard, that, wish, lists, are, helpful.

So, we do that.

>> We, don't, say, anything, else, uh, because just plans for other platforms are not completely finalized at this time.

>> Okay., Um,, but, we're, we're, working, on other stuff and so we want to simultaneous ship hopefully on some consoles.

>> Yum.

>> And, and, Steam., Uh,, and, it's, just, it's not fully figured out. We may have more to say um in in the coming months about that.

>> Well,, I, think, maybe, here's, a, better question because I'm I'm trying to think about my audience and and also myself thinking like, you know, what do I what what would I what should I >> So,, as, an, indie,, here's, here's, what, I'll

say. If you're just trying to figure out

say. If you're just trying to figure out this question from first principles >> yeah, >> like, where, should, I, try, and, get, my, game on for people to play it? I do think Steam is still the most accessible place

where you have the easiest access to the largest audience. Like you can get on

largest audience. Like you can get on consoles. It's harder. It's not as hard

consoles. It's harder. It's not as hard as it used to be, but you do have to do certification and all these things. You

have to get like you have to get the game rated, although that's easier now.

>> Yeah., Um,, but, then, it's, just, very, easy to get lost on those stores as well.

Like completely lost, right? And that's

easy on Steam, too. But I think I think if you're like fighting to get people to notice your game, there's just a little bit more of like a a ledge hold on Steam that you could grab and climb up

>> and, it's, harder, on, consoles., And, so >> if, you're, an, indie, and, you, have, a limited amount of effort and you're like, I I can do one platform, where do I go? I I think Steam is still the

I go? I I think Steam is still the platform. And you know, like if you do a

platform. And you know, like if you do a Steam build, you've also done almost all the work to do like an itch.io or an Epic Store builder. You get access go right? You get access to these other

right? You get access to these other stores >> because, most, of, that's, just, making, it run on Windows or whatever. So, uh

that's what I would recommend to people who are asking this question. The

consoles are great, like for sure. Like

they they have a lot of people who play them, right, still, but they're just harder to get on and it's it's more effort and you can end up disappointed.

>> Yeah., Can, I, ask, you, um >> and, again,, you, can, cut, this, out, whatever. Why did you go with a

whatever. Why did you go with a publisher this time?

>> Two, reasons., Um

>> one, was, uh, a, lot, a, lot, of, the, things that we've been talking about is just the internet's really noisy. It's hard.

It's hard for people to um it's hard for people who are even interested in your game to know it's out now. And so one of the things when we we released the anniversary edition of Braid, one of the

things that we heard was like months after it came out, people were like "Oh, that came out." Like I I didn't even know, right? And so we wanted to

avoid that. And so knowing myself, so on

avoid that. And so knowing myself, so on the one hand, I do think that marketing and promotion is pretty important. I

mean, to me it's not the most important thing because the quality of the game to me is the most important thing. But then

if you make it and you fail to get it to people, that's an issue.

>> Yeah, >> it, is, just, harder., Like, back, in, the, day, I think back in the braid days obviously there were all these gatekeepers on platforms and stuff who would reject your game, but they were

still pretty reasonable. And if you made something that was good, you probably could find success in some way because it just wasn't that crowded. And now you can make something good and nobody will

ever know about it and it'll just languish. And so knowing myself though

languish. And so knowing myself though I'm like, look, this game is massive.

It's bigger and more complicated than anything else that we've done. As we

come toward the release date, we're going to, be, scrambling, to, get, everything done like you do. And like I'm just not going to do that good of a job at marketing the game cuz it it actually

takes a lot of time and energy to like set everything up and organize everything and figure out who to talk to and how to talk about it and all that.

And I was just like look I just know that given the choice between working on marketing and working on, you know making the gameplay better in the beginning, of, the, game,, I'm, going to, make the gameplay better. Yeah. and then we

might end up falling down on on this modern internet and not selling enough copies. So that was number one is just

copies. So that was number one is just teaming up with somebody who's who's going to do that, right? And make that engine run and I still have to participate in it, right? Because if you

know like when we do press interviews or something, I'm definitely there, but I don't have to take the initiative to set these up and figure out how to who to talk to and all these things anymore.

Um, and then the second thing is uh there was a a money component. We did

get some some completion funding for the game. Um, we put a ton of money into the

game. Um, we put a ton of money into the game so far. Fortunately, because, you know, both Braid and The Witness were financially successful games. Um, but it

turns out if you work for like nine or 10 years on a game, take take whatever yearly burn rate you think is reasonable.

>> Yeah.

>> And, multiply, that, by, 10.

>> It's, like, a, bigger, number., Now,

>> can, I, ask, you, what, is, a, reasonable, burn rate annually?

Well, it dep I mean, so for us, it's probably varied a little bit um depending on what we were doing and how many people we had full-time and then sometimes we would be launching stuff

where we have a bunch of contractors doing ports for a few months and then burn rate goes way up for a little bit.

Um >> you, don't, have, to, give, a, number, if, you don't want to.

>> No,, I'll, give, a, I'll, give, a, range, just to just to let people know. Um because I do think I do think this is useful information for some people. So again

we're a small company. We have around 10 full-time people right now and then plus contractors. If you if you count

contractors. If you if you count everybody who's working on Order of the Sinking Star with a partial part-time contribution and all that, it's probably 20 people around there.

>> Gotcha.

>> But, that, size, has, varied, and, with, the size, the burn rate has varied. So over

the past 10 years our burn rate has probably varied between uh $1.8 million a year and $4.4 million

a year. Like 4.4 is when we're really

a year. Like 4.4 is when we're really burning all the jets to like get something out.

>> Yeah.

>> Um, so, let's, cut, that, in, the, middle, and say it's like three million a year right? If you work for nine years at 3

right? If you work for nine years at 3 million a year, it's $27 million.

It's a lot of money, right? And so you could imagine like I could use, hey, I kind of ran out. I could use a little more just to finish the game.

>> Yeah,, I, think, I, think, that's, great.

Honestly, I don't I don't I I don't have that burn rate at all. Twisted Tower.

>> I'd, be, glad, that, you, don't, because >> Twisted, Tower, is, uh, less, than, which, this probably, does, a, surp, you you, you, took, a look at Twisted Tower, but um someone told me you did or something, but

>> Oh,, yeah., I, played, it, on, stream actually.

>> Yeah., Yeah., Yeah., That, game, is, is, like less than maybe less than 700K um total. So that's great over like

um total. So that's great over like three or four years.

>> I, mean, if, you, like, that's, the, thing., So

braid original braid like I was living in San Francisco which was not the cheap. It wasn't as expensive as it is

cheap. It wasn't as expensive as it is now but it wasn't the cheapest place.

But it was mostly just me and I was paying David Helman to help do art.

Edmond McMillan did some art on it.

>> No, way.

>> Um, he, did., and, uh, you, know, a, few, other like sound effects, a few other things but like I think I computed the final budget of that game as like $220,000.

Again, that's that's $2,08.

So that's more today, but it's like wow, I used to be able to do stuff in that budget range.

>> I, miss, those, days.

>> The, fidelity,, this, is, what the fidelity of Order of the Sinking Star is. At first at first when I saw

Star is. At first at first when I saw it, this is what's so striking about Order of the City Story and I want to get I want to get your thoughts on the game. I want to hear your pitch about

game. I want to hear your pitch about what the game is. But

>> when, I, first, saw, it,, I, was, like,, "Okay, it's like a 2D topown game with really pretty art." And then I like it's 4K.

pretty art." And then I like it's 4K.

The trailer's 4K, which I think it's 4K and 60 fps.

Um >> probably.

>> Yeah.

>> Which, is, beautiful., It's, so, cuz, I, So, I zoomed in and I was like, "This is all 3D. Every little thing is intricately

3D. Every little thing is intricately 3D right?

>> Yeah.

>> Gorgeous., And, the, shaders,, I, can't, even quite put a finger on them because, you know, I work in Unity and I I'm like I guess you wrote the shaders.

>> Um,, not, me, personally., Um,, I, wrote, a small number of the shaders, but you know, we've had a team of programmers working on them.

>> So,, the, question, I, have, for, you,, because I was talking to >> Ken, Lavine, earlier, this, year, >> okay.

>> And, also, American, McGee., Both., Okay.

>> Their, games, are, art, centric,, right?

Sure. Yes.

>> And, he, was, like, both, of, them, said you can't really make a game cheaper these days if you want to do art like that or because art is so expensive.

>> It, is, expensive >> even, now., Um, and, so, I, guess, my, question for you is is that what took up so much of the budget?

>> Um, I, would, say, it's, it's, a, mixture., Um

certainly right now if you look at the composition of the team, you know, as we try to finish up the game, who's doing what, uh we have more art team than

anybody else. We have five people

anybody else. We have five people full-time doing environment modeling for the levels. Um and you could easily wish

the levels. Um and you could easily wish that or I could easily wish that we had double that because the game's so big.

It's just again, you know, you got to keep you got to keep the budget somewhere, in, reason., Um,, but, the, the the game is really big. Um, so you know, you mentioned there's this kind of sort of top down is a little extreme, but like

let's say Diabloesque view where it's it's what in the old days would have been isometric, right?

>> But, it's, perspective, camera,, right?

>> It's, perspective, camera., Yeah.

>> Okay.

>> Um, >> sometimes, you, can, move, the, camera,, but usually you don't need to. And it's it's just this static camera. Gotcha.

>> You, know,, most, of, these, levels, fit, on one screen. Of those one screen levels

one screen. Of those one screen levels there's there's something like 900 of them. And so you can just imagine doing

them. And so you can just imagine doing art for that is a lot. And then we have this big overworld and then some of the levels are not one screen. Some of the levels are quite extensive themselves.

And so that's also more work. And so it just takes a lot. And so okay, so there's just a lot of raw work that people need to do. And then like early in development, you're just figuring out

what your game is. And so it's not like art that you do early on >> Yeah., is, going, to, be, good, in, the, final

>> Yeah., is, going, to, be, good, in, the, final game because like what's the look of the game? I don't know. Like I So when we

game? I don't know. Like I So when we started the project, I had highlevel directives that don't exactly translate into pixels on the screen. So the highle directives

are things like, dude, I hate it when I play grid-based games and the characters are all stick figures and it looks like they could walk past each other, but like they block each other and whatever.

