Why He Chooses Humans Over AI

This is Alicia Keys. She has a bespoke piano and we recreated that in 3D. So this is not a real piano.

It looks so real. It looks like she's there, she's playing.

So everything you see here is running in real time, in Unreal Engine. And we are pushing the buttons and making this whole world come alive.

It's definitely not something like a Google's not on Banana and to just generate some images and stitch them together, it's definitely not.

the way Will Smith eats, spaghetti in AI eight months ago and now it's incredible what they did.

Our job, we need huge resolutions.

You still miss that control.

Like if I have my 3D artists create a whole world and animate the trees or birds or certain things. We have complete control over that. And with ai I am prompting you don't have complete control yet.

But if you like, would have a machine create a show, it would probably, for now just give energy all the time.

A show should be like a sinus. And for a machine to replicate that, there's no way. I mean, you need humans to build a very good show.

I prefer having a real person play a violin over a robot.

Why a

robot? Was trained in AI to play the violin and that person trained 25 years before the person could play the violin perfectly.

I think some things shouldn't be touched.

My next product is gonna be exactly about that, bring classical music, but in an immersive experience, we have real musicians. We recorded an orchestra of 120 musicians. It's still humans at the basis and technology enhancing what the humans are doing.

We shouldn't use AI because it's popular right now. No. Use it when it makes sense.

Today I'm talking with Yuris , CEO of Primax. they're the team behind the visuals for Tomorrowland They created some of the most demanding visual experiences in the world. So 16 K resolution, real time synced to live music. So for anyone building AI vision products, he holds the highest possible bar.

thanks for joining.

Thanks. That's quite an introduction.

I do think so. I think You are the highest possible bar, looking at all those AI visuals, right? We

always try to do something new, something that nobody has ever done before, and using game technology to do so.

I'm, uh, Yuri,I founded the company. Exactly 20 years ago. so we started out as a, a graphic studio, I would say, but we were mainly making visuals for the entertainment industry.

So when you have a fashion show or a nightclub, all the visuals that played there, that's what we were creating using, now, I call them old school tools like, after Effects from Adobe or Cinema four D. From Maxon. we used to create lots of visuals with that for say, 18, 19 years.

but still before COVID. And then, all of a sudden COVID hit. And since we were in the event industry, all, everything stopped. we were a small company, still eight people. But I had really had to start looking like, woo, till this is not good. what should we do?

When do we go bankrupt? Even? And then I just, on one evening, me and my business partner,we're both gamers. We were playing a game online. We were like, we've been playing this game now for a whole month, just because there's no work. Let's just reinvent ourselves and use the technology that we are playing in right now.

Unreal Engine, let's start using that. For, for what we do. so we contacted some of our clients, the biggest one of course, being Tomorrowland. And we told them like, Hey, what if we, your festival can't go? We can't happen this year because it's COVID. should we do that online? Should, let's create it in Unreal Engine and see what that does.

And they said, yeah, sure, let's do it. So we thought, ah, we're gonna create one stage or two stages and we are gonna record it. Eight or 10 DJs, but, in the end we, it was eight stages with a hundred DJs running for two days. So it was, a big undertaking, but we managed to do so. And from then on. We just started using unreal engine game technology in everything we do.

we started creating lots of, immersive, experience with it. We started, using it for more virtual production, virtual festivals, but then when the world opened up again, the virtual event thing kind of died. we were like, we have so much know-how now in Unreal Engine, let's create a product.

Where we can run visuals on a big screen at festivals, at concerts, but not run movies like we do, like small video clips that we mix together. But let's do a real time 3D world that we can manipulate on the spot. And that's exactly what we're doing right now all the time.

the best of both worlds.

Rightly,virtual and live together.

Both together and thanks to game engines like Unreal Engine, we can really get that graphical quality that we will get from our old school way of working. The only big difference is it's in real time. Doesn't have to render anymore.

What does that mean?

