Tech that Finds the Shot You Forgot with Lucas Igel

Cameron (00:54.207)

Welcome to On Production, the podcast where we explore the art and business of film and television production from every angle. I'm your host, Cameron Woodward, and today's guest is Lucas Eigel, CEO and founder of Kino AI, a startup that uses artificial intelligence to tackle some of the most time-consuming parts of post-production. Lucas got his start at MIT where he studied computer science and simultaneously nurtured a passion for film.

‍

He actually made a documentary called MIT Regressions. I've seen it, it's awesome, which digs into the university's history and culture. And in the process, he discovered firsthand how messy and disorganized large-scale editing workflows can be. That revelation led him to co-found Kino AI, pardon me. That revelation led him to co-found Kino AI, which automatically logs and organizes your raw footage so editors can find the perfect shot

‍

or SoundBite with a quick search instead of manually scrubbing through hours of material. The goal is to give production teams more time for creative decision-making and less time to lost grunt work. In today's conversation, I'll ask Lucas about his journey from a tech forward upbringing and NASA internships to the indie film world and how Kino AI has shaped the hurdles of making MIT regressions, had he had the tool when he was building his film and how AI can free up creative energy in the industry.

‍

that historically resists rapid change. Lucas, welcome to the show. I really appreciate it. I have a bunch of questions for you because all of our customers, all filmmakers are so curious about AI and it's awesome to get your perspective on this.

‍

Lucas Igel (02:35.82)

Thank you so much for having me here.

‍

Cameron (02:37.833)

So Lucas, can you walk me through a day in the life of an hour of footage entering Kino? What exactly does the AI do from the moment a video file is added and which parts of the process were the hardest to engineer?

‍

Lucas Igel (02:53.398)

Yeah, throughout the entire lifecycle of Kino's history as a product, we've always aimed to try to index it in every single way that is useful to filmmakers. Transcribing it, describing footage that comes in, detecting who is actually in it, and then all sorts of sub-details within those three categories are things that we aim to do in a way so that not only can you retrieve absolutely everything that comes in, but that you can view human-readable logs.

‍

of all of the footage that you have. It's almost the sort of necessary initial steps that need to take place. And when I look back at working on and co-creating that documentary that you mentioned, I was looking through this giant Google sheet we had called Video Index. And I was scrolling through the 80 plus years of archival footage that we had actually sifted through back in the day. And I don't think we were looking at that spreadsheet that we had made back then when looking at the more recent versions of Kino, but it feels very similar, right?

‍

know, breakdowns of interesting moments that took place, in-depth logs of what this footage actually is, all these things that meticulously had to happen before, you know, we can make a coherent story when making that documentary that you mentioned earlier.

‍

Cameron (04:08.467)

That's awesome. I'm curious, in building Kino's search function, what did you learn about how editors really search versus how computers search? Are there any surprises about the phrases editors typed or the results that they expected?

‍

Lucas Igel (04:26.222)

I think it's a good question. There are a few different parts to it. One is we notice that people want to search for things that they know for a fact existed because they were behind the camera when it happened or when they pulled that footage from some library, they did it because of that crucial moment. And it's a lot of things that you would expect. There's a particular shot of a sunset that you absolutely care about. There's someone sitting on a beach while talking to another person. I think what's been a bit more surprising is how open-ended and broad things can tend to be, right? Where once people realize that

‍

It has the open-endedness of a Google, but it's across all of their own footage. But it's far more powerful than perhaps Google Photos, where you're very much constrained to a few set of keywords that you can search by. You get prompts that almost start resembling something that you would type into Mid Journey, soaring sunsets, that sort of thing. It's been cool to see. It completely depends on the sorts of things that people are trying to produce. But those are the two types, recalling things that you know happened and then

‍

completely just seeing what it is that Kino can produce, given that you have a pretty large corpus of things to search through.

