We Talked to the Producer Behind Natasha Lyonne’s AI Movie

0:00Very few, if any, studios would probably bankroll that at 70 million. But if you made that movie for 10 or 12, you got a shot at it. And you got a shot to like really knock it out of the park. All right. Hello again, everyone to The Noise. This is going to be part two of our interview with Bryn Moser. So this is a two-part series. In the last episode, we started our conversation. Bryn Moser, founder of Hysteria and Moon Valley. In this episode, we're going to get into more detail about Uncanny Valley, the movie they're making directed by Natasha Lyonne. AI is such an emerging field. And in order for you to innovate, you actually have to put a production through it.

0:32and that's exactly what they did as stereo yeah they're gonna learn a lot by putting some actual project through the production process yeah and like we mentioned in part one the enormous cost savings that you can potentially achieve and bring glimpse us some of the glimpses of that yeah so uh enjoy the listen and we'll catch you at the end

0:54well yeah do you want to talk more about i can't even value and sort of how what um the role ai plan? Yeah, we just, one of our partners is Natasha Lyonne, tremendous filmmaker and a futurist, you know, from everything from Russian Doll and beyond, like, she's very interested in how technology shapes our world. I think when she was working on Russian Doll, she got very deep in quantum physics and sciences, and has this crazy network of brilliant people like Jaron Lanier, Ted Chiang, who, you know, are arguably building the kind of the quantum sciences of the world that we're living in are great thinkers.

1:26And she's interested in that. And she also, I think, understood what that AI was going to be inevitable. And I think as a filmmaker and any filmmaker was like, shouldn't we be at the table? Like, if we're not at the table, we're being served. What's that thing? So like, we should we should be there. And we should know what's good about this helpful about it. And what's really bad about it. Because if you remember at the strikes, everything was bad, bad about it. Yeah. And and what that means is, is if everything's bad about it, we end up fighting the wrong fights, right? Then we're mad at Brady Corbett because he like, used re-speecher to make the Hungarian sound better.

2:00But somehow we're like not mad that Meta is scraping all of our movies to make their model. Like we're sort of looking the other way, we're looking the wrong way. And so I think what, you know, Natasha was really interested in as she started to think about what the technology can do is that idea of democratizing those studio films. So Uncanny Valley is this big adventure sci-fi film with these two folks who, two people who go into a video game, right? And so half of it will be shot practically like any film. And then the other half will be in this hybrid AI using actors to drive, etc. And that movie would be like a$70 million movie at the studios.

2:35Wow, okay. And by the way, the movie is going to be incredible. The world should see it. It's going to be a smash. It's a hit. Very few, if any, studios would probably bankroll that at$70 million. It's very hard to build a financial model where that movie, where you're going to put$70 million and then actually make any money on it. Especially in the market we're in today. Totally. Very conservative. But if you made that movie for$10 or$12, you got a shot at it. And you got a shot to really knock it out of the park. Yeah. And you're not, you know, the savings come in really kind of this like long tail pulling out of how long it takes to kind of get this back and forth from production houses or overseas production houses or whatever else versus like a smaller team.

3:17Still, your DP that you love, your art director, your head of VFX, you need those people. Technicians are more important than ever. People who are great at something are more important than ever. You still need your actors, your writers. Don't compromise on that stuff. But just being able to do things quicker in real time makes a big difference in terms of budget. Technology makes a big leap forward when there's, especially in our industry, when there's a film up against it being road tested at the same time. Is any of Uncanny Valley being made in this building? Yeah. And if so, then talk to us a little bit about what you're learning and how you're adjusting on the fly.

3:51Yeah, I would say everything we're learning and putting into the model comes from practical problem solvings that we're doing on films, whether it's Uncanny Valley or beyond. That's why I'm always like, when I thought about this, it was really about like, find the people all over the world who are figuring out these little pieces, bring them together under one roof and be like, okay, guys, the collective problem that we're trying to solve is how does this technology actually build the future of our industry? I used to think about contemporary art a lot too, where you have people like Jeff Koons or Damien Hurst, and they have a big factory with like 50 people and all the money in the world manufacturing their art and they make amazing stuff.

