AI Directing

00:00:00: I'm hoping for a future where AI is also encouraging the creatives and the clients in the

00:00:08: very beginning to think of stories that we can only tell with AI. This is the commercial

00:00:17: content podcast. What marketing decision makers need to know about the future of content. In this

00:00:22: episode, Norbert talks with Florian Meimberg about AI directing. Florian studied graphic design but

00:00:29: just for one semester because he grew impatient and entered the advertising industry straight

00:00:33: away. As a creative director, he was responsible for numerous campaigns and commercials which

00:00:39: became the starting point of his directing work. In 2009, he created the Tiny Tales, publishing

00:00:45: hundreds of 140 character stories on Twitter, which later won the Grimme Online Award and were

00:00:50: released as a book. He is a founding member of the directing association direct, member of the German

00:00:56: advertising film academy and the art directors club Germany. Ladies and gentlemen, let's dive

00:01:01: right in. Here's Florian and your host Norbert. Welcome Florian. Thank you Norbert. Nice to have

00:01:07: you here to talk about you and what you're doing. And to warm things up a little bit, I have a

00:01:12: format called seven questions which I will ask you. And the first one is which media had the

00:01:17: greatest influence on you as a child? Which media? Television, games, or books? So, why did you leave

00:01:27: out cinema? Uh, good question. Let's see. So, I did watch a lot of television, I guess, the

00:01:36: couple of channels that we had back then. But when you talk about influence I would almost

00:01:43: say the movies, the cinemas, because it's such a different experience. I will for example

00:01:51: never forget the first movie I ever saw at the cinema which was the never ending

00:02:01: story. It was the first movie I ever watched and I will never forget the

00:02:07: effect. Really, it's like a photographic memory. That one moment that I will never forget is I looked

00:02:13: at that screen and then I looked over where my mom and my brother sat and their heads were

00:02:20: tiny. Like the contrast between seeing the huge people on screen compared to a real person next

00:02:27: to you is something that struck me, right? Because that moment was a new experience to

00:02:35: see the huge big cinema screen. So that was obviously a major impact moment in my

00:02:44: life. Other than that I for example was never the gamer. A lot of my friends were always like

00:02:50: into what was it Amiga and stuff, that was not me. I had a Game

00:02:57: Boy at some point but that was it for gaming and then a lot of television as I said. Okay, so

00:03:04: but officially we're going to take the cinema now onto the question because actually I also

00:03:09: maybe have to add internet and social media later because but as a child we grew up

00:03:14: without it. I know that so I didn't ask it. Okay so what was your first encounter you remember with

00:03:20: actually advertising. Ah, a first encounter with advertising? I mean, next to obviously advertising

00:03:29: being everywhere, not as much as today, but of course also back then you had TV commercials and

00:03:36: outdoor advertising and stuff. But when it comes to and I assume that that's the

00:03:43: base of your question when my idea started to maybe go into that direction was I think

00:03:50: when I was doing my high school exchange year in the USA. I spent a year in the US when I was 17.

00:03:58: So my 11th grade I spent over there and at the high school there was a course called marketing

00:04:04: right. And that was interesting, that was also scratching topics like advertising and stuff and

00:04:11: I remember seeing that as or finding that very interesting so that could have been one of

00:04:18: the sparks. Okay nice. Yeah I remember this 11th grade I never did one but my brother did one. This

00:04:23: was common sense to do a year abroad right? Yeah, my son is 17 now, just finished his year also in

00:04:30: the US. A very important year in my life actually. Yeah. Nice. So the next

00:04:37: question is where do you currently prefer to spend your time? The movies, the console or social?

00:04:47: The movies not enough because it obviously involves getting up and getting out and me having

00:04:54: yeah a job and a family. It's not always easy to find those slots, which I regret

00:05:01: because I still find the movies to be one of the most intense experiences you can make.