So before, in the traditional way, what we used to do is we would create a whole world in 3D, put all the lights, put a camera, and then to create the actual movie, you have to render it. So GPUs or the back in the day were cpu. Now it's more GPUs that are rendering this. They are calculating how the light falls, how it bounces, how the reflections are.

like Hollywood movies, like Pixar movies, they would take like a day per frame sometimes to render, just to calculate. But when you have a real time gaming engine, everything calculates in real time. 'cause video is. 25 pictures. 25 frames. in Europe it's 25. In America it's 30, pictures per second that makes a movie.

And each and every one of those frames needs to be rendered. And in a game engine, you can render that in real time. You can manipulate it in real time. Like before, if we had a big theater show and the director would come in and he would see our graphics and he would say, yeah, I want this forest to be a little bit lighter and have.

A bit of a different, vibe, then we would have to change that in our software and he could come back in 48 hours when it was rendered. Now we can just adjust all the lighting, adjust the textures, whatever parameters, the result will be instant. There's no more rendering.

The rendering is happening in real time.

Why is real time needed?

flexibility, I would say, no more waiting on rendering, and more creativity. For example, when we have a huge stage. We can put big flames. you have these big fireworks and flames. and if you have say, 10 big real flames that are, that are going, we can multiply those flames in our virtual world on the screen and have 50 more.

And the data coming from the fireworks guy, pressing the button, doing the real flames. That signal will also go into our software and translate it to Unreal engine and make our virtual flames go off at the same time. And then you can really extend the real world into the virtual world.

So there's an interaction component that's happening.

Absolutely. We

can use whatever data coming from whatever place,to interact with our visuals. We have another, projects, it's a world that is being influenced by the real world. So it's a virtual world, but it like has the same weather system as our real world.

So what we do is we have an AI miner. He's mining all the weather data,of that area and sending it into our software. so that the weather in our virtual world displayed in somebody's home is reflecting what is happening outside. Mm-hmm. A window, but a window on a virtual world with the same weather as the outside world.

Wait, it's a window right? it's

screen, but I call it the virtual window well, a window into a virtual world, I would say.

What do people use that for? I sitting my home and I wanna experience Mars weather. I'm just kidding. Maybe somewhere in the desert.

That's exactly what they're using it for. For, they want to be Mar, they wanna be on Mars, they wanna be in the desert, they wanna be in Paris. They wanna be wherever you could imagine, and you can just simulate. What is happening there in 3D in real time. You don't have to put, a million cameras around the world.

You can just simulate in a 3D world what is happening on the other side of the world with data coming from there, that's being mined by an ai.

I'm curious,what are, the users actually wanting to experience?

They're looking for something different most of the time. they're looking for something in their home that nobody else has.

and these are people that don't have lots of spare time. They're CEOs, very high placed people That wanna do something completely different. Like in the morning, if they wake up, they wanna be able to have a run in the Amazon Forest in the afternoon. In the evening, they wanna have dinner on times square with their wives without moving.

Anywhere in the world without,

without actually going there.

Without actually going there. Without taking the private jet and flying there. Yes.

how is this experience created? it's not on my tv. It has to be a,

it's a huge screen. like now we have a concept where all the walls in a room, we can create a room and all the walls are a hundred percent video screens.

LED screens. That's crazy. An LED screen is made up out of the word says in itself, small LEDs right next to each other, right? And you can make them as big and as small as you want in very high resolution. For example, the one we use now for this experience is here. the distance between the dots is 0.7 millimeter.

That's very sharp. But it creates very big resolutions. like one room can be. thousands of pixels wide and thousands of pixels high. So we need a lots of computing power to then generate this world that they're standing in, in real time in 3D. So there's lots of Nvidia GPUs working, working on those world all at the same time.

All synchronized.

So luxurious, I have to say. I guess it's not a,

it's

a niche.

Do you think this technology will go into like ordinary people's lives at some point? It'll be cheap enough to like everybody can experience that?

I think, yes. if you look at it an LED screen, when I started out 20 years ago, it would, nobody could afford it.

Now it's starting to pop up in these people's houses. Like they don't watch regular televisions anymore. Just the whole wall is a big tv. It is becoming cheaper and cheaper Give it another 10 years and it will be accessible to everybody.

I will send you a message and oh, I finally can afford this virtual TV room.