‍

Cameron (05:35.839)

That's awesome. Now, if I understand it correctly, Kino runs locally, processing terabytes of footage on standard hardware. How did your experiences at NASA, SpaceX, where performance is mission critical, shape the way you engineered Kino for speed and reliability?

‍

Lucas Igel (05:55.106)

Good question. I think working at SpaceX was fun because they gave you intentionally, I would say, like a bad developer experience, intentionally bad, or I would say like intentionally secure. It was like an on-prem instance of Microsoft Teams that we used. We couldn't use beautiful AWS hosted Slack, right? It was like a really old version of Linux that I was not used to, or Ubuntu that I was not used to using. In order to watch cool videos of recent launches,

‍

or streams of the satellites, because I was at Starlink, it was just a giant NFS drive that every single SpaceX office had connected access to. It was very 90s. It was very web one. And it was cool. It felt like because of these incredible hardware constraints and these incredible security restraints, in that case, like national security constraints, you had to use not only battle-tested technology, but very much anti-cloud technology.

‍

I think throughout the entire development history of Kino we've made, in trying to bridge this gap and trying to work with Hollywood, is very, I would say has resisted the cloud more than like other industries that like, I think some peers have chosen to work with. Really trying to figure out which amongst these things is good for the industry is good for productivity is good for making sure that privacy is kept intact was a massive portion of the job.

‍

There's another element, though, that I think can resemble a bit of a cargo cult. Embedded software is very important. When I was interning over at JPL, we working on the Perseverance Rover. And I was on one of the mini teams that produces a bunch of code that can run on the Rover itself, just in raw C++. And if it's good enough, then about two years from now, at least two years from then, it will be sent over, then tests will be run. So this requirement is to do native

‍

raw embedded C++ or Rust or other things. This requirement to do all of it using only like the most battle-tested libraries, I think is not what a startup should be doing. However, there's an ethos to it that is so important to preserve, especially in Hollywood and in other places where there's such a emphasis on on-prem, such a huge emphasis on security. So as you can see, this is something that I think we've been wrestling with for a while. How do you deliver the new to people?

‍

Lucas Igel (08:19.695)

while also preserving all these other things.

‍

Cameron (08:23.091)

That's awesome. here's another interesting technical question for you, which is what types of footage or scenarios today stump Kino AI? Are there edge cases like avant-garde footage or noisy audio where it struggles and how do you plan to improve or flag those?

‍

Lucas Igel (08:41.432)

think music videos slash really, really repetitive footage is kind of the final, it's almost the final boss, right? Where if I search for an extremely complex dance move, which thanks to the models that Kino uses, and because Kino has an understanding of video, not just like individual frames in the footage, you can search like some complex move that someone made. But you're going to find tons of duplicates.

‍

And some of those duplicates will be repeated attempts within that exact same footage, or it'll be across. And so I guess this isn't really stumping Kino as a search engine, but it is stumping Kino's UI. And so it's why we have put a lot of work into making sure that you can actually understand across all the different film strips that you have access to where this thing actually took place and how to, I would say, make sense of it and actually

‍

get it in the timeline as fast as humanly possible. I think that's where Kino went beyond simply kind of spotlight on your Mac, hit command and spacebar at the same time, you can summon files really quickly. You need an entirely new interface in order to do something like video, where motion is very important. You have many highly similar acts of motion scattered everywhere. And it's very important which day of shooting this was, it's very important which source this was. This is where we start entering

‍

difficult territory that we spent a lot of time more on the design side.

‍

Cameron (10:09.375)

So Lucas, I mentioned it in the introduction that you actually have produced a documentary yourself. It's called MIT Regressions. And it incorporates lots and lots of archival footage, lots of images that have degraded over time. And I was doing research for our interview today. And you had mentioned that in the process, you actually had built a pipeline of Jupyter notebooks to upscale archival footage for the film.

‍

Do you plan to incorporate any of that sort of image enhancement tech into Kino's core features or is that a totally separate domain?