00:05:10: Um yeah I do spend a lot of time on social media, also job-related somehow right, because there is

00:05:20: nowadays for example LinkedIn is a platform that I spend a lot of time on right now

00:05:27: because this new AI film bubble that we are I think going to be talking about today

00:05:33: also is very concentrated on LinkedIn. The algorithm in my feed on LinkedIn is let's say

00:05:41: 90% all of a sudden of this AI. So if you count LinkedIn to social media then

00:05:51: yeah, but also the other ones. I also do TikTok a lot, probably too much. Yeah, like you have

00:05:57: an account? You don't? I don't have an account, I just consume it and every day I realize how

00:06:04: addictive that is right. Yeah, I think that's a big topic. So, I don't have TikTok. I mean,

00:06:10: I logged in once, but no. Yeah. I mean, it's literally dangerous. If you

00:06:17: maybe do it as an experiment. Go on TikTok and then just for let's say 3

00:06:24: days and then compare how fast the algorithm learns you. You wouldn't believe it.

00:06:32: Oh, well maybe I don't have to try anyway. But actually, I'm watching people, I mean a lot,

00:06:36: in public transport and I watch people a lot, especially young people on their feeds and

00:06:42: it's a little scary I think because the speed they watch it. It is. As I said I have two kids

00:06:48: myself. So this is obviously also a topic with us at home in the family and I'm also following

00:06:53: the news with what was it Australia. Yeah, it was pretty easy with the law and stuff like

00:06:57: starting at 16. Now Germany is talking about it and yes there is a reason for people and

00:07:03: politicians to talk about it I think so. Yeah. Okay. So if you were on an island and you could

00:07:10: only use one channel or app to take with you which one would it be? Interesting question.

00:07:21: Which one would it be?

00:07:29: I mean, there's two sides to the question. So, if you talk about like consumption and

00:07:35: what info I can get, I would probably almost say TikTok because as I just said with the algorithm,

00:07:41: right? When you think about TikTok, most people think about funny dance videos and

00:07:48: stuff. My TikTok feed, for example, is I would almost say more than half political stuff. Okay.

00:07:56: Especially US related stuff. There's a couple of like very young influencers

00:08:01: that do a lot of really really good and serious quality on TikTok,

00:08:09: right? And I'm not saying I get my news from TikTok. I watch, I'm very old school in watching

00:08:15: the news every single night and also insisting on my two kids also watching the news with us. So

00:08:21: we all as a family watch the news every day. Um but yeah that one app on the island. So either

00:08:29: TikTok or when it comes more to apps regarding communication to stay in touch with

00:08:37: everybody. I know phone app is not allowed. A phone app is not allowed. Well okay. Okay.

00:08:42: If you talk, I mean if you take the phone app. Okay. But that's actually interesting because

00:08:46: I always catch myself thinking in the patterns which fit my age you know. So

00:08:52: communication is phone. Remember now when I say reflex phone is not allowed but if you have, I

00:08:56: mean you know this. So what would be your channel on the island? Probably news. Yeah

00:09:03: probably. But the news you watch on television. So okay for example that's the one. Okay so

00:09:14: little different topics. So how do people outside the industry usually misunderstand your job?

00:09:24: Probably depends on who you talk to right. I mean my close family by now more or less

00:09:32: understands it because I talk a lot about it and they're very interested in learning what

00:09:36: I actually do. And then yeah, you know, I come, I was almost going to say back in the old days,

00:09:44: home from a shoot abroad, let's say Cape Town, and show them pictures and show them, you know,

00:09:49: look at all the restaurants and all the beautiful places and stuff. So they,

00:09:54: of course, what was the question again? How do people misunderstand your job? Oh,

00:09:59: they misunderstand my job. Um, let's say probably a lot of people don't know how

00:10:09: much stress is actually involved. Of course it's a dream

00:10:16: job but you're going to Cape Town for work, it can't be that bad. That's what I

00:10:20: mean. That's what I mean. But then maybe I can tell my wife or my son then we swap and

00:10:26: you do that client meeting late at night in Cape Town where the client gets nervous

00:10:31: and stuff, you know. Of course, if all that happens in a beautiful surrounding, I'm not

00:10:37: complaining at all. But the funny thing is when you just said back in the old days or when you