You could achieve the same with projection if you have a completely dark room. with wild walls, it's less immersive than an LED setup. The LED wall is much more powerful, but. You can get the same sense.

for people who doesn't have any visual understanding of, for instance, what is Tomorrow Land? do you have an example?

Tomorrowland is, the biggest and the most beautiful dance music festival on the planet. I would say

they exist as long as we do, and they've grown from, a festival being attended by 18, 20,000 people.

Per day to now. a festival that is being attended by, in one weekend, 400,000 people. So it's a big one.

is this similar to the UK's uh, Glastonbury

Glastonbury is more like a rock festival, I would say. Tomorrowland is all about electronic music and Tomorrowland, they really build like a fairytale world.

With the stage that is like 150 meters wide and 40 meters high with an attention to detail that no other festival can match. that's what makes them so special. You should be able to see my screen right now.

Yeah, I can see your screen now.

So this is our demo reel of, past projects, that we've been working on.

See big, this is a big dance event. This is, more pop music. It's a virtual festival. This is Alicia Keys.

Are these real or like the person? on the screen,

it's a mix of real things and and unreal things I would say

It's hard to tell.

This is also a real festival. This is Formula One. This is tomorrow and last year. This as well. big stage using, our real time software. I would say this is the opening of DOTA two, a very big gaming event. Speaker:

So those are some of the projects we've been working on.

And then one of the biggest ones we did, last August we start, oh no, this is the virtual festivals. I'll, tell you a bit more about that. So here we put at the DJs are real and all the other things that you see around the dj. There not real, they're added by us. It's a complete 3D world, and this is what we used to do during COVID.

people used to watch these festivals from their computer. they used to watch them on projectors in their home. Have a small private home party. And this is a virtual studio that, where we created the graphics together with the Belgian national tv. And everything is interactive and it looks like the people are standing inside this virtual world.

Same with Alicia Keys. We put her in a virtual world. The only thing that's real here is her. Even the piano wasn't real.

Really? The piano was completely real to me.

no. She wanted to bring her piano, but it's very heavy. and we didn't have lots of time to record her, so we recreated her. she has a bespoke piano and we recreated that in 3D.

So this is not a real piano.

it looks so real. It looks like she's there. She's playing.

Yeah. in August, we did the biggest real time show we've ever done. Uh, 16,000 by 16,000 resolution all in real time. so everything you see here is running in real time in, in Unreal Engine.

We are pushing the buttons and making this whole world come alive, and we're using data from the light guys, from the laser guys to sync up all the effects on the screen with the effects that are happening down there with the lights and the lasers.

So you get a very interactive, worlds that really is animated on the music during the performance.

So those are some of the things that we've been working on lately.

these are amazing technologies. crazy. it's definitely not something like I can use Google's nano banana and to just to generate some images and it's stitch them together.

It's definitely not, it's,

it's a different, I mean, we don't use nano banana, but we use Midjourney for now. but I've been using it for a long time. We mainly use that, in the pre-process, the product, the pre-production process, I would say, like a client comes to us, they say, ah, we have a festival and we want to have, lions as a theme, for example.

Then we will go to Midjourney and we will generate, a whole storyboard and a mood board. We used to draw that out. I really had a guy sketching everything out by hand.

Now we prompted out, so instead of spending one week, two weeks, making a proposal, we now spend one or two days and it looks 10 times better just because you can prompt it.

And the result that's coming out, I've started using version 1.0. Now we're at, I don't know, 7.2 or something. And the results they get, they keep getting better.

And we've been doing experiments with, what comes out of mid journey to then animate it as a video, with all kinds of software.

software with all kinds of a AI models. The results are okay, but it's not there yet. if you look at, I always use as a good reference.

The way Will Smith eats spaghetti, if you looked at it at him eating spaghetti in AI eight months ago, and now it's incredible what they did.

But for our job, we need huge resolutions.

what's coming out now is an hd, not even a 4K sometimes, and you could upscale it. There's lots of AI upscaling tools, um mm-hmm. Video, but

they're not doing the resolutions that we need. And you still miss that control. if you, I have my 3D artists create a whole world and animate the trees or birds or certain things.