4:28And I was like, what if you could give access to that type of factory to other great artists, like who might be as great, but not have access like Ai Weiwei or something? Like what would they make if they had three months, unlimited resources and an amazing team of manufacturers? And like the world should know that. Like how come some people are limited, are able to do something in a factory like that, are able to make a movie with ILM, but then a lot of other people aren't. Like our industry has very high walls. It's meant to keep a lot of people out. For sure. And it's, you know, not everybody gets to make a Disney movie.

4:58Not everybody gets to work with ILM or something. And the line to make a Disney movie is 10 years long. Yes, exactly. You got to wait your turn. Exactly. And the risks on who's going to make that it's you, you start narrowing and narrowing it down. You're not going to give those things to people who are risky filmmakers, but they should be making the movies. I mean, you know, like I have a five-year-old son, there's not enough animated movies. I would go to a movie every night if I could with him. I can't, I've seen now Minecraft seven times. It's too much. But I think that like, you know, that part I think is where we are always pushing everybody we're working with to be like, give us the problems that you're trying to solve because we want that.

5:33Right. And what I always tell the team here, and we used to think about it as well, too, is I don't want people to ever be like, no, you can't do that. And I learned this from Tony Hsieh, who started Zappos, who told me that he never wanted people to say no. It was only yes if. So you would say, somebody would go, hey, can we turn this into a rotoscope? And then can we extract these into 3D layers? And then can we turn it all into claymation? And people would be like, no, I don't think you can do that. First, I'm always like, no, no, think about it. And then it's always like, yeah, I mean, sure.

6:06Yes. If you had 40 other people in this other room working on it and$5 million and it's like, okay, good. We're not going to do that. But like, at least we know what it is or what it would take. And so the team upstairs, I think that's their special powers that they're always saying yes. And we want these complicated problems. And there's a lot now that we're working on that are very unexpected. Like it's a good questions where some of the best filmmakers in the world are like, can you do this? And we were like, we never would have thought that that was something that people would need. But that's very interesting.

6:36And yeah, let's think about it. Has this, has the process with Uncanny Valley changed the structure of like the filmmaking process? Like, you know, traditionally it's very like rigid, you know, you got pre-pro, you know, production, and then you're in post. With AI and everything blending, have you found, is it blending the stages, condensing some of the stages together, being more of like an iterative process? Yes, I think 100 % where we're seeing a lot of excitement from great directors who are working on big tentpole films is that they're saying, hey, I want to be closer to this creative. And I think that this tool could, I think it's interesting.

7:13I want to learn more about it. And they come in here. It's like a teaching hospital. We're bringing a lot of people. And there's a lot of filmmakers who get excited about the idea that they could sit at a table with their head of VFX, their DP, their, you know, whatever it is, their cinematographer, whatever it is, a small team, and be able to just do a lot with sitting and brainstorming and getting materials out there so that they're not waiting in those times to do it. So I think that that part compresses a lot, this part where you have people actually sitting in a room for a couple days and brainstorming and the output of that, it's not for the film, but it might be for the financiers or for the studios, like you can get very, very far.

7:46right but that starts with you know original uh scripts original art artists who are doing it like it it's not like you're coming in and then you're like okay let's i don't know do bluey but they're cats or whatever like it's really like this is a movie and by the way this story happens all the time this is a movie that uh was at warner brothers for eight years and then they just killed it and like they i get reverted back to me and like i've been this is my dream movie and nobody's gonna make it for 80 million dollars but like could we try to make it for 12 like is there a world where we make it for seven like yeah that's a story time and time again that we're hearing and people are coming in here and that's a movie that i would argue is not going to get made without trying to figure out how technology can do that so that's like that's not taking a job away or taking a movie away that is like a net new movie that's not going to get made without this thing yeah i agree with you a lot of the costs in feature film production is the iterations the reviews the meetings like the stuff that you don't think about yeah at a very crude guesstimation would you say that current technology with AI today is one-tenth, one-seventh, one-eighth of what traditional production is?

8:50I think that it's funny. I'll tell you what other people think. Katzenberg famously said 90 % reduction for animation. Then Cameron said 50 % reduction. Then Ted Sarandos, who I love the most, said he hopes AI can make our movies 10 % better. Big difference, right? If you think about that. One person is saying, you don't need 90 % of the people. The other person's like, we got to get rid of half the people. And then the other one's like, no, no, we just want our movie to be better. Like who loves filmmakers more? Like that's the part that I think is so, is so interesting about it. And it's so true.