‍

Lucas Igel (10:44.84)

I love it. I think that it caused so many issues. were huge adopters of Topaz, which is this very cool startup that's been around for a minute based over in Texas. And what they do is they have been working for a long time since the era of GANs to make massive, to create like super resolution versions of your footage. With the age of all these kind of this huge wave of AI that's been happening in the past few years, their models have gotten even better.

‍

If we were to run Kino, or if we were to run our documentary corpus through all that stuff again, it would look even better. But it created so many duplicates, and it created so many different variants that were kind of me moving a slider over by one would completely botch our attempt to upscale some Apollo 11 footage that we had access to. That was on raw XD cam that NASA had sent over to a bunch of universities back like 10 years ago. So you're on this precipice of being

‍

making something feel almost more historically accurate because you took most of the case, like really crappy digitizations of really important footage. You're constantly on the verge of botching it and making something that is literally like false, like where it's, it guesses people's faces and like kind of assigns faces to people that just don't exist. Right. That was kind of like a fun balancing act that we had to play the entire time and required these really high stakes runs where overnight on my own computer that most of regressions was rendering on.

‍

When trying to run a bunch of NASA space shuttle footage or trying to run a bunch of footage of some of bigger protests during the 60s over at MIT, making sure that the inference run the night before went correctly, that there was enough space for it, that it was actually organized correctly was so much of the process. And the more time that that stuff takes, the less time you have to make sure that it's a good movie.

‍

And that it's good, right? And that the ordering of clips in your timeline leads to something that is enjoyable to watch. All of those problems I just mentioned, very much want to be a part of. Managing duplicates is a good example of that. Being able to call really good models like Topaz and others is exciting to us. Although we don't have a relationship with Topaz and many of these other places, things that start resembling full-on modification to your footage is something that we want to step into once we're ready for it.

‍

Lucas Igel (13:09.794)

Having a good foundation is important to us as well.

‍

Cameron (13:13.759)

How do you measure success for Kino AI on a technical level? Are there metrics or stories that to you confirm it's cutting significant time and complexity from post-production?

‍

Lucas Igel (13:25.23)

I think anecdotes from people who were in those early design partnerships with us. When we go on trips to Hollywood, we would knock on people's doors over in Burbank and others. And some people would let us in, and then we would show the people at the front desk what it is that we're working on. In the weeks and months after those initial demos, and then receiving word from them over email or by Zoom call that, oh yeah, this thing and this media that is now released to the public, that shot was because of Kino. That was cool to see.

‍

Very anecdotal. as you can imagine, are ways that you can, know, lots of software, right, is based on success and based on like, did the user actually accept the changes that my software requested, especially as a tool. That was something that more so like just put a lot of wind in ourselves and made us realize that this, this wedge of search is good. And this wedge that is search is a good way to expand outward into something that is both venture scale and

‍

something that has a very wide potential surface area. I think what we really do want to do, and I think what always is the North Star, is can you reduce the kind of mental barrier and just the very real barrier to trying out new ideas by having a tool that is just so much faster than otherwise? You'd be much less likely to look up some random fact about a past president if you had to go to a bookshelf.

‍

before the age of Google plus Wikipedia, right? By reducing all this friction, can you kind of increase the supply of art and increase the supply of media is what lot of what I think about is important here.

‍

Cameron (15:07.143)

I think we're gonna see an acceleration across everything in every industry, which is really exciting. You had mentioned sort of the studio systems approach to technology, cautious, there's even historical reasons for this. You look at the hack of Sony, really hard security and workflows that are tried and true are very important to the industry. And then on top of that, Hollywood is famously relationship driven and traditional. And here...

‍

You are, and so am I as a technology entrepreneur in this space. We like to, you know, the old adages, move fast and break things. Which side was harder for you to convince to either build Kino alongside you in Silicon Valley or use Kino in the industry, engineers or filmmakers? And how did you or are you bridging the culture gap?