We have complete control over that. And with AI am prompting, you don't have complete control yet, which I'm missing and you can't run it in real time. And it probably on the other side uses so much power just to generate these images that it's. It's not there yet, but I'm sure that in a few years it's gonna be a completely different story.

I remember we, I chat a little bit with you around. this LinkedIn post from like Google, and they were doing this resolution, the movement of particles.

So, can you see it? yeah,

yeah, yeah, definitely.

and there's an interactive component around it, right? people can like move these particles and things like that with your hand. And I feel that's some similar experience you are creating. Unfortunately, the resolution level.

they are not good enough.

Exactly. But I mean, it's a start because if this is, this is completely generated by ai. and that's completely different from what we're we are doing in real time. We use a game engine. We would use a, a particle system in a game engine, and then we would use our hand and a lidar camera.

A LIDAR sensor to then read out our hand and influence those particles

In the game. We would then manipulate those particles. But this is, every frame is probably, I mean, it is rendered with ai. So it's a set of AI generated. frames all played after each other, and the hand influences the AI image, which is a completely different way of, um, of calculating the way these particles move.

for us to use it as in our business. It's not there yet, but I'm sure that it'll come.

let's say if someone from an AI video startup pitched to you today. What would you need to see before you actually use it in a show?

Total control and high enough resolution. But funny enough, an AI startup texted me two days ago that they want to talk.

So maybe

can you share? What are they trying to show you?

they just said like, yeah, we're working on v We are vs. You have a dj, he's playing music and mixing. tracks together.

We are VJs video jockeys. We are mixing video together. so he said I'm creating a VJ software, but there's no visuals there and no video clips that you would mix together.

and there's also no realtime world like yours. That's a realtime 3D world that you can manipulate. But all the videos or all the effects, not videos yet, all the effects that are there in my software. They are, generated in real time by ai. So that's again, the difference. for example, you have doom the game, the very old game back then.

It was also a 3D engine, one of them, one of the earliest. But there's people now that managed to simulate Domin ai. So you're not having a graphics card render out a 3D world, but instead you have each frame being generated. In real time. buy AI depending on the input of your keyboard and your mouse.

So you can play doom, but you can play it in an AI generated environment instead of having a 3D simulated environment.

What do you think? What are your thoughts on that?

I think that will demand a lot of power, a lot of rendering power, because Yeah, you need to. If you look at, if you use mid journey now and you need to generate one frame, it takes a few seconds and probably lots of power in the background to generate that image.

If you wanna play a game, you need 25, 30, 40 images per second for it to run smooth. So the amount of, I'm not sure if the amount of, of power that needs the amount of GPU power. Will be enough locally to ever run that in AI compared to running a 3D simulation.

It also sounds like a complete waste of,resources.

Yeah. But it's a very nice experiment. for the, it is very nice also what you just showed that's running just in AI generated frames. Yeah. it's amazing. It's a very good experiment.

are you gonna meet with the AI company? They texted you. Are you gonna meet with them?

yeah. I'm I'm like going abroad for a week or two now. somewhere in February we're seeing each other.

what did they say to you that motivated you feel like you're gonna meet with them?

We were thinking of, I have a whole team of programmers working on our real time, and developers working on our real time system.

In Unreal Engine. but we're also quite looking at ai, generated imagery because yeah, that's the future and can we integrate that? and we were always so busy working on, on actual products and actual shows That experimenting, sometimes there's no time for it. And it's been very busy with these fear shows that we haven't been experimenting as much as we would like.

And this guy, he just started, he sold his previous company, and now he just. He's a maker and he just started experimenting with it. And if I'm sure that he will, he did a good job. 'cause his previous company was completely crazy what he did.

so I'm very interested in looking what he did, because we didn't have the time yet to experiment.

Mm-hmm.

It's a bit of problem. We started now going into this real time 3D using game technology. But we're kind of stuck in it because we have jobs,shows to make and we need, yep. To make the next step, you need something like COVID, where you need, where a company stops, right?