00:10:43: started when you try to start your sentence with back in the old days, we're going to be

00:10:46: the guys who've been to Cape Town, you know. I know, someday like we shot on film. Okay, we've

00:10:50: been to Cape Town. That's one of the guys who's been to Cape Town, right? Yeah. Yeah. Okay. So,

00:10:55: if you were no longer allowed to work in advertising, what would you do? Oh,

00:11:06: tricky one. I never learned anything else. I actually also

00:11:13: didn't go to any college or university. After school I just dipped into that

00:11:20: world and stayed there forever until now. So, um, it's the only thing I'm really good at. Does that

00:11:29: sound terrible? No. Uh, let's see. Let's see. I don't even know. You know, my wife, for example,

00:11:36: is a writer. She writes novels. Okay. And every day I'm telling her how I could probably

00:11:45: never do it, you know? I'm really admiring her for that skill to create a story

00:11:54: that is so complex and to work on a story for a month, right? Um before it goes out there and then

00:12:03: I look at myself on the other hand that does like 30 second commercials for a couple of weeks, very

00:12:08: compressed, also very complex, also challenging and stuff. Uh, where did I want to go with that

00:12:16: answer? I don't know, try to become a writer. I mean, I could. I think I would probably yeah,

00:12:25: most definitely look for something that goes into the same direction of I don't know. Yeah,

00:12:32: it's like the overused word storytelling. Tell stories. I mean, create something. It would

00:12:38: have to be something where I'm able to create something. I think that's the core of what I

00:12:44: have always been doing and what keeps me alive and keeps me doing that. Good. Nice. Thank you. So,

00:12:53: um let's go to the actual topic of this. Um tell us a little bit what's your life as a

00:13:01: director of commercials right now? You know, I mean you head up for Korea and you have been

00:13:07: doing this for some time already. So, what would you say? What is your job right now? Actually,

00:13:14: that job has changed a lot recently. I would almost say it almost turned 180°. So,

00:13:23: um, right now all the projects I'm doing involve AI. Some of them full AI productions,

00:13:32: full AI films. Some others hybrid productions when it's a mixture of classical shooting and classical

00:13:41: together with AI which for me is by far the most exciting direction to go into. I think we will

00:13:48: go a little deeper into that. Um yeah. So my, if you ask what does my job life look like

00:13:57: today, different than a couple of years ago. Because a couple of years ago it was much more of traveling

00:14:03: obviously when you shoot. Um and the AI stuff, a lot of it you can do from home which also

00:14:13: obviously has its good side as a family father and husband. So yeah, a lot more time at home,

00:14:20: but that doesn't mean that it's less work or less stressful. It's more the opposite. Okay.

00:14:26: Why is that? Yeah, because I mean the thing I'm working on right now regarding the

00:14:35: number of meetings and yeah client contact and production company and agency is like 10 times

00:14:44: more as on a normal production. So a lot of these jobs feel almost more stressful than a classical

00:14:52: one. That might have to do with the fact that this whole world still feels new to everybody, right?

00:15:01: You discover new things every single day. When you work, when you go into it and stuff, the

00:15:06: workflows that you work on are developing while you do it. So it's like this whole, like a huge

00:15:15: door that all of a sudden sprang open with a new world behind it that we're all just yeah, crawling

00:15:20: into. And is it a lot of moderation and explaining in between all the departments and the clients

00:15:28: and everyone involved? Because yes I heard that sometimes it's actually more work now because

00:15:33: you have to constantly tell people what actually is different now. A big part it

00:15:40: also obviously depends on the job itself. Of course, sometimes when you work with somebody or

00:15:45: for somebody, they know a little bit already about AI and then there's clients that have no idea. Um and

00:15:53: then yeah there is a lot of explaining to do and that needs time. But it's also very very important

00:15:59: because when you look at our industry and you've been doing it for many many years, I have been

00:16:04: doing it for many many years. The structure and the like the different steps that a project

00:16:13: goes through were always the same right. You have the script in the beginning that everybody has to