‍

Lucas Igel (15:56.206)

Well, will say, I think there's this idea that engineers and science people and math and kind of like STEM people are not fans of art and therefore it's like these things don't mix. But I think the recruiting side has been the easiest part of this company of recruiting like like-minded engineers to work with me on this thing. I think what has been far harder is making going truly from zero to one with any individual customer where

‍

Not only the engineering and technical requirements are high, but the security, specifically the security requirements are very high. I think the real why so much of, I guess, like this cloud boom of the past 10, 15 years has been resisted amongst video production is because video is so difficult. Not only is video very sensitive, like if you watch an unreleased piece of footage, it's a very potent form of media, right? It's very, but it's just so dense. It is so much.

‍

heavier than text. a massive, massive, massive portion of all internet traffic is video. Even though the amount of time that people spend on the internet is probably not proportional to it. It's just that videos just create so many engineering challenges in itself. So we have many, every single person loves the idea of on-prem, who we talk to because of how heavy video is. And then many people view it as a complete requirement just because of the privacy concerns that they have.

‍

So they're both technical requirements and privacy requirements that make this a difficult area to do engineering in in the first place. And that is where so much of our work comes from and so much of our thinking is rooted. But I think it takes me back to having worked in workplaces that are almost pre-cloud, kind of web 1.0, right? And making sure that we can selectively do this correctly, right? Making sure that we can still take advantage of just how unbelievably easy it is to protect

‍

create software now if you are starting a fresh company on brand new infrastructure, much of which is just built on top of AWS, right? Many of the different Y Combinator companies that we were in the same batch as have created so many amazing tools that are based on like cloud first development. How can we take a lot of those things and still be able to move fast in a way that allows us to be competitive has been a lot of what we think about.

‍

Cameron (18:18.847)

That's great. Now thinking about the creative impact of Kino, do you think that using Kino changes how editors and directors tell stories? Have you seen users discover new story arcs or found footage they otherwise wouldn't have noticed?

‍

Lucas Igel (18:34.54)

Definitely. We've seen a lot of people discover footage that they wouldn't have otherwise, or things that they knew existed but they didn't really bother to look into otherwise. Friends working on documentaries who were using past versions of Kino that were available to the general public, Or people in some of these larger studios now using Kino to browse through and make sure that they have access to absolutely everything. I think there's a peace of mind that comes from it that allows you to...

‍

move a lot faster, allows you to be a bit more daring.

‍

Cameron (19:06.921)

How do you ensure that in speeding up these post-production workflows that you don't also speed past serendipitous discoveries? Is there a risk that hyper-efficiency might reduce the slow-thinking editors sometimes need?

‍

Lucas Igel (19:22.062)

I think during earlier conversations, editors would say, listen, I'm not going to not watch all my footage. And our goal is not really to mess with that at all. I think our goal is to be able to help people who already have gone through that watching process. We create text. We create logs, very high quality, human readable logs of every single detail that took place. Every topic that Cameron brought up, as well as the way in which he was walking, perhaps if it was like a video.

‍

that was recorded on the street or something. We are trying to not only jog your memory, if you are already deeply familiar with the footage, but also kind of expand who is allowed to do it. Editors are not the only people who benefit from Kino. It's also producers. It's also anyone whose job involves that footage in some capacity. That is where things have opened up. We are on...

‍

a variety of projects where it is the editor's responsibility to very meticulously watch every single piece of footage that gets recorded. But it is not the responsibility of all of that editor's colleagues. And making sure that they can also have full context has been really powerful for us so far.

‍

Cameron (20:34.559)

So, you know, I just recently recorded an episode with a gentleman who runs a post shop in Los Angeles, indie films exclusively. And in his work, he oftentimes hires musicians because he loves to have a strong beat and undercurrent in all of the films that they're making. And something that's interesting is like, he's not worried about AI at all because he has seen the post-production process constantly.