It comes to a halt and you say, now we need to reinvent ourselves. But we still have quite a lot of work with our game, tech-driven, solutions. we're just seeing the tip of the iceberg there. we just started, and before AI catches up to the level that we need. Yep. It's gonna take some time.

So we're at a good spot right now.

Yeah. Yeah. And, and are you actively watching any AI tools or startups, anything you feel is, close to being useful for your work? Other than Midjourney. I know you're using that already. Oh

yeah. Lots of things. we're looking at nano banana, looking at VO three.

I closely monitor like the big ones. Google, Hey Jen. Looking, been looking into that. been using it as well to change people's voice. Yeah. To trend. I had the CEO, uh, he wanted to come onto the stage as a hologram, but a hologram is a projection on a, if you project on a see-through. Gauze, then you can create a hologram that's technology behind that, but it is just a plain video playing there.

What we did was we recorded him in a green key studio. So the wall behind is completely green. Then we removed the green and then we set to Hagen ah, this. Needs to say these sentences, hello, welcome to,to our presentation needs to say it in eight different languages.

He doesn't speak eight different languages, but because of Hagen we could make him speak eight different languages. he had the same voice, but he spoke different language and his lip sync was very good. So that was a nice use of that.

I think now. this is, I'm talking about a year and a half ago.

I think if I would have to do it now with, and it, it evolved like crazy. it will look a lot better. other things we are using,mid journey, runway, they have a nice set of tools,with video and, I haven't used it myself, but lots of people in my office are constantly playing with, the video.

Generator of, they're a big fan of that because it's not that censored as all The other ones, as soon as you put something political or wanna make something funny, it might say ah, that's against our guidelines. Let's not do it. And Groc is owned by Elon Musk, if I'm not mistaken, and he just says There's no censorship.

You can do whatever you want.

okay, so yeah. So what are they using it for? Is it just for fun or they're actually experimenting? We're

we're experimenting, definitely. I call it old school 69 videos because the screens we have are very seldom the dimensions of a television, I would say the aspect ratio of a television, which is 16 by nine.

Right. But when we do have to make a 16 by nine video, we started experimenting with using a combination of all these. AI video models. One is better at,one is better at futuristic worlds, more like 3D animated looks. One is more like Google VO is super good, in,in real persons and making them talk.

And so we've been using them, to experiment and make some actual. actual, products with that, for example, we don't really do that. We don't make commercials, but I think the people who are making commercials, they have a real problem. also actors. And you don't need actors anymore. if you're shooting a commercial with neutral actors, just a beautiful man, a beautiful woman

AI generate them. Why would you pay? for,for a real actor, while if you go into something like VO O three, you can just do whatever you want and it's consistent. It stays consistent.

Consistency seems to be like, getting really better like last year now. So it used to be a problem.

Now you can have consistency in the people, the nuances in what they're wearing and then the surrounding.

Exactly. There's just some, like you can tell sometimes, if you have a hallway and there's lamps. these lamps might differ. The model of the lamp might differ a little bit from the left side and the right side.

And then you can still see, ah, it's a,

the lighting, The lighting, yeah. I've put stuff on fire in ai and then the way the fire reflects on the people that are there, it's amazing.

Oh, that's crazy. Wow. So what does your team use it for?

like for smaller projects, for like little commercials?

We don't really do commercials, but when we get asked to do a commercial or like a promotional video, which is 69 or goes on Instagram or whatever, we start using parts of it will be AI generated because it's, it becomes possible.

And why would you spend, the time? Yeah. Why would you spend if our client says, yeah, we want to have a da say something stupid, A dancing bunny, that is throwing Easter eggs at the screen. Why would you, if you need to model that Easter Bunny and then animate it, put somebody in a capture motion, capture suit, it's costing a lot of money.

It's cost. It takes up so much effort. While in ai, sometimes you get a very nice result with two hours of prompting.

this is not your core business though, you don't do a lot of those, right?

no. But sometimes the events we do. the event where we have the big screens on the stages and there we create the visuals.