So, we're, going to, have, kind, of, squat characters., We're, going to, have, squat

characters., We're, going to, have, squat objects. you're going to be able to

objects. you're going to be able to look. I hate it when I try to play a

look. I hate it when I try to play a puzzle game and I kind of can't tell where things are exactly because the art's getting in the way or I can't tell where I can walk unless I like go there.

So like let's preemptively solve all these problems. Um and then so so you have some set of these are the goals right? And then of course there's

right? And then of course there's implicit goals that you should restate sometimes like of course we want it to look really good and all these things.

And then you have to find your way from there to what do you actually do when you sit down to make a level, right?

Like what do the meshes look like? What

are the, you know, how do you texture them and stuff? And then especially if you're making the own engine, your own engine, you get the blessing of well, we get to make whatever shaders we want to

help these things look good. But that's

also kind of a curse because that's a big pit of effort and like this paradox of choice where like look because we could program any shader there's so many

things that we could do like how do we decide what we're actually going to do and all that takes a lot of time and you know um I I would say that because this game on the one hand it doesn't seem

like that weird of a game because it's like characters moving around on a grid based thing we know what that is right but like it is when you start seeing what it's about and how it works, it is

really weird and different from anything else. And it it just took a long time to

else. And it it just took a long time to figure that out.

>> Yeah.

>> Then, also,, here's, the, thing., When, I, work on games, both when I program and when I design and whatever, I have a very pragmatic approach

based on when I started in the games industry, it was so hard to get anything done. like there weren't existing

done. like there weren't existing engines that you could use. I mean

maybe you could you could license for $500,000 within a year or two of when I started, but like that was like literally for Quake one and so if your game was like Quake one, you could do

that, but if you're doing any other game, it wouldn't really help, right?

Um, and I didn't have $500,000. Um

>> it, was, 500k, to, license.

>> That's, the, number, that, I, remember.

>> That's, wild.

>> Yeah., I, mean,, I, think, it, seems, low, in retrospect, but like there just weren't that many 3D games back then >> compared, to, what's, available, now., It's

like Yeah.

>> because, most, stuff, is, like, free.

>> Yes., In

>> quotes., It's, a, very, different, world, now.

>> Yes., Yeah.

>> Um, but, so, just, the, point, being, everybody was making their own engines and figuring out how to make games and stuff and most teams who decided to go make a game just completely failed to even ever

have anything that you could play. Now

it's different. Now you could fire up Unity on the the test demo project whatever it's called, and have a guy walking around in 5 minutes.

>> Yeah.

>> It's, just, going, from, there, into, a, game anybody wants to play is the hard part right? That is.

right? That is.

>> But, back, then, it, was, like, getting anything that even runs on your computer was hard.

>> And, so,, um,, I, was, just, like, forged, in the fire of that. And now anytime I go do anything, it's like, okay, first step, get toward the goal as much as we

can. and in the process of doing that do

can. and in the process of doing that do some stuff that we know is maybe not good enough and what like if I'm programming something I'll be like okay here this function is a really stupid

hack I don't I can swap it out later I know I can swap it out later we're just going to get toward where we need so that we can then see what it looks like from there and see if we need to adjust and and if this is really going to give

us the experience that we think it will be you know is this going to be fun whatever so the problem is um for the beginning of this development so we made our own game engine. And the reason we made our own game engine is cuz I made a

new programming language. I wanted to make the engine in the new programming language. And our previous engine that

language. And our previous engine that we also made for the previous game was all in C++. And I was like, well, we could port parts of that. But really

we're kind of making our own engine. And

>> did, you, make, the, the, the, language?, It's

JI, right? Is that how you say it? J.

You made >> it, doesn't, have, an, official, name actually but >> Oh,, whatever., It's, the, code, name.

>> What's, it, most, similar, to?

Um just so I can >> what, I, would, say, is, that, it's, it's, in it's a descendant of C and C++ in that it is a systems language. It's not

managed. So you take pointers to stuff and you can destroy all the memory in your computer by doing that if you want.

Um but like syntactically it's a little bit of a descendant of some functional languages like uh there's a language called standard ML of New Jersey. um that it's a little bit

New Jersey. um that it's a little bit not that much but a little bit of influence and and some >> standard, ML, of, New, Jersey., And, by, the way, you're talking to a layman here.

I'm an art guy. I I know C#. I've been

>> Nobody, uses, that, specific,, but, that that's a dialect of a language called ML, which is not machine language. It

stands for meta language or something.

It's like a one of the branches of functional languages. A modern modern

functional languages. A modern modern descendant of that line would be like objective camel.

>> Yeah., I'm, trying, to, figure, out, and, let's shift gears and talk about um Order of the Sinky Star. I want to hear all about it. Why is this game using a custom

it. Why is this game using a custom language? What's what's the goal? And

language? What's what's the goal? And

tell me about the game as it relates to that and just rant about the game.

>> So, So, back, when, I, was, working, on, The Witness, you know, there comes a point where you're almost done with the game and then there's still probably a year left of development after that point right? But you get to the point where

right? But you get to the point where the game is done enough that you're like, "Okay, I see what the game is.

>> There, were, these, big, risks, in development like this game will suck if we don't do something well." And then you at some point you get past those risks. You're like, "Okay, we need to

risks. You're like, "Okay, we need to polish this a little more, but it's good enough." Like it does the thing that the

enough." Like it does the thing that the game needs to do for it to be a good game.

>> And, so, we, were, past, that, point,, but there was just like a lot of work left maybe like the last year and a half.

>> And, I, would, go, into, the, office >> and, I, was, programming, in, C++, and, it would just take like C++ is an ugly language. it would take a long time to

language. it would take a long time to compile the game because there would be changes overnight and then every time I make a change it takes a long time and then that's combined with these other things that I wish we had done a better

job about back then but like you know you always have these build processes that are part of working on your game like we've got to wrangle the data somehow like oh the art team made a

bunch of stuff so I'm going to come in in the morning and I'm going to I'm going to update I'm going to pull all those down from the central server somewhere and then that takes time and then maybe those assets need to be

processed locally in some way and that takes time. And so all these things were

takes time. And so all these things were stacking up and I would go in and I was kind of tired from working hard for years and I was just like felt really like blocked and unable to do work

because everything was so slow, right?

>> And, that's, a, feeling, I, felt, several times over my game development career.

This wasn't totally new, but it was new to have it like in a project that I that was my project completely. You know, I set it up and started it and all that.

And obviously there's a big team working on it, but you know, I made all the choices. So, they're all kind of my

choices. So, they're all kind of my fault at some level is is what I'm trying to say. And I and so I was sitting there, I would go in and we had things. So, we had a little, you know

things. So, we had a little, you know whatever you call it that runs down in the corner in Windows, whatever those are called, little widget that runs there.

at 2 update and pull down the data so that when I come in hopefully it's fresh but like >> that, was, broken, a, lot, or, you, would, turn it off a lot because you would be working on the code locally

>> and, the, bug's, not, fixed, and, you, want, to come in in the morning and resume from there.

>> Okay,, this, is, yeah, this, is, turning, into a long story but the point is yeah >> can, I, get, clarity, really, quick?, I, need clarity for your story.

>> Is, it, the, engine, that, was, the, problem, or the language? I would say that the

the language? I would say that the language was one of the problems. One of the problems was the way the engine handled assets like

so asset is a technical probably people know what that means if they watch your podcast but it means >> it, means, any, chunk, of, data, that, your game uses right sound texture map mesh whatever.

>> Yeah.

>> Um, I, only, say, that, because, the, first time I heard that word I didn't know what it meant and I pretended like I knew what it meant for a couple weeks till I picked up on it.

>> Yeah., We, we, just, could, have, made decisions better about the engine as well and it all added up but but very you know it probably only took like two minutes to compile the game or maybe

four minutes I don't know like if so in C++, okay, this, is, one, of the, pernicious things there's if you just change one file and it's the right kind of file in the right

place and you just make a small change you can do an incremental build which takes less time but if you there's these things called header files that are shared everywhere And those tend to be defining the concepts that are in common

across the game. So like if I've got a certain kind of entity and I want to change something about that entity, I probably change a header file and then I probably need to recompile most or all of the game. And so those are the things

that take minutes. Um, and that was that was minutes with us having put effort into making it not compile that slow right? Like some some games it takes an

right? Like some some games it takes an hour to compile when you make a change like that. I think people are aware

like that. I think people are aware enough that that's a big enough problem.

Anyway, so whatever. So, so go back to like 2013. I think it was 2013

like 2013. I think it was 2013 2014. So, somewhere in there, I was

2014. So, somewhere in there, I was starting to get really unhappy with C++.

And, I, was, like,, you, know what?, Computers

are so incredibly fast. And you prove this by running games. Like if you ran the latest Call of Duty that was out at that time or whatever, it's putting all this stuff on the screen, incredible

amounts of data. It looks great. It's

running shaders on all of it. And I've

made compilers for programming languages before. I know how much computation that

before. I know how much computation that is. Like if you were to somehow weigh it

is. Like if you were to somehow weigh it as abstract computation, it's way less than what we do in like one frame of rendering.

>> Yes., Um,, now, part, of, that, is, the, GPU, is implicitly parallel and it can do a bunch of stuff faster and whatever, but part of it is just somehow these programs are all a lot slower than they should be and a lot more the language is

a lot more complicated than it should be and they keep and then another part of it was I have all these things I'm trying to do that that I'm writing code to do like if you have a game you want

to write out your entities to disk and read them in and have a UI in the editor. Uh and and writing that kind of

editor. Uh and and writing that kind of code involves knowing what are the fe like if you have a position on the entity that's like some floatingoint numbers and if you have a a texture ID somehow right

>> the, system, needs, to, know, what, that, is, so it could pop up the right editor widget or whatever and like C++ doesn't help you with any of that for example they they maybe are starting to add something finally like 14 years later but but I

kept looking around at like what okay what are my problems as a game developer that I have that's making this hard and then what are people talking about in the programming language world? The

problems they're talking about solving has nothing to do with all the problems that I have. And so why don't I just sit down and make a language that actually addresses the actual problems that I

have really >> not, the, problem, fake, problems, people talk about and then maybe that'll be useful to other game developers as well.