9:20I think there's an easy 20 to 30 % time and cost savings across the board. I think if you look at things where you have to think about where the models are very powerful, right? So animation is like every frame is running through a potentially a model, right? There's, there's rendering in every, there's those costs associated with every frame versus if you think about a movie that has maybe 20 % of the movie has special effects or something, that's a 20 % thing. There's, there's not, you're not getting like a, all the way down to zero. Again, nobody's trying to do that. Maybe some people are trying to do that, but I don't think anybody's going to watch those movies in the same way.

9:53Yeah. So I think there's like a big reduction right now. That's like 20 to 30, maybe 40. That's undeniable right away right now. Um, and I, and, and, and I would say that like the way that we see the world, And I think this is going to happen is that in the very near future and to, and to beyond every film and TV project will have a custom model and that's what we're building. So we build a custom model. It's fine tuned on all the materials with the artists. If it's done with a studio, the studio owns that, that data doesn't go into our foundational model. They own that, that model. That model is a, a product that they get at the end of the film.

10:27and if you're doing that on a tentpole film and you have a model that's totally fine-tuned and can generate things on model then that becomes a tool for marketing for video games like it's quite powerful thing but i think even indie films will have a custom even our documentaries now have a custom model and that could be the way to think about it is like it's like a smart series bible meets chat gbt or like a smart google drive that you can just start to like have as your partner you're making it it's like an extension of the intellectual property that has the model yeah and it's a, it's an, like an agentic partner for you as you're making it.

10:59So that's what we're building. That becomes that sort of artist interface. But I do believe that whether it's us or somebody else, that's the way that it goes is that you have a model. So when you're starting an indie film, like right now you have an indie film, let's say we go make a movie tomorrow. We're like, we start an LLC, we open a bank account. That's how you do it traditionally. And then we're in production. Now it's like also open a model, start a model. Um, and, uh, that's, that's i think where it's gonna head and the idea of the model like a chatbot you could talk to about give us marketing materials that are in the voice is it for like generate pickup shots that look are in the style of what the movie looks like what does that i mean i think that a lot of people are already using ai in this way as a partner for them if they're a writer they're using it for research if they're uh you know an artist they might be generating images on mid-journey that helps them on the storyboard like we're seeing now how ai in the way that you can use it right becomes an extension of what you're doing, right?

11:54It becomes like a helper for you to make this stuff, a little producer that you're with. And so I think that that just becomes kind of like the bigger version of that. But what we're talking about that's different is that the models are really fine-tuned. We call it like a Laura model, but it's fine-tuned around exactly what you're making. So say we're gonna make a space movie is our movie. Like that model is not just a random chat GBT. It's like, it's the space movie. It's got all, it's got the script. It's got the characters that we've been designing that are original. And so then it starts to know that world.

12:25It knows what cameras we're shooting on. It knows the locations. And then it can start to generate things. So it might generate, that's helpful for us, plates, background plates. It might generate elements within it. It might generate laser guns or whatever it is that we're doing. That becomes your helper. But it's ultra specific. Now, you wouldn't take that model and move it over to your kid's movie that you're doing because it doesn't know that world. And I think that's like the massive difference that like the AI companies were missing. They were trying to build these models, which is why they were scraping everything, pirating everything from all the studios is because they were like, we need all the movies so that you can say, make me an MGM film, a comedy and whatever else.

13:06And it would make you that, you know, as we were building the model, we were like, we don't need the model to know how to make a movie. We make a movie. We actually need the model to just know how the world works. and then we're going to build something on top of it that's going to be our partner to help make the movie sure because now nobody's time nobody wants like uh yeah it's an mgm musical it's a universal this it's like that's not yeah it's not what we want we're not making stock videos we're making a very specific yeah yeah yeah and i i think there's lots of room where uh the copyright and other things don't matter like i think people who are still making tiktoks or you know stuff on youtube and wherever else which is a big market like then you know they might not care.

13:43They just want whatever the latest model is that can do the thing that they're trying to do the best. But our industry cares. Yes. And real filmmakers care and studios care in a real way. And so I think that you have to think about how do you build solutions for the industry that you're working in. And it's not just one size fits all. Have you had to rethink talent deals with these projects as far as like having rights to replicate their voice with an AI model or replicate their likeness in a model and then like restricted to just that project or like restructuring deals in that way? I think that everybody is figuring out what these deals look like going forward in terms of AI, both in terms of training and in terms of likeness, right?