‍

get better with tools, but there's also another school of thought that like, hey, AI tools are going to homogenize style, surfacing, maybe the technically best, so that it makes everyone's final cut look the same. Do you have a view on how you keep Kino from narrowing creative choices?

‍

Lucas Igel (21:23.726)

I think about it a lot. think the current frontier of a lot of these language models is certainly in coding and somewhat in writing in general. No, it's text. We're trying to do video. It feels like we're in a bit of a different world where the requirements are completely different. So it's very useful to see the absolute edge of things. And I was trying to write some multi-page document the other day. And Claude was just out of curiosity having Claude generate the next paragraph.

‍

It awful. It was mid. It was the average of all writing that has ever happened in this topic, right? Yeah, yeah, yeah. But I asked it to criticize what I had written and it was very good. Whether that is what that says about my judgment, given that I said that its writing ability was quite slop is, I'll let you decide. But it was certainly competitive with what a lot of had also texted, criticism that a lot of friends had also texted me at that exact same time.

‍

Cameron (21:57.385)

Slop.

‍

Lucas Igel (22:20.77)

I think using these things more as discriminators than generators is really one way that you can massively, I say, increase your abilities. That is a huge thing that I've noticed.

‍

Cameron (22:34.307)

that's sort of a wonderful philosophical insight there. Because if you actually look at something that's interesting and even the wisdom of crowds, so to speak, it's always very interesting that the large corpus of known knowledge is quite good at criticizing, but then deciding the next best steps to make in a truly chaotic present and future is much more difficult. And so to necessitate truly novel conjecture is hard work.

‍

Cameron (23:04.219)

is what drives innovation though. So like that's a really cool, interesting way for you to frame it.

‍

Lucas Igel (23:11.182)

I appreciate it. I'm reminded of the ending of the children's movie Ratatouille where the critic, he has a meal that's so good that he starts to wonder what his role as a discriminator really is. But you end the movie still being like, not gonna, I should go easy on creators, but also I have a unique ability to criticize this Pixar movie that I just watched.

‍

Cameron (23:32.531)

Or even there's a, there's a famous like line from Ira Glass of this American life that like the role of a creator as they're starting is that their taste is very good, but their ability to execute the craft is not yet up to the level of their taste. And so you just have to keep pushing, keep refining, keep, you know, keep poking on, what you've built to see if you can make it better. So that's super cool to think about like applying some of those sort of nuggets of the creative process to like a large AI system.

‍

Lucas Igel (24:01.528)

Well, it's changed a lot of how I think the fact that things like attention and context are now like quantifiable things. And we have benchmarks showing that like these models ability to these models ability have very quantitatively increased in in recent years has changed the way I think of writing and the way I think of drawing things that I didn't really think computers were able to really make a dent in. Right. I thought it was more about like I think when I was in the early earlier years of college.

‍

The previous AI waves were all about how AI could tell you if your movie is going to excite audiences enough. And there's this much more daring, but also much more disturbing era where I think the hype people are now saying, no, no, no, the AI is just going to generate everything. Every single pixel is going to be created by the machines. it's not about, there's no pretense of, I would say, this kind of human machine interaction that's going to go on.

‍

Obviously many people, especially in Hollywood, very distinct opinions about this, but it feels different. I understand why people feel very different about this than they did five years ago.

‍

Cameron (25:11.325)

Yeah, I mean, every dinner table in every industry, people are having this conversation as to the impacts of these technologies in their roles and in their work. You know, on that point, Lucas, from like ethical societal concerns on these types of things, as AI takes on more and more of the creative process, do you see a need of any ethical guidelines or guardrails or how do you think?

‍

the industry should be thinking about this stuff. Should filmmakers disclose AI assisted edits or is it simply another tool?

‍

Lucas Igel (25:43.714)

You know, in software engineering, in like tech Twitter and just amongst friends who are also building software companies, I think people actually are very honest about whether the code they're showing me was made by machine, or like 95 % of it was. And I, because I think people are less precious about software in particular, I think. So I think that is a field that is very distinct from filmmaking where you can kind of see how a very different culture treats these things.