And so I started that I think in 2014 uh just as a part-time thing like while we were still working on the witness it was like one of my side projects and then once the witness shipped you know we did

support on that and porting for a couple months and then I started the new the new language full-time uh in April or May 2016 and then uh the new game as

well at the same time. I have a question about coding languages as it pertains to AI and it's not what you think because typically typically it's like is AI good for coding and that's

>> no, I, figured, you'd, say, no, and, that's, I would love to talk about that but there's so much to talk about >> there's, a, lot, of, lot, of, conversations about that already >> how, much, I'm, going, to, I'm, just, going, to

sort of put together a series of questions to get to a final question.

>> Yeah., Yeah., because, I, was, talking, about with this this with my brother and he and I were talking about something that Primagen said and I think he's going to come on the podcast here. I don't know if you know Primagen but um

>> yeah, I've, talked, to, him, briefly, before.

He he we we were we were looking at some of his tweets and we were thinking how how much of binary how much of code how much of the process of code

>> how, much, of, it, is, weighted, down, and, by the way I'm a layman okay so I'm I'm trying to explain this but how much of is it is it weighted down >> by, being, wrapped, in, semantics, that, are

readable >> does, that, make, sense, what, I'm, asking >> yes, yes, um, I, would, say >> well, you, can, keep, going.

>> So,, so, my, question, is, >> yeah, >> if, the, what's, the, point, of, a, coding language? The point of a coding language

language? The point of a coding language is that it's readable to you and the computer?

>> It's, a, way, for, you, to, communicate, with the binary of the computer. Yes.

>> And, with, the, human.

>> Doesn't, AI, suddenly, remove, the, need, for languages?

>> It, actually, So,, it, doesn't, because, So, so do you mean a hypothetical interface where I just type into the AI?

It writes binary on the back end, but it spits out, it could spit out any other language you want. You could say >> "Yeah, >> tell, me, what, you, just, wrote, in, binary, in C so that I can read it in C#, but

what's actually being input into the game itself >> is, binary, so, that, it's, lightweight, and there's nothing wrapping the code in something readable."

something readable." >> There's, a, bunch, of, reasons, why, that's not >> going, to, be, true, anytime, soon., The, the first one is just so we use language models for all this stuff right now. Um

I don't think that that's the most promising form of AI for doing smart problem solving. Like I think I think

problem solving. Like I think I think the kind of stuff like like Alpha Zero right? The program that learned to play

right? The program that learned to play games and from from nothing that's why it's called Alpha Zero. Uh

>> that, is, not, a, language, model., It, does use some deep learning techniques. So

it's related, but it's not trained by like throwing all the text on the internet at it, right? It's like it's trained by like playing itself at the game. I think if we have something

game. I think if we have something that's good at programming eventually it would be more based on something like Alpha Zero, but then it would use a language model to talk to you, right?

>> Correct.

>> So, it, would, be, a, fusion, of, these, things.

>> Okay.

>> The, problem, is, just, Okay,, there's, a bunch of things. So using the L going back to LLMs for a second like if you ask an LLM you know please write me some

JavaScript code to do a thing or whatever right it's going to give you back source code which is in in the programming language so it's not doing the thing that you're talking about yet I think you're talking about a

hypothetical future >> a, hypothetical, where, it's, actually writing binary but it's readable >> so, so, the, problem, so, far, is, like, anytime it gives you that sample code it's going to be using So there's a thing called an

API, right? Which is just it's an

API, right? Which is just it's an interface to some code that somebody else made that already does some sub part of your problem, right? So like

let's say I want to make a 3D game. If I

have if somebody wrote some code that could draw an individual triangle for me, then I could use that many times to draw a bunch of triangles to put something on the screen, right? That's

not actually how it works anymore, but that's an example. Uh so the problem is those APIs also are only accessible through the interface of the programming

language. So like let's take C++ for a

language. So like let's take C++ for a second because that's a closer to what you're talking about where a C++ program is actually a binary on the disk whereas

JavaScript like is text and requires the browser to turn it into something every time it gets loaded right but so so code that's in C++ it's not just pure binary

that tells the computer what to do because a lot of it is like how to interface with other C++ modules and other parts of the system and all that and, so, I, mean, it, would, have, to, at least emulate all that stuff in the output

binary. But if it did that, it would

binary. But if it did that, it would still sort of be that binary from that language >> until, we, got, to, some, point, where, we're past that. I don't know. But I guess

past that. I don't know. But I guess what I would say, here's the thing that I would say about all that is if the computer is so smart that it can do the

whole thing from one prompt. Like make

me something. Make me a firsterson looter shooter, but the the combat is like Tarov but not broken and the net

connections are good and what you list out all the things you want and it'll actually oneshot it for you. if we ever get there, who knows, right? That's

amazing. So amazing that like none of us have jobs and all that, right? Um but

the problem is if you can't oneshot it how do you how do you refine it? Okay

one thing is you try to play the game and it's not like the experience that you wanted. Why isn't it like the

you wanted. Why isn't it like the experience that you wanted? Well, I

don't It doesn't feel right. When I'm

running and then I try to vault into the window, it feels sticky. And like why?

Why is that? Why does it feel sticky and weird?

>> Yeah.

>> If, you, say, to, the, AI,, "It, feels, sticky and weird," is it going to have any idea what you're talking about? Is it going to be able to fix it? Probably not. And

so, you need something that you can look at.

>> Yeah.

>> To, figure, out, why, it, feels, sticky, and weird. Even if even if you're not going

weird. Even if even if you're not going to edit that, but like having some human readable code representation in the middle,, you, could, at least, say,, "Hey,

AI, I think in function, you know, vault onto window, it's reading the wrong variable or I don't understand why it's doing, this., Can, you, at least, explain, it

doing, this., Can, you, at least, explain, it and fix it if it's whatever?" So, for the foreseeable future, that kind of programming is going to involve converging on something by steps. And

you just would need a representation like that as the human to like understand what's going on.

>> Yeah.

>> At, all., Right.

>> That, makes, that, makes, a, lot, of, sense, man. Um

man. Um >> Yeah., I, don't, think, I, have, any, other

>> Yeah., I, don't, think, I, have, any, other questions about that.

>> Okay.

>> It's, uh >> that, was, a, long, answer., Sorry., But

>> no,, it's, great., It's, great., The, last podcast we did, some some people in the comments were like, "Thomas doesn't know how to keep up with John Blow." And I was like, "Yeah, that's true.

um you you're on a different sort of level. I don't even know what

level. I don't even know what >> I, don't, know.

>> The, thing, is, programming, is, so complicated.

>> You, can't, expect, someone, who's, not, a professional programmer to like >> Yeah.

>> get, it.

>> I, can, I, can, program, C#, from, scratch.

That's I know how to do all that, which is not impressive to you. I know.

>> But, you, don't, spend, most, of, your, time doing that, right?

>> I, did., I, mean,, I, did, for, for, like, a decade.

>> Okay., Well,, that's, different.

>> And, admittedly,, you're, going, to, hate, me.

Admittedly, I got co-pilot running all day.

>> Well,, that's, fine., I, mean,, look,, if, if that >> all, day, >> if, that, meets, your, needs,, you, know,, go for it.

>> That's, what, that's, what, I'm, struggling with, man, is it does like it I I have knocked out I kid you not, I've knocked out a thousand feature requests and bugs

from beta testers, play testers with Claude running on the side. And the

reason why it works for me is cuz I've got my codebase for Unity all done over four years and it's written from the ground up custom with me and my team.

>> Claude, comes, in, and, what, Claude, does, is Claude is like this. I'm not trying to sell you on it. I promise. I'm just

trying to tell you this is I'm this is my struggle because I hear you say and I hear Prime say it sucks at coding yet I've got it running and it's like

solving all these problems. the game is still running 60 fps plus and I'm like I'm sure it got a little bit more bloated and I'm sure there's some junk in there which would probably drive you crazy.

>> Well,, I, mean, what's, an, example, of, a, bug that that you fixed?

>> So,, let's, say, um, for, some, reason, Thomas can't figure out why play testers are getting jittery movement on a cons on a gamepad, right? And so I can tell Claude

gamepad, right? And so I can tell Claude I can say why is this happening? And it

just scrubs through all of my controllers, my weapon controller, my camera controller, my character controller, the physics. It scrubs

through all of it almost like it's writing documentation and then it goes there it is and it finds the it finds the bug. It's incredibly effective for

the bug. It's incredibly effective for me and I'm not arguing with you. I'm

trying to I'm trying to learn from you and figure out what am I doing wrong here.

It's really good at finding bugs, which on the tail end of development. This is

what takes so long for me when I'm play testing and there's a thousand bugs to try and you don't even have to replicate them. You could just tell Claude this is

them. You could just tell Claude this is happening. Why is it happening?

happening. Why is it happening?

>> Yeah., I, mean, I, think, this, is, one, of those things where there's a there's a wide spectrum in programming, right? So

so one of the things that you'll hear say, one, of the, truisms, is, like, oh, in programming you actually spend 95% of the time debugging or something. Um, for

me that's actually not true. But it's

because I've got a lot of experience.

>> Um,, and, because, now, I, work, within, a system that, you know, when you're when you're in your own programming language especially like you wrote all the common code that gets you're just very familiar with it. And when something goes wrong

with it. And when something goes wrong there's this way that you can just be like, "Oh, I bet I know." Right? And

I've done this on stream before. I'm

just like, "Oh, I bet that's this thing." And you go and it is that thing.

thing." And you go and it is that thing.

Right? But, okay. Okay. So, so getting back to the the question, I think that um there's different kinds of bugs.