00:16:19: agree on, meetings meetings meetings, at some point you have the final script. Bam. Then

00:16:23: you go into production. There's the pre-production phase and you know, you know the game, right? You

00:16:29: have like the meetings that you've always had, the PPM before the shoot where you confirm

00:16:36: everything, the locations, the cars and everything. Then you go on the shoot. So

00:16:41: it was always the same steps and everybody knew them. Right now these steps have totally

00:16:50: changed. So in an AI production for example, one of the most important meetings you will

00:16:58: ever have with everybody involved is what we right now call the first frames approval. What

00:17:05: is that? Most people will know but very quickly: producing an AI film mostly involves creating

00:17:14: stills in the beginning, the so-called first frames, because that still you then feed into another AI

00:17:20: tool and bring it to life and animate it. So to define these first frames is so important because

00:17:27: these first frames contain everything that the client needs to approve. So when I'm

00:17:34: in the next phase and I already have animated pictures and a running film and then the client

00:17:40: notices, ah I want that guy to have a yellow shirt instead of a blue one, then that is something

00:17:46: that we should have talked about last week in that first frames meeting. To explain

00:17:52: to clients these new structures and these new ways of important milestone meetings within

00:18:02: the process is very important and also of course takes time. Can you say something about

00:18:08: or explain a little bit about the process combining AI with a shoot actually because

00:18:15: um then people are used to stuff but there's a layer coming on top. Is it more difficult than full

00:18:21: AI or what would you say? Yeah, that's a big topic so this is a term that we also right now

00:18:29: hear more and more: the hybrid productions. So the combination between classic shoots and AI.

00:18:37: I myself have done three of those in the past couple of months. Um, and they were all kind

00:18:44: of different. So, let's, I mean, one of them, for example, was one of the biggest ones. It's also

00:18:51: one that's really heavily running on TV. So, I can also say it is the Persil film for the

00:18:57: washing detergent Persil. Uh, which was production wise very complex and very lengthy. It went over

00:19:04: a month and there for example when you talk about hybrid approaches the first thing is that from

00:19:11: the beginning the client insisted on casting the main role. The main role is the character, the lady

00:19:18: in white, which in the film, just very quickly, is who doesn't know it, um everybody probably knows

00:19:25: those Perwoll clock towers that are still standing in big cities around Germany, like really high 7

00:19:32: m high green clock towers as advertising for Perwoll and they have this illustration of a woman in a

00:19:40: white dress from the 1920s and this very iconic figure, the white lady, is like one of Perwoll's

00:19:47: core DNA brand assets, right? And in the film which starts in the 1920s and you see one of

00:19:58: those clock towers and then that illustration all of a sudden wakes up, becomes alive and the white

00:20:03: lady as a real person steps out of the tower and starts walking and then travels through all the

00:20:09: decades and stuff. So that's the film. Um so yeah, where did I come from? The client said,

00:20:16: "We need to, I want the character based on a real person because" and this obviously also

00:20:24: has to do with Henkel being of course a very old school and conservative client that's been

00:20:30: doing advertising for a hundred years, right? But always the same. They always casted their models,

00:20:36: paid them their buyouts, and that was one aspect that the client obviously wanted to keep. So,

00:20:43: we said, "Okay, let's cast the main woman like a real casting."

00:20:47: Yeah, we did a classic casting with set cards and stuff. And then we had a woman that was confirmed,

00:20:53: right? And also got paid normally as a model with all her buyouts and stuff. Okay. But technically

00:20:59: what we did then is take that woman into a photo studio, make like 500 pictures of her from all

00:21:07: angles in different light setups and stuff and then yeah create a so-called digital twin of her

00:21:15: by feeding those still images into an AI model to create that digital twin, with which then you can

00:21:24: do anything by prompting, right? And the woman can go home and it's done. But that was the yeah, the

00:21:32: classic aspect at this point. And then still talking about hybrid productions, for this film it

00:21:39: was very important to be um historically accurate right, within the scenes she

00:21:46: walks through different decades. Berlin 1920s, Berlin 1950s, 1970s, '90s. So, and we all know

00:21:54: that nowadays the shitstorm is waiting around every corner. So, as a classic brand like Perwoll,

00:22:02: you better be very accurate in what you show. Which is why then in the first phase of the

00:22:10: production that we call world building. It's a term that comes from the gaming industry.