Now they started asking, yeah, it would be nice if the social campaigns reflect a little bit what is happening on the screens at the, going to be at the screens at the festival. And then we create, some movies. but it's seldom, so seldom that we do that. So for us, it's still, to actually use it in production.

I would say 0.5 0.05% of the time because it's not there for us yet. But in the concept phase, yes, most definitely.

And among all the hard things you're doing that AI is not,capable enough. which one is the hardest thing? if you were to pick one thing that AI is like furthest away from?

The high resolution, staying consistent and giving that crisp image quality that we get from a game engine. if you do something in AI and you blow up the resolution and then you have a big screen and you're close to that screen, you'll see that it's AI generated like.

I see it because it's my job. But for example, you go onto Instagram and you see these models popping up.

They have half a million or a million followers and everybody's commenting, wow, you're so beautiful. And then she moves and I just see in the hair the small details that she's ai.

But it's very well done. and if you're not into. the business you don't see, but I see like the hair, very small things like if we would make mountains with lots of small leaves, and we do that in a game engine. Every small leaf stays the same. Small leaf stays in this place when we move the camera.

But if you do that in ai. The consistency of millions of leaves being in one shot, it cannot keep the same consistency. The leaves will every frame differ a little bit and you see that. I see that. Mm-hmm.

I think it's ultimately a problem with AI is not deterministic.

Exactly. Not yet.

You can't control it like very consistently.

Yeah, exactly. Yeah, exactly. or now I experiment, sometimes I take a picture of somebody holding a can with a straw in it, and I say,drink from the can. And it'll just put the straw completely through. it would physically go through its head and then it's standing there like it doesn't know yet.

Well enough, like how this should look like to calculate it correctly. But it will get there. if you see where we were eight months ago and where we are today, it's crazy.

What do you think is the benefit of it actually gets there you know, based on what we're seeing, it could consume, more compute.

if AI is to generate every frame, then it could be very expensive.

The amount of energy that a movie generated in AI uses, its, It's not serious. it is also not sustainable, I think for open ai. we're also using Jet GPT Open ai. SOA is not available yet here, but the results of SOA are also amazing.

But yeah, it's not sustainable. If everybody in the world starts making videos for fun, you're gonna need a few nuclear power plants built next to your company. to have that amount of energy and, yeah.

coincidentally, I just interviewed, someone from Denmark who wrote a book, a professor from, the University of Copenhagen.

He wrote a book about sustainable ai and he was sharing like, the mound of water that the company Open AI used to train. GPT-4 is basically 300,000 human beings, one month's worth of use of order.

and now it's not only open ai, but every, all the big companies are training and investing, and so the amount of power that is going there is, it's too much.

same with, for example, basic hardware,rem random access memory, the price went up three times,times three. something that was 300 bucks is now 900 because all these, AI forms, they need, more, random access memory. So yeah,

that's another problem.

Sowhat's your gut feeling about where this is going, or do you feel like in your pipeline, AI is probably not gonna take a place for all these reasons.

If they keep on, if they, if the big companies keep on giving it away so cheap, it could be interesting for me if the quality gets better.

But there has to be a point when, these big companies start asking the actual money that it costs in energy to generate what you just made. And then it could be very uninteresting. for what we do. Because you might have to. Make 50 iterations before you get the actual result that you need.

And then cost-wise, it's probably more interesting to have a human create this,in 3D software. then again, if you have a commercial that doesn't need to be high resolution and has actors in it, and it does a very clear storyline. And it's very clear what needs to happen then I think it is worth the cost versus, hiring a camera team, renting out a whole restaurant to shoot a commercial.

Yeah,

it's gonna beat that. But for me, we have all these show that we make. It can be a crazy 3D on the water world that we're flying through in very high resolution. But it could also be a show that's completely based on a futuristic world with cars flying around. And it's also specific Yeah, that and in that high resolution that I'm not sure if it's gonna be feasible.

I mean, the amount of energy that's gonna be wasted to create that show compared to putting some humans to work and making that show from their own hands, with their own hands and their own mind.