‍

And it's allowed me to see what is healthy about the way they're doing it, what is unhealthy about it. I think it is... I think these things should be disclosed, insofar as it makes sense. I also think that it's gonna become so normalized so quickly. And the goal with Kino is certainly to make it feel like a natural next step, where it would be as if there was some sort of like moral panic where...

‍

animators at Pixar had to disclose that they were using some distinct rendering method inside of RenderMan, right? You know, people were getting yelled at online because, you know, they're using some new realistic lighting technique that they didn't disclose before. It should be, I think, the rate at which things are going and the rate at which all these tools are shifting is going to make a lot of the controversies that have happened in recent years and months seem a bit silly. I do understand people's reservations. I think there's that big movie

‍

the Brutalists, were using voice techniques to augment someone's accent to make it sound more Hungarian. I don't know what to say. I don't know what's When it comes to human performance, I do enjoy the feeling of reality and things being real. However, I think that when the writer's strike took place in 2023, AI was sort of on the menu. A lot of it was about the previous decade of innovation. It was about streaming.

‍

The way that we as writers and later the way as we as actors get paid was based on the era of, you know, blockbuster and like movies and shows being distributed the normal way. And we have been, you know, the way that we're getting compensated now, it doesn't make sense. And everyone who was leading those strikes said like within a year of those strikes beginning, the question was also about the subsequent decade of progress that is now happening, the AI question. It didn't feel like they're really talking about

‍

Lucas Igel (28:06.824)

novel tools for thought, novel tools for filmmaking that are kind of extensions of the human hand and extensions of the editor's eye. They were very concerned about, you know, I don't want some studio to scan my face and now they own it indefinitely, right? There's some kind of funny examples of this happening from the past 10 years where actors that have been dead for 50 years get resurrected like a multi hundred million dollar movie. And I understand it. I think that the debate is going to get significantly more blurry, though.

‍

Like I said, as the stuff gets more metabolized and these kind of flashpoints related to performances of actors, flashpoints related to scripts that are so clearly just based on, you know, every single episode of Law and Order SVU was just run through some model and now it's being outputted. I think those are kind of like the easy ones. I think much harder and more interesting ones are yet to come.

‍

Cameron (28:56.937)

it's well stated. mean, I think it's an interesting thing where you're seeing both, both threads of the argument getting mixed today. And like, that's not really fair. One is almost a set of economic questions related to human labor and compensation. And then one is, is it art or not, which is sort of irrelevant. Like you could almost make the argument even more convoluted by asking, were the Coen brothers wrong to say at the start of Fargo that it was a true story? Like, no, no, it adds to the

‍

the experience of the film. Like I could also fully get behind an argument that augmenting a performance via AI for the voice to make it sound more Hungarian is in service of the story. It's not necessarily like an impediment to the economic benefits of the cast or crew behind the film. And what's interesting, it's like by that same logic, like is it unethical or

‍

Does it negate the beauty of film craft to do multiple takes and then only edit in the best one? Some people would say yes, and that's why the theater is so wonderful because, you know, the actors on stage and really experiencing the emotion and the energy of the crowd. I think what's really interesting also on that line too is I've even been seeing interesting news of, yes, I want to see an authentic performance from the actor, but on the stage, I'd like it fully rendered.

‍

Cameron (30:22.505)

through like, you know, an LED screen so that like, it feels very, very immersive when I'm sitting in the theater in terms of the sets, but that the acting is authentic and based on the energy of how the actor is performing, which is pretty neat.

‍

Lucas Igel (30:38.392)

I haven't heard about this.

‍

Cameron (30:40.093)

Yeah, it sort of makes sense, right? Like imagine instead of like changing the stage props, just, know, by a flip of the switch, they're sitting on a different planet or on a different scene, you know, immediately. Speaking of this, you know, fast forward five or 10 years, Lucas, what does the editing room of the future look like? Like if Kino has its way, how will day-to-day post-production operate in 2030?