00:22:16: World building in the AI context means you define and collect the ingredients for each

00:22:24: shot, right? Architecture, cars, wardrobe, very important stuff. Like everything you see, you need

00:22:31: to know how to build up that picture. And for this phase, we booked a real production

00:22:38: designer and a real stylist to supply us with accurate moods, mood pictures, right, of real

00:22:47: Berlin architecture. For that um the production designer went into like archives to really

00:22:54: research accurate real photos from that time. The same with the stylist who

00:23:00: gave us yeah mood worlds and pictures of all the outfits that are accurate. Um so yes we

00:23:08: used those classic departments in this very important phase of the production before we actually started

00:23:15: prompting a single image. Well, important point of course. Yes. And this is to sum it up what

00:23:21: made this production hybrid because from then on all the rest was full AI. So the film that you see

00:23:30: is actually 100% AI generated. Yeah. But you also shot hybrid commercials with real people and

00:23:38: uh I did. And yeah, as I just said, I mean I can tell you all three. There was the first one,

00:23:41: right? The second one I just recently finished. It was for a German supermarket. And the story,

00:23:49: it's also on TV right now, so some might have seen it. Very simple story. Family at a lake

00:23:54: doing a picnic, right? At some point, dad wants to go fishing. He takes out his fishing rod,

00:23:59: says, "Okay, guys, today we're having fresh fish." Throws out his fishing rod,

00:24:05: and then something bites. What does he drag out of the water? A whole supermarket, the building.

00:24:15: Like a whole building. Ah, he drags it on land, goes in and gets out with a fresh fish. Right.

00:24:22: So, for this one, we actually had a very classic shoot with a casted family. Also, the father,

00:24:28: everybody cast, we shot in Spain on a beautiful lake in Valencia. And then the AI part

00:24:36: obviously is that building coming out of the water that was integrated into the real images with AI.

00:24:45: So I mean and then it's obvious why this was maybe not done in a classical VFX pipeline because it

00:24:51: might have been timing wise or budget wise or I don't know the reasons. But comes to the point why

00:24:58: do people actually approach these AI solutions and this is also always with whoever you talk, uh

00:25:07: of course it's always that question: is it only out of budget reasons? Of course it's a reason. So yes

00:25:12: of course you could have done it in 3D and also in the very beginning the involved people were

00:25:18: also looking at different approaches. The 3D option was also on the table but yes it would have

00:25:27: taken obviously much longer and of course would have cost much more. So yeah I mean the

00:25:35: budget question is of course always a part of the equation. Yes. Yeah. Um so but before I ask

00:25:43: you this there was a third project you mentioned. So the third is yes also um my third yeah hybrid

00:25:49: production again a little different and this was very hybrid. So the from the whole thing that was

00:25:56: almost my favorite one. It was for an Austrian agency and production company um for another

00:26:05: supermarket and they over there in Austria every year they have the sticker album right where you

00:26:12: put in like the Panini album in Germany, very very popular and they always, every year at

00:26:19: the beginning of the year they do a commercial. It's always the same basic idea: the female

00:26:24: protagonist is like a really famous TV host over there and a couple of kids traveling to

00:26:31: some world. Every one of the sticker albums has always a topic right and this

00:26:37: time it was like the Nordic Viking world right. So the story actually was the

00:26:45: film starts in a living room where they check out the sticker album and all of a sudden they get

00:26:50: beamed into that world onto a ship, so they're on a Viking ship on the ocean and then there's like

00:26:56: a huge whale jumping out of the water and like an octopus, a pink octopus with the

00:27:02: arms coming out of the water and stuff. Um, and this one we actually did a real shoot for the very

00:27:10: first scene in the living room. Real people, cast people, everything real location with obviously

00:27:16: all the departments involved from DP to production design to everything. Then we spent another day