And what I would expect is that LLMs do better at some of the kinds of bugs than others. And maybe it's that a lot of

others. And maybe it's that a lot of your bugs are in this one category. And

so the kinds of bugs are, let me call them um I mean I want to say like shallow state versus deep state bugs but that sounds like I'm talking about the federal government or something. But

the point being like, okay, whatever whatever this gamepad bug is that you're talking about, it's like maybe the input was an integer and it was being cast to a float too late or something. I who

knows who knows what it actually was but that's an example of the kind of thing that could cause jittery motion.

That's an example where the code is just kind of doing the wrong thing like all the time. Um, and and you can tell that

the time. Um, and and you can tell that it's wrong by looking at the source code, right? Um, hopefully like if

code, right? Um, hopefully like if you're perceptive. And I'm actually

you're perceptive. And I'm actually surprised from that description that that this is an example of something that it found. But let's let's take that for granted. Um, most of the bugs that

for granted. Um, most of the bugs that are actually that I find that are time consuming and difficult to solve uh are what I would call what I wouldn't I mean

again I just said deep state bugs but like things that involve state of the game that's kind of complicated also with the compiler like state of the

compiler like we're compiling some subphrase of the programming language that has all these nodes in all these particular places they're all pointing at each other and whatever right

>> diagnosing, those, bugs, is, time, consuming because the thing that's going wrong is kind of driven by the data being wrong or

the data is in a format even though the data is right, it's in a format that is causing some combination of code paths to run that like isn't what happens in

the general case. th those to me tend to be the problems and those seem to hit the spot where LLMs are bad at. So my

experience with language models is, you know, um, and I think so, so a lot of people's experience with language models, which I agree with, is let's say you're just starting and you're like you know, make me a web page with some

buttons on it that does this or that.

It'll it'll do that, right? And

>> it'll, probably, kind, of, mostly, work, and you can fix up what doesn't work, right?

>> Yeah., But, the, thing, about, a, program, is um as a program gets bigger and more complicated um then mo the process of

programming the process of modifying that code and adding new code is about making like the new lines of code that you make have to agree with everything

that's there in the following way. Like

let's go back to annities or something right and this is a simple example like the way that we do the coordinate system in this game is it's like one unit okay grid based game like order of the

sinking star it's one unit per square uh the origin's somewhere near the corner of the level right whatever and so um oh and there's a gameplay position that's

floatingoint coordinates but they happen to be rounded to integers but they're still floating point to make math easy and then there's like a visual position which is a thing interpolating around between the

squares, right? But because it's a grid-

squares, right? But because it's a grid- based logic game, you don't use a visual position when you want to know if you are colliding with something or whatever. You use the gameplay position.

whatever. You use the gameplay position.

So these are decisions that are made early on. Yeah.

early on. Yeah.

>> And, so, any, further, code, that, you, make has to understand those decisions and interoperate correctly with them. Right.

And the problem even before you talk about LLMs, just one of the core problems of programming is that as a program gets bigger and bigger, there's just more and more of those things that

you have to keep straight in order to successfully edit the program. And if

you don't, the result is bugs, right?

Um, one one class of bug.

>> And, so, so, as, a, program, gets, bigger, and more complicated, there's like this really intricate like webbing. You could

imagine a big messy spiderweb connecting all these little pieces where if this piece has to agree with this piece about how something works. There's like a really thin line and and you know

software engineering is about trying to keep that from getting too messy. But

like it's only possible to do so much.

And part of the magic of games, part of the reason games are fun is because stuff in the game all interacts with each other, right? And that just means it's this system where where things mix

together. And so at some point when

together. And so at some point when programming becomes hard, it's because both the number of these little spiderweb rules that you have to track in order to add new things to the code

becomes very very hard to keep up with.

And then also the more code there is the more of this internal state there is representing the objects and all that.

And that becomes more and more complicated. And the more complicated it

complicated. And the more complicated it is, the easier for it is for that to be wrong. Now, the thing that I find about

wrong. Now, the thing that I find about LLMs is just that they can't track all those uh I don't think I said the word constraints yet, but that's how I think

about it. There are constraints that

about it. There are constraints that edits to code need to follow in order to be correct. And at some point very

be correct. And at some point very quickly, um the like you can start asking, oh, you you

you know, um I usually use I don't know.

I've tried a few different things but what whatever LLM you say hey please add a function to do this and it adds a function and you're like okay you did that but we need to also do this and

that because of these constraints with other systems and like it might kind of do what you say but it's not necessarily that good at really holding those constraints and then it'll

also start forgetting the old because it's hard to do all these things at once this makes programming hard and it'll stop start dropping the old ones And like at some point you just can't get it to do because it's it's not I

mean like I said it's a it's a language model right? It's not like uh it's not a

model right? It's not like uh it's not a chess player.

>> And, so, so, what, you, really, need, is, a chess playing AI and that's not what we have.

>> We, don't, have, that >> to, program, successfully., Now,, now, again, so whenever you make this kind of statement about AI >> everyone, wants, to, turn, it, into, like, some

extreme stance like, "Oh, you're >> you're, very, anti.", I'm, not, doing, any, of that. I'm not anti- AI. I think to the

that. I'm not anti- AI. I think to the degree that it's useful, you should be happy to use it. I just I do, however >> get, irritated, at, what, I, perceive, as, the

excessive claims made by a lot of people about what it does. Now, that said you're also sitting here telling me you have a lot of success with it. So, I

don't know. I mean, maybe there's just something about your code base that makes it more >> I, don't, know,, man., I, I, think, I, just, need the game, right?

>> I, know., Yeah., I, mean,, it's, my, game, is, is my game is admittedly very heavy. Um

you know, it it it struggles where it shouldn't struggle, you know, because it looks like a double A game from 200 nine. I mean, that's like the best case.

nine. I mean, that's like the best case.

>> But, that's, just, that's, hard., That's

independent thing, right?

>> Oh., Oh,, you, mean, just, for, the, amount, of geometry on the screen, you wish it were >> No,, I, think, the, geometry, isn't, even, the problem. I No

problem. I No >> you, know, what, the, problem, is?, The

problem is lighting. Unity's lighting is horrible. Um, it's because it by default

horrible. Um, it's because it by default every mesh casts a shadow and by default every light casts a shadow. So if you're not an experienced developer and you don't know that shadow renders cause

such a problem.

>> Yeah.

>> Um, >> so, you, put, a, lot, of, shadow, casting, point lights in the scene and it just and you're not you're not realizing that a point light is for spotlights, that kind of stuff. Whereas when you build your

of stuff. Whereas when you build your own engine, I'm sure you have to think about all this stuff. And

>> so, the, programmers, realize, that, stuff, but then you give this stuff to the art team and they just do whatever they want with it, right?

>> Yeah., Well,, let's, uh, let's, Dude,, I, like Do you have um let's say 25 minutes to talk about your game and show it and open it up and stuff?

>> Yeah., Let's, let's, make, double, sure,, but um I'm pretty sure Yeah, I've got time.

>> Okay., Can, I, ask, you?, Yeah.

>> What, is, the, pitch, for, this, game?

>> If, you, if, you, I, like, to, challenge developers.

>> That's, That's, your, Steam, tagline., It's

complicated.

>> We, do, try, to, distill, the, pitch, down, into concentrated form in the trailer. So, I

encourage people to see that. At the

same time, I'm not sure. I mean, it's 90 second trailer and it's hard to explain everything in there. But the Okay, the point is, uh, let me go back to my previous game. Um, maybe some people

previous game. Um, maybe some people have played that. It was a game called The Witness, and it seems very different from this game. At the start, you like solve some puzzles and the puzzles are like these abstract symbols that are on

panels and the first part of the game is you see new symbols and you don't necessarily know what they mean and then you sort of figure out by experimentation what they mean and how to solve the puzzles. Right? In this

game, it's like that. I got interested during The Witness in this idea of gameplay elements combining and generating more complex situations and I said, "Okay, what would it be like if

instead of combining like simple puzzle symbols, uh, we were to combine things that are themselves much more complicated, like for example, um, fully

complete games already." And so, uh, you know, here, um, this is the overworld from the game. And um when you start there's these four territories that

belong they have these like title screens, right? They belong to four

screens, right? They belong to four different games.

>> And, when, you, go, in, there,, each, of, these has uh totally different game mechanics.

And there's a big >> overworld, that, you, explore, uh, and uncover things and um you know, you venture out, you clear the fog, and and you explore more and more things. But

but at first everything that you explore belongs to one of these uh one of these four territories. And so for example

four territories. And so for example >> if, I, go, into, mirror, iss, Oh,, you, know what? Uh let's skip this. Um I I have

what? Uh let's skip this. Um I I have that turned off for it's like a placeholder for a cut scene that we would do or whatever. But like this game, uh this level is temp graphics, by

the way. It's like really old. Um, the

the way. It's like really old. Um, the

previous level has good graphics, but I I wanted to skip the the fake cut.

>> This, looks, great,, man.

>> But, yeah,, so, it's, it's, got, this, core mechanic about you teleport around and do stuff. Um

do stuff. Um >> Oh,, I, love, that, teleport, that teleportation shader is awesome. Or with

the particles. It's so cool.

>> Yeah., and, and, um, there's, some, fun surprises in here that I don't want to spoil, but uh you know um the basics are you know you can push stuff in the

water, you can push the mirrors around you know, you teleport and >> oh, man, >> you, know,, you, solve, some, puzzles., And, so this one world is about this guy trying

to deal with some situation and uh you know, and it's all about mirrors and all that stuff. If you go down here um,

actually, so I'm I'm trying to think about how do I best avoid We're just going to see some some levels without >> And, like, I, said,, John,, we, can, we, can, cut out anything you want.

>> No,, no,, it's, fine., It's, fine., So,, this is really old. This is like early temp concept art that doesn't match what we're gonna do. Um, but this is like a different world that has different

characters in it. And this one, um, so all of these four worlds are based on uh, games that were done a long time ago now, like back when we started development that you actually could play

free on on the web. And so Mirror Isles um, if you search for the Mirror Isles um, I don't know why the game's

>> hanging, right, now., So,, so, it's, all, based on games that already existed, and the idea was to take those things and combine them. So this is uh Heroes of

combine them. So this is uh Heroes of Socon which is a game made in a language called puzzle script uh by Jonah Ostroth. So okay, what's going on?