Mm, that's a good point. Maybe, people working with you guys for the quality requirements, their jobs are safe.

but everybody's afraid. Also in my company, they're like, oh, AI is coming. AI is coming. Yeah. But if you're creative, like we are, we are creative slash technical people. if you use your creativity, embrace AI as a tool and make your product and yourself better with ai, then you have lots of steps ahead from your competition.

I don't read my contracts anymore or whatever. I get in, I agreed, my contracts of course, but I throw them in ai, and AI gives me its opinion. I throw them in chat, GPT chat, GPT. Knows everything about my company and I asked the opinion about certain contracts, and then I checked in with my lawyer.

Same opinions, 90, 95% of the time. I mean,

it's good enough for that kind of things.

Definitely. I don't write my emails with AI because I think you can see that when an email is written with ai, but I'm not a native English speaker, so sometimes I have. Gemini, clean up my emails or give me suggestions like, I think for that it's amazing.

because it's still all just, just language models, that's what it is. Language models and AI generators, it's not,

If you think about text, it's so low resolution, right.

Text, it takes up nothing, compared to an image that needs to be stored.

Text, text is really nothing. But I'm amazed. sometimes

when I ask Chachi PT a difficult question, sometimes it also tells me something that's completely wrong, and then I'm like, and then I tell you're wrong. And then it corrects himself itself, But then the next time I ask the same question, it'll give me the same wrong answer.

Just because it's been trained on a specific data set. And only the next time it'll be trained, it'll maybe take the correct answer. Yeah. So it's all not there yet. All use it with caution, but I think it's gonna, it's gonna have a huge impact on our lives.

I think so too.

so let's say, if I start an AI studio on visual arts or videos.

I approach you to collaborate. I ask the question, let's build something for Primax,What would you ask me to build

Yeah. first we, run, worlds in 3D worlds in real time, and the big difficulty there is. Assets come from several artists.

an artist will make a tree, another artist will make a car. somebody will make a metahuman that is walking around. Somebody will make flowers. But, it really depends on the qualities, game qualities of the 3D artists. if that asset is usable in real time or not. If an artist is not used to create assets for a game engine.

He will, probably use way too many polygons and then the, as the asset becomes too heavy to run in real time, especially if you put them all together. so we've been looking into creating something ourselves now that optimizes it through the use of ai. Now we're doing it without ai, but things like that.

Optimizing 3D models, even generating 3D models that are completely optimized for real-time workflow. That's something that will be very interesting for us. Or a system that takes our whole database of everything we ever created in 3D for in specific themes, say we have, like we created 50 different kinds of trees, 50 different kinds of castles.

For example, we created 50 castles already and we say, we have an AI train, on those castles, and then we say, now generate us. A castle, from the 18th century, with three gates and no windows, then it would use all the data that we created already to build in 3D by itself, a 3D castle. That would be very interesting.

That runs in real time, that doesn't have too many polygons. That could be a very interesting technology for us.

So taking your data, vast amount of data and do the generation real time.

No, not real time. For use. In real time

For use. Use in real time,

yeah. Yeah. Because now I bet there's generators that already generate lots of 3D models, but.

They're probably too heavy to be used in a game engine in real time. They will have too many polygons, too many triangles that, builds up this,this 3D model. And then it's too heavy to use it in a real time environment, then you can only use it to, to render.

I wonder if it's a problem with the model itself.

it's usually the Human who makes the model that is not used to modeling for game tech. yeah. that, that's a thing we need modelers, humans that, know how to create a model that will end up in a computer game and then,

Got it. I have a lot of builders on my channel, so hopefully somebody will reach out to you. to collaborate. So you've been doing real time these visuals for 20 years. Mm-hmm. What do you know about human perception and the attention that, the AI researchers or the models haven't figured out?

Oh, that's also a thing. We create shows and a show should be like a sinus. People shouldn't be dancing the whole time. they should be dancing as much as possible. But if you like, would have a machine create a show, it would probably for now just give energy all the time. And humans can put.

Into a show. They can put a human factor, they can put a nice break, and then it goes up, and then it goes down, and then it goes up. same as a DJ does, a DJ looks at his audience.

And he sees like this track works, they're dancing. but he should have his crowd dancing, but he shouldn't wear them out.