14:22I think that this comes down to in the next year, consent. And everybody starts talking more about consent. Did you have my consent to train the model on my stuff? Did you have my consent to do this voice that sounds exactly like mine or this likeness? And so I think that consent starts to become a really clear way of thinking about how we build parameters around AI, but that's going to go into all the deals too. I would love to be able to find a way to build AI solutions. It's not what we're doing, but somebody should do it to help speed up the process of how long deals take in Hollywood. Because I actually think that like deals take so long.

15:01Yeah. Why do you think that is? Because there's a lot of people and there's a lot of money and there's a lot of people sitting around who are all participating in those deals. You've got an entire industry that's built up by people who are helping put those deals together and it's slow. And the challenge is you've got creative people who are really excited to lean in right now who are like, hey, I got this movie, I want to make it. And I'm a big director and I've made billion dollar movies before, but I've had none of the upside. And like my next movie I want to do here. And then you have financial people who are like from outside Hollywood for the first time, VCs and stuff leaning in thinking about Hollywood in this way.

15:37We're like, yeah, I'll fund that. If you can make that movie that you was going to cost 120, you can make it for 20. I'll fund it, you know? And if when I want a part of the toys or whatever afterwards, but then you have an infrastructure inside that's still slow. Yeah. Like that takes a lot of the energy away. And so if we can figure out how to speed up that, I have no idea how to do it. Or do you think we could just get more kind of permissionless filmmaking where it's like, Because if you could use the tools and you could do it for less of a cost, you can just do it yourself and you don't need so much permission for it to happen.

16:06I do think that there's a moment where the power dynamic shifts. And if you've got a director who wants to make something, you've got somebody who's got the money to go do it. It's like you don't need all those layers potentially to do it. I just got back from Senegal. There's a kid. He's younger than me. He's like 37 named Hussein Dembelso. And he's one guy in Senegal who started going viral because he was doing these things on using AI on YouTube or whatever. I think he's the best filmmaker in the world who's cracking AI. Really? It's extraordinary. His first one was about Senegalese soldiers in the Second World War.

16:43And somebody sent it to me like, did you see this? And, you know, it had a million views already. And I was like, oh, my God, who's this kid? I went home because I was in the Peace Corps in Gambia. So I was so excited about Senegal. I went home, opened my LinkedIn, and he had messaged me. I was going to message him. He was like, hello, I love Asteria. Like, I just made this movie. I'd love for you to see it. I was like, see it. I'm like, I love it. And so I went to see him. Okay. And we worked on a movie together that he's directing. And I, you know, it was incredibly inspiring. Like his stuff, follow him on Instagram, follow him on LinkedIn.

17:12His stuff looks incredible. And he's not like, yeah, one day I'm going to come to Hollywood and make a movie for Disney. He's like, no, I'm staying in Senegal. And we're going to tell these stories. And we don't need French people come down to write the scripts. We don't need a studio to come in. I'm going to go make them. and his stuff looks better than most other people who are already making stuff. Because he has, by the way, better internet in the car than we do in this building. He's got the right computer. He has access to all the tools. The playing field is level. It's not even like the Canon 5D days, right?

17:42Where it was like the Canon 5D was still 2 ,500 bucks. Like all of a sudden now it's like, you got a laptop, you can run these things. Like pretty exciting. It runs on the cloud, just 20 bucks a month. Yes. So then it's like, it becomes about who has something interesting to say. And that's very exciting because that's a lot of people in the world who are going to have to start selling your stories. That's also terrifying to the old system. Yeah. Speaking of the old system, I would love to hear your predictions on the next five years. Like how does our town, our industry change? People often ask, and I'm sure they do for you guys too, like, what's going to be the Toy Story moment where a movie comes out that's AI generated that everybody goes, oh, it's undeniable, right?

18:28Like people always talk about this Toy Story moment. Or, you know, remember ILM, Steve Williams made the walk cycle of the T-Rex using a computer and it changed everything. So what's that moment that breaks through? and I don't think that it's a creative thing. I think AI just sort of starts to come into everything. So it's not going to be one movie is like, oh my God, AI has arrived. But I think it's going to be a business moment. It's going to be a moment where somebody makes a movie independently like flow, but it's like the size of Lilo and Stitch. And all of a sudden it's like this small team made this movie for$4 million and it made$400 million at the box office.