Ostroth. So okay, what's going on?

Something about this is one of these things that's interesting about game development.

>> I, think, something, about, the, game, capture that's happening is interfering with the audio like because it's >> it's, trying, to, capture, audio, or something and so the subtitles are not proceeding.

>> Yeah.

>> And, so, it's, just, like, frozen., And, so we're going to have to figure that out.

This is a good a good play test because you're going to have streamers playing this game non-stop.

>> Oh, yeah,, we, got, to, make, sure, that, I, mean I I actually run it with OBS doing the capture and that works fine. But

something about whatever's going on right now cuz so you know, okay, here's a thing people may know, right?

>> When, you, play, back, sound, effects, to, like a speaker in a PC >> Yeah.

>> you, don't, know, exactly, how, long, that takes actually cuz the clock for the sound system is a little bit different.

>> Yeah., And, so, if, you, measure, the, time, on the CPU, it's not going to exactly match. And so what you tend to do

match. And so what you tend to do is actually check the time from the audio system like how is it proceeding?

And but that that leaves you in this thing. If the samples aren't going out

thing. If the samples aren't going out for whatever reason >> then, the, time, isn't, going., And, so, this guy isn't, you know, this is one of these things we need to work out. Uh but

so this guy is a warrior. He could push an indefinite number of blocks. Let's go

back to talking about the game. And this

guy uh he he lives in this world where there's like Dn D style heroes, right?

And they all but they all have like Dn D puzzle abilities. Um there's there's

puzzle abilities. Um there's there's like floor buttons and stuff that you might expect. So like I can weigh down a

might expect. So like I can weigh down a floor button.

>> Did, you, go, back, in, time, just, now?

>> I, skipped, through, levels., So, I'm, just hitting a developer peak key to go forward and backward through the levels.

>> Oh, man,, this, is, great., Okay., Can, I, ask you a really quick question, John? That

block out, is that how every level starts >> in, development?

>> Uh, when, we, design, them?, Yeah., So,, so here I'll open up the editor really fast. So

fast. So >> like, if, I, want, to, edit, the, level,, like, I can add walls and whatever, right? And

like you know >> what?, So,, wait,, the, editor's, in, play

>> what?, So,, wait,, the, editor's, in, play mode.

>> Yeah., So,, so, if, I'm, in, the, editor, and, I want to play exactly what I'm editing, I hit control F11 and it runs that in the game right?

>> That, is, so, confusing., It's, a, different paradigm than like Unity. Obviously

>> I, I, don't, know, how, Unity, works., I, guess in Unity what I know is you hit the play button and then you wait 20 minutes for your game to start up.

>> This, is, true.

>> So,, so, going, back, to, this, again,, this isn't final art. Um, but this is uh >> why, is, this, not, final, art?, This, looks really good.

>> This, was, just, an, early, concept, of, what we wanted to do. It doesn't match. So

as we develop the fiction of the world and what's going to be in the world there's that kind of stuff. And then

also, it's just not at the level of refinement. Let me show you uh before we

refinement. Let me show you uh before we go on.

Um, >> this, is, it's, beautiful., These, are, these are some levels at roughly final art style.

>> And, so, we, have, a, lot, of, these, like volutric or fake volumetric effects happening in the background.

>> When, did, you, figure, out, that, height, was a visual advantage for this game?

>> Not, that, long, ago., Like, a, year, ago.

>> Yeah., Which, sounds, like, a, long, time, >> but, like, in, terms, of, um >> the, history, of, development,, uh, it's relatively new. Dude, this is

relatively new. Dude, this is incredible.

>> Yeah,, these, are, just, examples., So,, you know, I can go back in the editor somewhere like here. I don't actually when I use the editor, I edit the gameplay objects and not the visual objects. And so, it'll be

objects. And so, it'll be >> But, like,, you, know, >> is, this, real, time, lighting?

>> Yeah., So,, I, can, grab, a, light, source, and move it around and stuff, right? Um

>> are, you, going, to, sell, this, engine?

>> We're, going, to, give, it, out, for, free.

>> Of, course, you, are.

>> Yeah., It's, The, thing, is, it's, an, engine.

Um, it's not totally general purp like with Unity or Unreal, you could make a game without a programmer.

>> For, this,, you, still, would, need, to program, but maybe your LLM can do that for you and then you're fine. I don't

know.

>> Well,, if, it's, written, in, JI,, I, don't think so.

>> No,, it, could., It, It, does., If, you, ask, it, it'll it'll do some There's enough examples of it on the internet. It's

just >> it, doesn't, have, as, much, training, data, on that as it does on like, you know JavaScript or whatever. Um, but so we do stuff like we we have notes that we leave in the levels like like, "Hey, if

you're going to do art for this, please make sure that this hole stays visible and stuff."

and stuff." >> That's, funny.

>> That's, really, useful.

>> Um,, I, don't, know., We, have like we can randomize monsters and stuff. I don't remember how to do this.

stuff. I don't remember how to do this.

Let me see.

>> Oh,, you, can, fly, around, in, the, the, scene.

Okay.

>> Yeah., Um,

>> what, do, you, call, this, engine?, I, don't know how to use this. It doesn't really have a name. Um, we're going to have to figure out a name obviously when we give it out, but I don't I don't know yet.

Um, but anyway, yeah, I mean, I don't like the thing is I I use this all the time and so I don't exactly know pretty far here. I Oh, there we go.

>> Yeah., So,, there's, the, scene., You, know, we have, you know, there's some skybox and whatever that goes in the background. And then

background. And then >> yeah, >> this, is, wild., You, made, you, all, made, this like completely from scratch. I mean

>> well,, as, we, noted, in, the, beginning,, it did take a little while.

>> That's, why, I'm, like,, no,, you, should, sell the engine, you know, make some money.

It's okay.

>> One, of, the, things, I'm, a, little, bit, proud of, it's kind of subtle here. Actually

let me try to find another level that like our water effect is pretty cool.

So, you'll notice we try pretty hard to keep the grid readable. Like I said in the beginning, it's like I want I want games to stay readable, right? um but

not ugly. So like with gridding and stuff, you know, we have this thing where we can kind of paint >> again,, I, don't, I, don't, do, it, all, the time, but like if I open like the paint

panel, um I forget what channel like you can sort of, you know, in this way that you do 3D rendering, you use old style color channels to represent new things

because that's how the data is set up and that's how GPUs work. And so there's a thing I can do here if I get it right.

Um whoops. Uh hold on. Is it T?

I think I select the thing. I hit T and Oh, here I'm I'm clearing some grid lines successfully. Right. So

lines successfully. Right. So

>> what, in, the, world?, That's, so, cool.

>> Oh, yeah,, there, we, go., I, can, paint, some.

It's just it's painting really faint right now.

>> It's, like, a, splat, map, or, something., Um,

it's actually, uh, there's an animated like flip book texture for the grids and then there's an overlay that's like >> yeah, >> just, an, intensity, texture, and, then, maybe

multiplied by some noise. Um

uh, so we also it's >> it's, maybe, hard, to, notice, closeup,, but uh, there's like a also like a 3D normal mapping effect on the grid

>> that, you'll, see, when, the, when, So,, it's like the w the waves on the grid actually stick up. Um

>> oh, man., That, do, you, ever, do, you, ever struggle with >> it's, not, isometric,, but, it, kind, of, is, right?

>> Yeah.

>> Do, you, ever, struggle, with, like, this desire to show all of the perspectives of a level and have the camera move around and show cut scenes and zoom in and fly around various models?

>> We, don't, have, a, lot, in, the, way, of, cut scenes, but sometimes level geometry gets really complicated. And we do have levels with multiple viewpoints.

Um, I'm going to So, this one I think might be simple. So, you you pick up this map and it'll give you an alternate view, which in this case is just overhead, right? But it like it it helps

overhead, right? But it like it it helps it be readable. Um, but in in other levels sometimes we have Well, here um here you get like a top down

uh just to help you read where things are. So, we do that sometimes, but we

are. So, we do that sometimes, but we also >> I, want, to, see, different, perspectives, you know? I want to I want to go down

you know? I want to I want to go down and look at that pipe.

>> Yeah., Well,, the, problem, is, if, you, want to go down and look at the pipe, it means we have to do a real good job on the pipe so that it derives close-up scrutiny. And uh

scrutiny. And uh >> we, don't, want, to, do, that., So,, but, also just I don't know. I just didn't totally love it when I played puzzle game. Like

some puzzle games will just map like gamepad right stick to like rotate the camera around.

>> Yeah.

It doesn't make sense from the gameplay perspective. This is just the artist in

perspective. This is just the artist in me that I just want to see. I just want to see all the pieces, you know?

>> Well,, I, mean,, on, our, art, team,, I, think people are very relieved when they have fewer viewpoints to deal with because they're like, "Okay, we don't need to worry about >> having, things, look, good, from, that, >> dude., Look, at, that, scene.

>> dude., Look, at, that, scene.

>> Oh,, it's, so, pretty.

>> I, mean,, it's, Yeah,, it's >> that's, incredible.

>> I, mean,, it's, all, it's, only, right, in, the camera, right? I know it's smok and

camera, right? I know it's smok and mirrors but >> literal, smoke.

>> Literal, smoke, and, mirrors., That's, an incredible smoke. Wow.

incredible smoke. Wow.

>> Yeah., Like, the, way, the, way, the, light sources hit the smoke and all that is is pretty nice. Is

pretty nice. Is I don't think it ultimately was that hard to do the effect, but it took us a number of iterations and guesses to get there.

>> So,, you're, saying, that, you, guys, just figured out how to create that sense of depth with >> Yeah., Like, a, year, ago., Yeah.

>> Yeah., Like, a, year, ago., Yeah.

the game. I mean, when you showed me the other shots where it was very flat versus this, it's like totally different.

>> Yeah., Well,, I, mean,, so, that's, what, I, was saying about like >> it, it, takes, you, some, time, to, find, the look of the game, right? Like what what are we doing here and how do we do Okay I could go into a little bit of like how

this is accomplished, right?