‍

Lucas Igel (31:06.752)

I'm sort of reminded of the namesake of our company. You know, it's silly because I think this is, it was inspired by a documentary that's like pretty much the first documentary they showed you, meaning like your freshman year of like film school. It's called Man of the Movie Camera and it was made by this documentary named Ziga Vertov back in like the late 1920s in Russia and where throughout the streets of Odessa, Ukraine, he just recorded everything that he saw and these like pristine

‍

trolley cars going by and these like amazing factories that were producing all of this surplus. His whole idea is that, you know, much is very similar to the earlier thread you were on. Theater is bourgeois or filmmaking that tries to resemble the theater is bourgeois. And what we should be doing is we should simply be doing what the mechanical eye, this camera thing is truly amazing at, which is recording reality as it really is. And the film and the cinema.

‍

should not have this notion of scripts, it should not have this notion of actors. The artistry should take place in the edit itself and the spectacle is in like just how much the mechanical eye is able to see, the Kino eye, as he wrote about, is able to see. And there's kind of this optimism about technology and this optimism about what his art form could do throughout like the essay that he wrote to correspond with this film and in just the edit itself. And most of it was about like spectacle.

‍

It was like, look at how I took a thousand different shots and I'm allowing you to see it in five seconds. Because he was able to interlace them and create these incredible montages. Like, you know, look at just how many people are out on this beach and how amazing it looks, you know, when I frame the camera this way and how every single one of these is like a real human person. I think the ability for people to create spectacle is going to increase dramatically, especially talented people who are experienced in this field and know what it is that they're doing.

‍

It feels like we have such a small number of people who are allowed to create something that is truly jaw-dropping. I think of like, and it almost feels like a lot of the debate is, our movie theater is going to continue to exist. There's this whole art form built around seeing things on the biggest possible screen and being able, giving these very opinionated auteurs full rein to arrange human.

‍

Lucas Igel (33:29.218)

beings as needed in order to create some like amazing spectacle that reflects how we are as people. I think these things are very intertwined with engineering. I think these things are very intertwined with whatever technology is able to give us at the present moment. In the case of that documentary from the late 20s that we based a good chunk of our branding on, chemical film and...

‍

move the movie camera, as they like to call it back then was sufficiently plentiful and sufficiently cheap and sufficiently portable for him to go around the whole city and just record absolutely everything he thought was beautiful. I think of digital cameras being able to, you know, do that times a thousand where there was no longer any consideration about whether the storage constraints were sufficiently bad. You'd asked about the editing room, all of those different forces that I'm mentioning of storage getting even cheaper, it being even easier to create ideas that are just in your head.

‍

the ability to summon footage that you need and increasingly the ability to kind of augment and create human performances. I think the editing room should be something that allows you to harness all those things in order to make good, you know, good movies, good art. And I think those rooms will exist. How many of those rooms exist is important. And I think what we're trying to do is preserve the spirit of, you know, when filmmaking was, its future was a bit unclear back in...

‍

when there was a series of tech demos in like the early 20th century. And, you know, very rigorous debate was going on as to whether this was, you know, how artistic this stuff even could be. I think it's exciting. I am concerned about, I would say, whether the business most known for putting these things out, the studios that support the Nolans of the world and the mini-mini-indies that are around, making sure that...

‍

the people who choose this as a career to do this completely full time as professionals, making sure that those people have editing rooms that are sufficiently stocked and allowing them to do what it is that they want to do. Making sure that that is kept intact is of course important.