00:27:24: in a green screen studio and in that green screen studio I built with the production designer half

00:27:30: of the ship in real right and then again we had of course the actors in the green screen studio

00:27:36: and the ship to shoot the shots playing on the ship. So and then the backgrounds for that

00:27:43: plus all the wide shots were then completely AI generated. Okay. So it sounds also like I mean

00:27:50: pink octopuses, that would sound like VFX heavy probably. Yes. So

00:27:57: um but as you tell all these stories about the shoots you did. I actually, so how did you end

00:28:02: up in this AI field because I mean people who followed you, they know that you're doing this

00:28:06: for quite some time already. So how did you start or why did you start to actually

00:28:12: explore the field? I was, you know, I was always somebody who, whenever there was something new,

00:28:18: instantly wanted to know all about it, right? Especially in our industry, right?

00:28:23: Everything that technically was something new, I was one of the first ones who wanted to at least

00:28:27: see it or understand it or try it out. And yeah, and then on some day I saw MidJourney

00:28:35: for the first time and realized what you can do: create a picture by writing a short prompt. So

00:28:43: I jumped into that head first from day one almost. So from the very very beginning I was fascinated

00:28:50: by the tools. I instantly started to play around with them and then of course also instantly

00:28:55: had results right and I just yeah, this just fascinated me. So I stayed with it. I just more

00:29:04: and more, I spent a lot of time just learning, checking out the tools, comparing and experimenting

00:29:09: around and stuff and yeah during that time that topic also became aware within our industry

00:29:18: and at some point there were first people that wanted to, ah maybe we can even try it out for

00:29:22: real on a real production. So I more or less yeah grew into it together with the

00:29:32: technology itself from the beginning obviously. Interesting would be in that context.

00:29:41: Yeah, context. Thank you Florian. Um to talk about, I mean you started to explore AI yourself. So you

00:29:50: sat down and used the tools. You're a director by nature. So do you think it is the future that

00:29:59: actually directors are in the position to you know build all these worlds, explore and build

00:30:05: and prompt themselves or is it the role of the director and other people are supplying the images

00:30:13: or the footage? How do you see this field? This is nothing that you could answer with one sentence.

00:30:22: It's very different. I mean also on my jobs, right? In the last couple of months,

00:30:28: I've done a couple of full AI films that I did all by myself. You know, one of them even without

00:30:35: a production company or agency between it, but based on an incidental meet up with a guy

00:30:45: that I met on a plane and we stayed in contact and he contacted me at some point and I made a film

00:30:50: for him. He has a company and I did this all by myself, all by myself in the end right. There's

00:30:56: other ones like the Brazil one that were much more complex, a classic big client, big agency, big

00:31:02: production company and in the production then also like a team of a couple of people also technically

00:31:08: around me at my side, also in the actual prompting AI world. So there is also, I mean, there you can't

00:31:18: really answer the question. There are people. There's young people, young talents also right

00:31:23: now, without almost no background experience, growing into the field that are also good,

00:31:29: right? And can also yeah work within the industry. There's guys like me, you know, that's

00:31:35: been doing classic films for decades and am now adapting to the new technology. There's yeah,

00:31:43: many ways into it, I would say. So, it depends on the project like usual actually. Commissioning

00:31:50: whom. So what does it do to the actual, you know, in a film shoot, in a classic film shoot there's

00:31:56: a lot of people involved. And so you mentioned that on some projects you work together with like

00:32:02: stylists and art department who like looks for moods and defines the style. So how do you see the

00:32:11: industry evolving or changing also in terms of who's actually involved in doing such a film

00:32:19: because of course it's connected to the size of a production, the production companies, the landscape.