>> Yeah.

>> Let's, um >> So,, this, is, the, level, where, I, Oh, god, it's doing it again. Um this is the level where I skipped the cut scene from. Um, so this if I go into like the

from. Um, so this if I go into like the designer mode, right? These are the islands and there's some water between them right?

>> Um,, and, so, this, we, just, sort, of, did this. This is an exact transcription. So

this. This is an exact transcription. So

the puzzle script game Mirror Isles which was designed by uh Alan Hazelden um, is exactly this level. And we just started by making it a 3D version of I

think it's we might have tweaked the island shape slightly, right? But so

>> so, this, is, like, tipping, your, hat, to, is this in the game?

>> Yeah., Yeah., Yeah., It's, it's, literally you start by playing that and then it it becomes more complex and merges. But so

so then how do you make these islands and the answer is well we have some variety of individual pieces. Um

>> like, there's, a, little, piece, and, we, just slam it onto the other pieces and then we just have some ways of blending the normals together. um you know the shader

normals together. um you know the shader sort of does that so that you don't because like any slight slope difference normally you would just see that these were two like blobs inter penetrating um

but you don't see that and then you can see that you know we have little detail meshes that we use >> hey, does, that, happen, automatically, or does the artist go in and do it >> these, are, hand, placed, right, now, you

could imagine making a mode >> where, it'll, sprinkle, them, automatically but >> I, have, no, idea, how, how, you, would, do, that I'm sure you'd figure it out >> and, then, part, of, part, of, the, art, style

was look, we want the island to look organic. Like, what's the point of doing

organic. Like, what's the point of doing good art if it's literally like cubes like this, right? So, we don't want it to be cubes. We want it to be close to cubes because you don't want any ambiguity about where can you walk or

where can you push an object or what's going to happen.

>> So,, we, have, this, kind, of, organic, style, but that still respects the grid. That

was very important. Um, and then you'll notice all the little detail objects also like respect the grid. Like these

little plants outline the grid squares.

>> And, that's, the, artist, who, did, that.

>> Yes,, that's, manually, placed, uh, by, the artist. Um, and then

artist. Um, and then >> so, so, one, other, layer, to, this, is, decals.

So um >> all, these, things, that, you, saw, uh, when the level opened, these are all projective decals, which is a a graphics feature that people might know about.

But if I grab this and move it around >> you, can, see, like, there's, this, big, corner line that I'm moving.

>> And, that's, independent, of, the, meshes, right? That just gets stamped down on

right? That just gets stamped down on them. And you know, you can scale these

them. And you know, you can scale these and wherever.

>> And, it, doesn't, have, to, be, a, flat surface. It can be any surface.

surface. It can be any surface.

>> Yeah., So, even, if, I, if, I, grab, this, me, Oh, well this is I mean it's projecting as I move it. Let's rotate this plane.

move it. Let's rotate this plane.

>> Yeah.

>> Yeah., It, just, Yeah., Um, so, that, was definitely a thing. We've actually had these in the engine since the early days.

>> Yeah.

>> Um, it, it, it's, it's, a, really, nice, way, cuz you know in the old days if you had some texturing on a mesh, if you were going to stamp out the same mesh a bunch of times, you would see the same texturing a bunch of times and that creates repeat

patterns >> that, are, undesirable, and, part, of, the look of a modern game maybe or I don't know. You could you could choose as part

know. You could you could choose as part of your style that it's a little retro and you see a lot of repeaty computery things, but for us that's not part of the style. And so a lot of these islands

the style. And so a lot of these islands are literally just these rectangles.

It's like these rectangles are stamped out >> kind, of, messy, like, and, then, these, edge pieces around it and then we use the decals and lighting and stuff to blend them together. See, this is the beauty

them together. See, this is the beauty of um a camera that's static because you don't have to really worry about optimization.

I don't think you're going to have problems because it's just like >> you, could, stuff, it, full, of, all, sorts, of assets.

>> Yeah., Um, that's, what, I, thought, in, the beginning. It turns out sometimes you do

beginning. It turns out sometimes you do mostly around draw calls. Um so we've as we've worked on the engine, we've pivoted it toward something where Okay.

So, so let's So, a scene like this, for example, uh is a little more complex because you could you could think of this plane down here as being like that island, right? And then there's walls

island, right? And then there's walls and stuff. So, it's more 3D and all

and stuff. So, it's more 3D and all that. So, what's in this scene? Uh

that. So, what's in this scene? Uh

there's some gameplay objects, right?

There's like characters, right? There's

like mirrors, there's like some light sources, but actually almost everything in this scene is this entity type called inanimate, right? And that just is like

inanimate, right? And that just is like a mesh with some colors and really not even that many other fields, right? You

can override some materials and stuff.

And so if I say how many inanimates are in the scene, there's like a thousand and you you see them. It's like almost everything visual. And so

everything visual. And so what we realized is >> static., Is, that, static,, John?, Is, that

>> static., Is, that, static,, John?, Is, that the same thing as static?

>> Uh, probably, like, these, things, Well, okay. These things for the most part

okay. These things for the most part don't move except that for some materials there are shader effects. So

like I don't know if we have wind turned on in here. It looks like we kind of don't. Um

don't. Um but these plants would move in the wind if you like made a wind entity and stuff.

>> But, that's, a, shader, thing,, right?

>> That's, a, shader, thing., So

>> you, can, have, a, static, object, with, a dynamic shader.

>> Yes., Yes., I'm, just, trying, to, clarify that.

>> Yes., Okay.

>> So, and, then, you, know, this, thing, back here is probably like a particle system or something, right? That's a different entity type. Um, these little bells that

entity type. Um, these little bells that are, Oh,, you, know what?, It's, not, moving because we're like not actually in game mode.

>> There, we, go.

>> Right., So,, um,, these, little, bells.

>> Yeah,, you, can, see, they're, moving slightly. Those are a different entity

slightly. Those are a different entity type that I mean the animation is not very complex, but it's this thing called a procedurally animating where you >> but, it's, not, a, physics, object., It's, just inanimated.

>> Yeah., It, doesn't, interact, with, other things. That's

things. That's >> but, but, the, point, being, uh, since, almost everything that you're going to render is this inanimate object type then if we just keep all of those on the GPU in

like one buffer once you load the level >> you, could, draw, them, in, one, drop, call, per pass. Now there's a bunch of passes and

pass. Now there's a bunch of passes and all that and so you know >> what, with, like, the, lighting, overlays, the the particles and stuff.

>> Okay., So,, a, lot, of, games, do, like, a, depth prepass first, which is just a way of >> making, sure, that, you, don't, draw, the, same pixel a bunch of times with expensive shaders.

>> Um,, so, we, do, that., Um,, we, do,, uh,, I mean, we're a forward renderer, right?

So, we're not doing any of this crazy deferred stuff. Although, what

deferred stuff. Although, what >> I, don't know., I, have, this, discussion with people and I guess there's there's a lot of question about what's forward and what's deferred. I mean, we

do have a number of offscreen buffers that we use for effects and stuff. Um I

could like flip through. So we have debug modes here. So this is just the colors >> of, all, the, textures.

>> This, is, um this is just the normal that we use for lighting. So like red is along the

lighting. So like red is along the x-axis, right? And

x-axis, right? And >> yeah, >> blue, is, is, up,, right?, Um

and I guess I guess that's tweaked from the geometry or I don't maybe geometry normal is in object space or something.

I don't know. Um, so vertex color is that painting that I was doing on the ocean um was writing into these channels. And so here in the ocean I was

channels. And so here in the ocean I was using it to do gridding, but for these guys it does like blends between materials.

>> Um,, so, maybe, I, let's, experiment., Like, I said, I never do the art, but let's I'm going to try doing this the same thing that we did for the ocean. I'm going to pull out the paint panel and I'm going

to zoom in here and let's see if I can change the way these look somehow. Okay.

It's Oh, wait. I got to select them.

Sorry. I want to select some things.

Yeah. See, I can like >> Whoa.

>> You, know,, and, that's, just, another, way, of uh creating variation among >> materials,, right?

>> I, love, that,, dude.

>> Yeah., This, is, actually, um this technique goes all the way back. I

think there was for Uncharted one or two they had a GDC talk that was a little bit about this but you basically pre-author some materials that correspond to each other right so I like

have I have fresh wood and I have weathered wood right and then um you know so as I paint I'm just crossfading from one to the other and

you can have that would be a simple blend um you can have more than one >> so, Um, I actually wonder. So, we have materials like blended two is between two

textures, blended three is between three textures and whatever. And you can tweak these channels to control the blend. I

just don't know. Okay. So, this is in the red channel because that's what I'm editing right now. Let's see. Yeah.

>> Is, there, something, in, green?, No., So,, I'm painting green. It's not changing. So

painting green. It's not changing. So

this is this is probably blended too right? But you can see how Okay. In the

right? But you can see how Okay. In the

old days, if you wanted a lot of variation, you had to go author a bunch of meshes or generate a bunch of textures or something like that, right?

And so with this, it's like I'm I'm using actually a bunch of the same mesh which is kind of efficient for the size of your game on disc, right? And then

you just have this other channel where you're writing in. And then you'll see so I'm going to go back to whatever that mode was. Um

mode was. Um it's not it doesn't put the text up. So, vertex

color mode right here is where I painted, right? Um

painted, right? Um >> yeah.

>> And, if, I, just, you, can, you, can, see, me painting that. And so, that's just one

painting that. And so, that's just one of the channels that gets fed in and then the shader does the blend.

>> Um,, I've, got, to, remember, not, to, check this level with all this messed up.

>> I, know,, right?, Would, you, uh, And, please tell me when you got to go, man. I I I could >> Oh,, yeah., I, forgot., We, don't >> But, let, me, just, Okay., Roughness, map,

which is how things respond to lighting.

Can I ask you a question really quick though?

>> Yeah.

>> Would, this, be, a, performance, nightmare >> if, this, game, was, open, world, where, you have all these >> I'm, going, to, call, them, splat, maps., I

don't know the right word, but just like these these masks that create all this texturing variation.