19:11And they own the whole thing. Yes, that's what's coming. And that's coming soon. And when that comes, that's like, oh, shit. For a lot of people, it's an oh, shit, like, let's go for independent filmmakers. It's an oh, shit for the studios when it happens. But I'm very excited for that. So I think that like, that's the moment where people go, okay, this is a game changer of what it can do. And Hollywood's always been sort of adapting to new technologies. How do you see the existing studio system adapting to your Senegalese talent and some of the rock stars that are out there already? Hollywood has witnessed a lot of times of change like this before.

19:52So if you're like the big studio, you weathered through, if you're the board or you're the longtime CEO, you've weathered through what happens when animation goes to computers or what happens when VFX or what happens when editing gets off of Avid. You know, like you've gone through these transformations before where your actual fundamental structure of the team changes. So the studios know what's coming. They have varying, I would say, degrees of understanding the magnitude of it, but they know that it's going to change how things are made. And I think that it's going to be a challenge to keep up and go as quick as you need to, to be able to adapt.

20:29They're all going to have to figure it out. Yeah. So I think that some will be more successful than others at doing it. And I also think that there's part of the, back to the old Hollywood thing, I think one of the things that's interesting that I notice a lot is a kind of like resistance, this blanket resistance to AI. And I always try to understand who that person is who's resisting it. And I think that the industry overall is like not working for 99 % of the people, but it is working for a top 1%. I find a lot of resistance with that top 1%. We're like, oh, I don't like AI because it's going to, job losses of these people that work for me are so great.

21:07And I'm like, I don't know if you really like fighting for the, you know, every man like that. I think you're just like, we've built the wrong thing for this moment that's coming. And how much resistance can you do? And it's futile in the end. So I think that like it's a moment where for those who are leaning in, there's tremendous opportunities. And I would encourage everybody to not lean in if you don't want to, but at least be curious enough to know what could be helpful to the stuff that I do. For sure. And what is going to be harmful to what I do. You can't look away either way. You can't look away.

21:42Yeah. And we're about to launch these free workshops and sessions for the guilds upstairs over some weekends in the fall where we're just going to go and show all the tools and what people do. I mean, I think it's part of it is just sort of like an education piece because there are a lot of things that are like going to be big changes that come. But to the people who are kind of like curious about it and leaning in, there's tremendous opportunities. People can start new companies, can leave wherever they are and like build something right now for the first time. Like it's very exciting. You know, you guys know like people in VFX, like it's hard to know Nuke and Houdini.

22:19It's hard to know these tools. It's like, it's a long process to be able to get to be at a place that's like one of the best VFX shops and you have to know these tools and you're then in this thing. But like, there's a lot of people at those places who are really creative. And if all of a sudden they could actually like do what they do so great in their little field, but do it on stop motion or whatever else or claymation. Like, it's like, it's a big unlock. The thing that amazes me is the VFX artists that we have today, they're already so good at identifying quality, making the frame better. All these intrinsic qualities they have, they can just transfer that to a new tool set.

22:57Yeah, I think so. I think people who are creators like that, it's incredibly exciting. Like we just did some work with this incredible stop motion artist named Kike in Puerto Rico, who was working on the new Bad Bunny music video. It has a little frog, stop motion frog. And as we were meeting with Kike and starting to show the technology we're building, we were like nervous because Kike is a guy. He's a master. He's a stop motion master, you know, and like we're coming in to show him something that like you feel like is the opposite of it. And he was like, yeah, I'll check it out, whatever. But then when he got in there, I think really quickly he saw that there was a lot of things that this thing could just help him do more of.

23:36We weren't like, don't make the models anymore. Like, actually, we still need him to make those models. So if you look at that video, like, he's still making the heroes. Like, the four hero frogs in there are done traditionally. But then we're training a model so that it understands the look of that world and what those frogs are. So now we're able to create all these other, like, he's at a nightclub. So it's like all these people in the back, all these frogs in the background. and it's just like it creates a richer world that he would not have been able to do in the timeline and the budget to be able to do it.