‍

Cameron (35:38.355)

I think it's a beautiful vision of the future of cinema and what is cinema is the meta question I think I heard you answering in your answer there, which is exciting in that we're still exploring that question. Like we still don't have the answer and it does align sort of in my life's work as well of trying to create technologies that enable more

‍

explorations of mind through creativity and art. And you mentioned it, and I think it's accurate that like the access to tell these really sophisticated stories and then make them spectacular through visions and through moving images has been historically very constrained. And alleviating some of those constraints is change in the world, which obviously can create concerns, but

‍

also historically has opened up opened up even more opportunities. Not only to create better art, but to create different types of economic models, to create different types of human flourishing. So that's a really awesome, awesome vision, Lucas. I appreciate you sharing that. I have just one last question, which is there is an explosion of AI startups claiming to revolutionize everything, including filmmaking.

‍

What do you think will separate the winners from the also rands in this space?

‍

Lucas Igel (37:10.446)

if they're useful, I think, is a big one. I love mid-journey. I love how weird the origin story is. This accidental startup that refused venture capital funding. For those who don't know, it just started out as a bot, an app inside of the kind of gaming Skype thing called Discord. And you just type something in, and then a beautiful image of that will appear.

‍

Cameron (37:36.691)

famous for producing the viral image of dripped out Pope Francis.

‍

Lucas Igel (37:41.164)

Yes, yep. And a very small number of people work at this company, and they very much work very hard to keep a distinctly mid-journey field, all of the images that come out of there. The founder revealed that a vast majority of these images people keep to themselves. It's like drawing. It fulfills the same need as drawing. It doesn't, I think that's perhaps what I mean by useful and fulfilling. The startups that...

‍

are producing these tools that people actually do want to and enjoy using every single day or as needed is what is going to set these things apart. There's a lot of discussion right now about incumbents being able to absorb all of the gains that are about to happen. I remember Avicii sent me a fun document that he wrote, spelling like a pretty crazy future ahead where he showed that the incumbents before the dot-com era and after the dot-com era did okay.

‍

But the completely brand new companies like Google and Amazon just completely crushed it. Companies that did not exist until cheap consumer broadband. I wouldn't say broadband, just cheap consumer access to the web was a thing. When mobile took over, you still had brand new companies. You had your Facebooks and others. But those incumbents from the previous era did very well. With the cloud revolution that happened in the following decade, similar thing. Incumbents did even better. And new companies, your slacks.

‍

your stripes also happened. But again, you just had this interesting era where the incumbents seem to be decreasingly asleep at the wheel. And I think what a lot of people, especially in startup world, really just everywhere are deeply concerned about is this idea of platforms and this idea of incumbents sort of incorrectly adopting these magical new technologies that are coming up. If this is the age where the computers can talk to you,

‍

I, you know, should we be concerned about whether Apple is going to competently integrate all these different things in order to deliver this vision to people? Or are we concerned about the likes of Google, you know, taking what was already good about the previous version of their business and kind of cannibalizing it by making it so you don't even need to visit websites anymore? How you actually build brand new tools in order to take advantage of this technology is, you know,

‍

Lucas Igel (40:01.932)

You have to do two really hard things now. Not only do have to make something that people want and that people can use and are willing to pay a significant amount of money for, you also need to compete with, albeit sluggish, but important incumbents who are also, you know, have the mandate to do very similar things. So that's a big part. I mentioned mid-journey at the beginning because, you know, the word taste gets used a lot lately. And I think that is what they have demonstrated. This kind of ability to generate images that...

‍

you know, every single large tech company now shoves down your throat. Like if you go on Instagram, there is like a button that allows you to like generate images of yourself and other people. And I don't know anyone who actually uses it. There has to be, it seems like there's a heart, there's heart to it. And there's like a, there's a sense of care that seems to need to exist for these really, for people to actually use these things with joy, not because they have to.

‍

Cameron (40:43.849)

Hmm.

‍

Cameron (41:00.497)

Absolutely. Lucas, thank you so much for your insights. Congratulations with all of the success on Kino. And I can't wait for more and more people to get their hands on this awesome technology that you've been building.

‍

Lucas Igel (41:13.08)

Thank you so much for having me, Cameron.