>> I, think, most, of, it, would, be, fine., I, mean a lot of this is is similar to things we did in the witness which was open world or is a descendant of it. Um

>> yeah., So,, for, example,, this, that, we're looking at right now is light mapping which there's some problems in here.

Like some of these trees aren't lightmapped and stuff, but but in general, this is um you know, in Unity you you'll you might turn on like ambient occlusion, screen space ambient

occlusion to put some extras. So, here

we actually have a very subtle screen space ambient occlusion, but >> for, bigger, like, see, this, shadow, down here, that's like a global illumination thing.

>> Yeah.

>> And >> so,, this, is, baked., Yeah,, this, is, baked, right? And I'm sure there's a way to do

right? And I'm sure there's a way to do that in Unity now, but maybe not all games >> real, or, global, elimination, with, baking.

>> Yeah.

>> Yeah., Yeah., Yeah.

>> Yeah., I'm, sure., But, so, we, did, that, in the Witness. Um we did some version of

the Witness. Um we did some version of this blend technique that I showed you a minute ago in the Witness, although it was probably simpler with fewer inputs right? Um the thing that might get the

right? Um the thing that might get the most expensive is uh the decals that I showed, but you could just fade those out at a distance right? And um and the classic thing

right? And um and the classic thing about like if I have tons and tons of light sources all over my open world scene, I have to find some way to deal with that. In really old games when the

with that. In really old games when the GPU was literally doing the lighting you had some limit like eight light sources at any time and you had to deal with that. Now, it tends to be more

with that. Now, it tends to be more about um you know um I've got to bin together how many light sources affect any given object and I've got to simultaneously

then make sure my shader is pretty fast up to some number of light sources but also make sure the bin doesn't make rendering slow and all that and then um well we have problems like what you were

talking about with uh you know if light sources cast shadows that's a whole other thing right?

>> Yeah., Yeah., Yeah.

>> So, these, light, these, shadows, I'm, looking at these are baked shadows >> except, for, the, character, ones, which, are real.

>> Okay., There's, so, there's, shadow, maps, and there's light maps. So the the the more well-defined shadows that you see are actually uh dynamic. So if I select the sun entity.

>> Yeah.

>> Um, where, where, is, it?, It's, over, there for some reason. But like um I can like you know change change all those. Right.

Oh, I see.

Light is dynamic, which makes sense.

Yeah, >> but, but, the, the, light, maps, um, are, uh are now invalid now that I moved the sun. They're still there. They're being

sun. They're still there. They're being

rendered, but like we would recomputee them, right, for the correct Oh, there's one other there's one other thing that I forgot to show >> that's, important, um, that, people, will

find familiar from other places. Um

what what's the panel? Is it this one?

Is it It's this one. Okay. So, um here this is just another way of uh

Oh, it's this panel's had a lot more stuff added. So, just like I was hitting

stuff added. So, just like I was hitting a key to pop between these modes, I can do this. But this has uh this has

do this. But this has uh this has different modes. So, so here um this is

different modes. So, so here um this is light maps. This is the baked part

light maps. This is the baked part right? So this is light probes which are

right? So this is light probes which are similar but different and they both combine to do the scenes, right? So a

light probe is like when you bake you put the camera at some point and you bake the surroundings into a cube map and then people nearby use that cube

map. And I think I could probably I

map. And I think I could probably I forget which one of these shows light probes. Oh, this does. So each one of

probes. Oh, this does. So each one of these spheres is like a a localized sample of the lighting in a particular part of the scene. and then meshes that

are near that use that cube map and then the artists just sort of place these based on where they think it's going to be most helpful to have them. And so the thing is I think it's important to have either

light maps or light probes. The light I mean it looks it looks kind of ugly here when you're just looking at the channel but um uh oh I think I could do Yeah. So here

we can look at the whole scene minus that thing. So this is without light

that thing. So this is without light maps. You can see it looks flat.

maps. You can see it looks flat.

Uh, here's without light probes. It

actually doesn't look as flat. So, the

light probes are helping a lot. They're

they're just not very good up in the foliage and stuff. And the light maps are good at at giving you >> So,, the, are, the, light, probes, in, this case, are the light probes only affecting dynamic moving objects?

>> No,, they, affect, everything., They, affect all the everything. It's just it's just in the shader.

>> It's, in, Unity,, light, probes, do, not affect static baked baked objects. If

there's a baked light map on something the light probe won't do anything. Oh

>> now, here, they're, additive.

>> Oh,, okay., Or

>> Okay, >> I, guess, that's, not, technically physically correct. Maybe we average

physically correct. Maybe we average them or something.

>> Yeah., Yeah., I, don't, know.

>> But, yeah,, you, can, you, can, go, without, one or the other. Um

>> so, we, we, focused, a, ton, on, art, here, and it I just want to let you know it's not because that's all I care about with this game. Well, I'm a very I love art.

this game. Well, I'm a very I love art.

I care so much about art. So, if I'm obsessed with it, it's just know that I do care about the gameplay. So, can we talk really quick about the gameplay?

>> I, sort, of, skipped, out, of, the, gameplay discussion accidentally.

>> Well,, I, want, I'm, so, fascinated, by, the ability to pitch a game idea in a sentence. I love to try and get

sentence. I love to try and get developers to do it.

>> Yeah., Okay.

>> I, can, imagine, you're, not, the, kind, of, guy who wants to do that.

>> No,, actually,, for, this,, I, actually, got, a lot of practice a couple weeks ago. Um

I'll try to summarize it here, but I'm also kind of tired of doing this because I did a ton of press interviews where I do this, but okay. The idea is part of the magic of game design is

when games are fun, right? Why are they fun? There's different reasons. One of

fun? There's different reasons. One of

the reasons that game designers uh work on when they design a game, and this is more obvious to see in puzzle games, but it really affects all kinds of games.

Um, is when you design objects that work together really well. And you know, in a puzzle game, it might be, hey, that mirror that teleports stuff, and then there might be some object that interacts with that in a cool way

right? But, but in all kinds of games

right? But, but in all kinds of games so like in a first-person shooter, uh you know, you design some guns and you design some monsters and do they create cool situations when you try to attack

this monster with this gun or whatever right? Or when you have monster A and B

right? Or when you have monster A and B in a room, is that interesting? And

monster A and C in a room, is that interesting? Right? So, so that's what

interesting? Right? So, so that's what game design is like. And so I wanted to make a game where we zoom in and really look at that and try to understand what

is that all about. Like that thing about things combining and creating delightful situations. And so in this game, we've

situations. And so in this game, we've taken these four worlds, you play them individually first, and then we mash them together, and you start playing combinations of the worlds.

>> Oh,, I, see.

>> And, we, take, you, through, all, these permutations of how things could combine. Um, and and we, you know, do

combine. Um, and and we, you know, do our best to design the levels to be fun to play and and delightful and all this stuff.

>> And, the, goal, is, to, like, really, shine, a light on what exactly is this fun thing about games where like stuff combines and generates situations that are

surprising, but but also you kind of should have expected them, but you didn't quite, right? And they generate >> delight, and, and, fun., And, so, that's, the ultimate point of the game. Okay. So, I

think I got it.

>> Yeah.

>> It's, It's, like, elements., It's, like, fire, ice water earth >> kind, of, >> but, different >> from, a, game, play, perspective,, which, is like >> if, somebody, masters, let's say, fire.

>> Yeah., Yeah,

>> I'm, not, saying, that's, what, this, is,, but it's like if somebody was to master fire and then suddenly that master went into water and then they had to master water then you take the both of them and you

combine them and then suddenly you out comes this whole new uh gameplay style. Is that kind of what this is?

>> Kind, of., I, mean,, but, you, could, think about it just directly from that elemental analog that you made, right?

So you could learn about fire individually and you could learn about water individually, but obviously water and fire interact, right? You could throw water on fire to quench it, which is not

something you ever would have known just playing in fire, right?

>> Yeah.

>> Uh, but, it'll, generate, steam, that, might scald you. It'll generate smoke. Um it

scald you. It'll generate smoke. Um it

>> it'll, have, changed, the, color, of, the surface that the fire had burned, which is not necessarily something you would observe, you know. And then maybe the water world is about all kinds of

liquids and not just water. And maybe

some of those maybe there's gasoline in the water world and you didn't it just smelled funny. But since there was no

smelled funny. But since there was no fire in the water world, you didn't think it's like water. It's just like water that smells funny. And then you realize when the things come together actually they're very different, right?

>> So, the, goal, is, for, them, to, master, it linear linear linearly.

So we start them in one world, they master that world, then they move on to the next one and master that one.

>> Yeah., So,, so, if, I, if, I, walk, out, a, little bit further. So if I start doing the the

bit further. So if I start doing the the RPG heroes one. So we showed the warrior, but there's like a thief who pulls stuff. There's a wizard who

pulls stuff. There's a wizard who teleports. There's a bard who charms

teleports. There's a bard who charms people and monsters. There's a druid who transmutes stuff. And a priestess who

transmutes stuff. And a priestess who protects. So there's like six main

protects. So there's like six main characters in this thing. And as you go from screen to screen in the overworld you drop into these levels where the different characters do stuff. Um, I

mean, I I should have probably showed that, but I'm skipping ahead. It's been

a long it's been a long podcast. So, you

know, you go through these screens where you're adventuring further and further out.

>> Um,, and, then, at, some, point, you, get, sort of to the end. So, by the time you get here, >> you're, like, by, the, time, you, finish, the screen, you're basically done except not all the levels are required. you know

some of them are optional and all that.

And um and so I've traveled a fair bit out to the north in the overworld and all the territory in the north belongs to the D and D Heroes game, right? Um

>> so, it's, like, four, different, games.

>> Yeah., Uh, in, in, sort, of, an, anthology,, but but then they come together, right? And

so >> that's, wild., So, you've, Are, you, saying that you've taken four different game genres and mashed them together into one? kind.

I mean, I would say there's subg genres of puzzle game maybe, right? But yes.