He shouldn't have them dancing at full force the whole time, because then. You won't have a good show. A show needs to be a sinus and for a machine to replicate that, there's no way. you need humans to build a very good show.

And if a show is too perfect, like you have these time coded shows, that means that you know what's coming and everything is pre-programmed,

right?

It looks very nice, but it looks very fake because everything is too perfect. If you have something that is being. Made by people in real time pushing buttons. Then it looks, and it feels much more human, and not fabricated because

at tomorrow end, for example, there's almo during the shows, when the DJs play, nothing is automated.

Everything is pushed by humans, by hand, and you really see that. it looks more human than some of the other shows. I see

the audience want that.

The audience. The audience are humans and they're looking at a show. And if it's too mechanical, too much computer generated, the feeling, then it.

It won't work as well as, as when humans create it. Humans that have a feeling for rhythm have a feeling, for color contrast, balance between video and lights. And that's something that you cannot learn a machine. It's something that only we humans can do. You can have, an AI write a poem, right?

But it'll never be as good as when you take one of the best poets in the world and have him write a poem. Because he will put some humanity in it. And the other one, the other is written by a machine.

What do you think of those performance now? I've seen like the robotics playing, violin, that there's an orchestra of, have the robotics playing those things, it looks like, wow.

Different as well. It's

different, but I prefer having, a real person play a violin over a robot. Playing a violin because why? The robot, was trained in AI to play the violin and that person trained, probably 25 years before the person could play the violin perfectly. I think some things shouldn't be touched.

And my next product is gonna be exactly about that. We are going to bring classical music. But in an immersive experience with a 3D generated world in our game, technology, but we keep the real, musicians played it. Nothing is AI generated. We have real musicians.

We recorded, an orchestra of 120 musicians, and what they bring is being translated onto the screen through gaming technology.

But. It's still humans at the basis and technology enhancing what the humans are doing. Not everything is just generated by machines.

Wow. So the musicians are playing, the real human being playing.

It's just the surroundings and the perceptions and the looks could be different. It could be in a desert that could be on top of a journey.

Exactly. exactly like that. We created worlds with our game tech. say we're playing, a very classical piece about the Alps, then our system will, we created Alps, simulation of Alps and these artists will be playing on the Alps virtually, so it'll look like they're sitting.

On the Alps together with a real weather system or everything, but it'll be played by these real people and not by anything that is AI generated or not real. No. 3D models there. They're the real artists.

Sounds amazing. if your event is coming out I'm happy to share in the video description,

we're aiming for September, re worldwide release.

There is time. Okay. you also mentioned that you're hiring, like what kind of people are you looking for and what's the pitch for someone thinking about joining?

Yeah, we're looking for, people, with 10 or more years of experience in, 3D modelers for games, 3D animators, that are used to game technology and programmers, c plus plus programmers, QT programmers, that are all.

Into the gaming industry, working with, unreal Engine. That's what we're looking for.

And, since we spoke, I hired three new people and I found some very good programmers. Yes. Very nice.

Awesome. Congratulations. thanks. Thank you so much, for the conversation. I really appreciate.

I know it's really late for you. It's 10:00 PM or 11:00 PM 11.

The evenings just started.

Yeah. you are like one of the hardest working founders I've talked with and I think what you do is a craft. it's just not like I, I have full respect for people who are honing on their crafts, like the violinist.

The real human doing that. And you're experimenting with ai, but you're not chasing technology for ai. Sick,

We shouldn't use AI because it's, popular. Right. No. Use it when it makes sense.

Yes. Good point. And you are chasing the moments that make people feel something. And I feel, I wanna go to one of your event and I ping you.

I'm here.

come over to Belgium. we have tomorrow in there. Where are you from?

I'm in San Francisco Bay area.

Yeah. we're probably coming out, uh, later this year to do some shows over there,

I would love to go to one of your events and, I know that again, you're holding a really high visual bar for whoever is building the next generational visual experience, Yeah. And thank you for sharing how you think about what you're doing, your craft and being so generous with your perspective and time today.

Thank you. Thanks for having me.