>> Um,, and, so, we, could, show, this, is, one, of the things that we do in the trailer.

Uh, we've sort of been doing hero stuff in the in the north. Let's just show one other character really fast so you get the idea. Um

the idea. Um >> I'm, just, hitting, some, cheat, keys, to, like shoot levels so I can go in them. Um, so

this guy's a wizard. We do this fun thing where we reprise the levels over and over. This is the same level you saw

and over. This is the same level you saw before, but it's a different character.

And um this guy uh actually you know what let's do all of them is real fast.

So we saw the warrior he pushes he could push a lot of things. He could push an unlimited number of things right. Um we

have the thief. She can't push anything at all. So objects act as walls to her

at all. So objects act as walls to her but she could pull. Right. And her

pulling is a compulsive aspect of her personality. Uh so she can't just walk

personality. Uh so she can't just walk to the right here. She has to pull.

Right. Um

>> Oh, see., And, then, with, the, wizard,, he's also got this ability. He teleports, but he has to teleport if he can because he's so interested or obsessed with his teleportation ability or whatever. So if

he tries to walk left here, he'll just teleport swap with that object. And you

know, we go through some levels where we combine them two two characters at a time. Here's like a three character at a

time. Here's like a three character at a time. And what what comes out of this is

time. And what what comes out of this is like a comedy of errors kind of thing.

So we saw this floor button briefly.

It's like, okay, I hold down the floor button. It'll open the gate. Um, I can

button. It'll open the gate. Um, I can get out. Uh, what? And and but but we

get out. Uh, what? And and but but we can't stand on the button because then that character can't get out. So, we

need to get that crystal that's in the corner. I'm going to send the thief

corner. I'm going to send the thief around and she can like pull it, but she can't actually pull stuff around corners cuz she pulls stuff directly behind her.

So, it's like, let me send the wizard to like teleport it.

>> So,, he, tries, to, walk, south, and, he, just walks swapped with the warrior and he's like stuck in this cubby hole. So, the

warrior has to come down. the wizard can swap with him and swap with the thief and get the thing around the corner. And

now the thief could pull it again, but now she's stuck in a little cubby. And

if the wizard tries to come back to help her, he's teleported the block back, but not as far back. And so she can pull it again. And so there's these combos of

again. And so there's these combos of the character abilities. And this early in the game, it's still um the puzzles aren't super deep conceptually

because you're still learning about the abilities and how they work. But as we go further into the game, it starts to be more and more about surprise. So

then okay so >> so, suppose,, well,, I'll, show, one, other fun thing since we showed uh the characters now. Um, progress in the

characters now. Um, progress in the overworld. At first, it has this kind of

overworld. At first, it has this kind of general unlocking mechanism, right? So

like I cheat solve this level with a developer key. It opened this, right?

developer key. It opened this, right?

But then once you've got the basic mechanics under your belt, it starts exploration and progress becomes more about the ideas that you saw in the levels, right? So you play the levels

levels, right? So you play the levels you see something like, oh, wizard teleport works a certain way, and then you use that in the overworld. And one

of the ways we do that in the heroes zone is we have these wardrobes. We

didn't talk about the character you're playing here on the on the overworld but she's different from any of the characters in the games. She can walk through these wardrobes and take on the character abilities by putting on the

outfits, right? And so

outfits, right? And so >> so, she, could, become, the, wizard, and, now she could teleport, but she gets stuck in just the same way as that level. So

>> hm,, if, I, want, to, get, to, that, level, on the right, I could become the thief. I

could like pull this over and then maybe I become the warrior and I can push it some and and so forth. And so, progress on the overworld starts to look like that. Now, so as I to get back to your

that. Now, so as I to get back to your question, as I went linearly north, I would have solved enough of heroes. I'm

just going to walk through the walls brazenly right now. I would have gone uh I'm in cheat mode, so this otherwise would have killed me when I walked in front of this dragon. But I would have

solved this. I would have got around to

solved this. I would have got around to this back passageway. Um

that lets me uh clear out another way to connect all these rooms, which is helpful. um you know this is back uh

helpful. um you know this is back uh this is these two rooms to the south but then also I get exposure to this gold

room that I can solve by uh doing things that, involve, both, of, this, is, kind, of a spoiler all right so if you don't want spoilers of the game skip ahead or maybe

we're close enough to the end uh hit the stop button but um I don't mind spoiling things because the game is so big and there's so many surprises that it kind If you get one thing spoiled, it doesn't

matter. But the point being

matter. But the point being >> if, I, got, far, enough, to, get, to, this, back alleyway from heroes, it means I basically solved enough puzzles with all the heroes that I get it right. I get

how it all works. So, here's another wardrobe where I can drag around this.

And there's this floor button, and like I want to get through those gates, but I can't get through all three at once because they're like in opposite modes.

So, >> yeah., Um,, but, I, could, at least, maybe, put

>> yeah., Um,, but, I, could, at least, maybe, put this crystal on it and then there's one gate, and, I, could, at least, get, up, to here, but I don't know what to do. Okay

so uh if I This is a fast travel area by the way, which sort of gradually materializes as you explore more of the game. If I go over here, and again, I'm

game. If I go over here, and again, I'm just cheating, but if I went through mirror aisles and I did a bunch of these puzzles, I would have gotten up here and then opened my way over here, and then

there's this mirror out here sort of with conspicuous placement. Um, and if I solve these levels, it thaws out the mirror.

And then remember how I was teleporting the the guy in mirror aisles. You can do that with these crystals. Um, and so I can toggle the crystal back and forth

without touching it anymore, which means I can go back here and by using, you know, I use the thief ability to position the crystal and then I use the mirror to toggle it, I can actually get through these gates.

>> So, the, overworld, is, a, a, huge, puzzle, box as well. Yes. So, um I'm only going to

as well. Yes. So, um I'm only going to show this because we spoil it kind of in the trailer, but like if What's going on? Oh, yeah. Wrong thing. Um if I clear

on? Oh, yeah. Wrong thing. Um if I clear out all the fog and I zoom out, you can see like the scale of >> Holy, crap.

>> Yeah.

>> Is, this, going, to, Is, this, going, to, cause some problems? Do I need to blur this

some problems? Do I need to blur this out?

>> No,, no,, no., It's, fine., I, wouldn't, The only thing I would say is if you look there's definitely some stuff that's blocky where the graphics isn't done.

So, avert your eyes. In the trailer, we have fog over those areas cuz they're not done yet.

>> Yeah,, like, the, brown.

>> Actually,, I, can, make, that, happen., So,

trailer, fog., All right., So,, now everything is obviously quite done >> right?, Um,, okay., So,

>> right?, Um,, okay., So,

>> what, a, dude,, here's, the, one, question that kept running through my mind.

>> Yeah.

>> Is, this, game, meant, to, be, beaten?

>> We, have, play, testers, who, are, just, about done with the game now. Uh, so it is possible. But did you have to give them

possible. But did you have to give them hints and tell them what to do?

>> No., No., In, fact,, the, version, of, the, game that everybody will play is easier because they they had to fight through like buggy levels and levels that had exploits and just broken stuff. And so

yeah. No, it is what I would say is um we're very conscious of this issue that we've built a super big game that is great for people who want to play a super big game, but not everybody wants

to sign up to be in like the puzzle army, right? So, So, for people who

army, right? So, So, for people who don't want the biggest puzzle game ever or whatever, I there's several games that could make claim to that, but whatever. A a very big puzzle game. For

whatever. A a very big puzzle game. For

people who don't want that, uh there's a multiple ending structure, right?

There's three endings, and the first ending doesn't require you to play through literally the whole game, right?

It is a satisfying, it's a good ending.

It's a satisfying ending, but it doesn't require everything, right?

>> I'll, end, with, this,, man., This, is, This, is beautiful. This is incredible and and

beautiful. This is incredible and and it's it uh it looks like it was it looks like a lot of sp money was spent on it.

>> I, wish, we, could, have, spent, the, money efficiently, right? So

efficiently, right? So >> no, no no, this, is, incredible, and, it's it's it's just amazing and I think what we what you've done here has just shown you know that building the engine and

the game that was that was the 10-year journey. It was the the engine the game

journey. It was the the engine the game >> programming, language., Yeah., Yeah.

>> All, right,, John., Thanks, so, much,, man.

>> Thanks, for, having, me, on., Good, talking, to you again.

Yeah, dude. Uh link is in the description for anybody who wants to wishlist. Let's get those wish lists up.

wishlist. Let's get those wish lists up.

>> Yeah,, the, wish, list, is, great., We, also, uh we have a Discord that you could drop in on and talk to people. I tried to link it from Twitter and then Twitter rewrote the link and like it was broken. So, but

we have that there's there's a Twitter account for the game and you can you can follow all these things.

>> Okay.

>> Yeah.

>> All, right,, man., Thank, you, so, much., Take

it easy.

>> Thanks., And, by, the, way,, if, you, want, to make your own indie games and get paid to do it in 2026, I have a massive gamedev program and it's called full-time gamedev and it's 50% off for New Year's. This is a big one. And yes

New Year's. This is a big one. And yes

these are actual reviews from real students on the front page of my website. Many of these students actually

website. Many of these students actually went full-time with the help of the course. That is so cool. See, in the

course. That is so cool. See, in the program, you're not only going to learn how to make a game in 2D or 3D, and you can pick whichever engine you want to learn, Unity, Unreal, or GDAU, but you're also going to learn how I secured

funding from publishers and crowdfunding or maybe even hit the Steamfront page multiple, times., And, I'm, going to, show

multiple, times., And, I'm, going to, show you as best as I can, how I did it, and perhaps how you can do it, too. Yes

I've gone full-time on three separate occasions with just a demo. But don't

take it from me. Click the link below and check out the reviews for the program. It's 50% off again for New

program. It's 50% off again for New Year's and you're also going to get my 2D and 3D art programs completely free but again only for this sale. It doesn't

last forever and there's just 100 seats available. Let's make 2026 the year that

available. Let's make 2026 the year that you master gamedev and who knows, maybe you go full-time. All right, click the link below and let's jump into the conversation.