00:32:27: You know, you've been doing this for some time now, so what do you see actually on the people's side, who

00:32:32: are involved or not involved in projects? Because you know, yeah, this is actually one of the aspects

00:32:39: of the whole thing that and I'm serious, also scares me the most. Right? As I said, I've been

00:32:46: doing films for 20 years plus, right? A couple of my closest friends I've met on film sets or within

00:32:53: production circles, right? And the thought of first me spending the rest of my life in front of

00:33:02: a monitor and secondly never having, and this is actually something that gives me my energy, is

00:33:08: the exchange with people. I love to travel. I love to meet new people. Whenever you travel,

00:33:13: of course, some of the heads of department you know already, you book them because you know them,

00:33:19: but there's always new guys and it's always the magic mixture of creative energy and you

00:33:26: create something, develop something together with other people. And this is for me still

00:33:31: where the magic of film lies, right? Um, and I don't want to lose that. I think this is also

00:33:38: the reason why I'm fighting so much for those hybrid productions and I'm doing really in the

00:33:46: last couple of weeks and months I have done quite a couple of those podcasts like here or as a

00:33:52: speaker on events and stuff and I always end up yeah saying guys let's be hybrid,

00:34:01: let's kind of save what we have there with all the classic tools and people we have

00:34:07: because of course it's a danger. You see those tools develop. Oh my god. Seedance 2.0 is coming

00:34:15: out next week and then there's a couple of first clips going around which makes everybody say okay

00:34:19: that's it. But now we obviously don't need anybody anymore. I don't know. I don't know what will the

00:34:26: future look like. It will dramatically change. It is just changing right now. And of course I'm also

00:34:34: talking to colleagues, right? Um I'm for example one of the founding members of our direct commercial

00:34:42: directors union that we founded here in Germany a couple of years ago and this is one of the main

00:34:48: topics and there's also there's different sides. There is a couple of also my colleagues that also

00:34:54: use AI very heavily and also jumped into it and stuff and play around with it but there's also

00:35:00: some that still can't relate and are more yeah, rejecting AI and actually I always end up saying

00:35:11: don't, be open because I mean the point of no return is far over. It will be part of all our

00:35:21: lives right. It will also, advertising is the first field of film that is being, you know,

00:35:31: yeah, not replaced, but I mean AI already there. I just talked about it. I made full AI commercials

00:35:38: myself. You see more and more AI commercials. So I always say also to my colleagues,

00:35:42: you better learn those tools so you at least know them. The same goes for producers as you,

00:35:48: right? Even though you don't necessarily have to actually physically work with the tools, you need

00:35:53: to understand them because if tomorrow your phone rings and there's a client asking you, can you do

00:35:58: AI films? At least you have to understand how the tools work, what they can do and also what they

00:36:04: cannot do, right? Because when a client asks you, should we do this in AI? The answer could also be

00:36:10: no because classic would be better. So I'm just trying to always um convince everybody

00:36:19: to at least get to know what it's about, how it works, to then think about and decide where and how

00:36:30: to integrate it. Because in our little bubble with the commercials, it will probably um step by step

00:36:38: become part of the process. I mean when I think of all my big productions, you know, a couple of

00:36:43: shooting days in Cape Town and then I always end up saying in the old days but usually you would

00:36:49: let's say there is in the script, in the storyboard, there's like the establisher shot, like one epic

00:36:55: wide shot of anything, a landscape that you would normally have spent I don't know a third of

00:37:01: a shooting day, drive somewhere, yes maybe with a smaller crew but to shoot that one frame.

00:37:09: Today that one frame, beautiful wide shot, is generated within a couple of minutes and you

00:37:17: save maybe half a shooting day. You know what I mean? So it will become more and more part of what we do

00:37:24: and I think we especially as directors need to be like a stand-in conductor.

00:37:32: Yeah. To level out and to also understand and yeah tell the rest

00:37:42: where AI makes sense. Yeah. So actually talking about the stuff which is now maybe done by AI in

00:37:50: the future, because you mentioned it not everything can be done by AI and you would also say this is

00:37:56: not an AI project. What would be the parameters where you say okay this is, we don't do it, we

00:38:01: better do it real. Well I mean whenever it comes to emotional human stories right, where it's about

00:38:12: people and also where I see people in closeup and where I want to with the film create empathy for

00:38:21: that person on the screen right, um then I would always try to go for a real shoot because of

00:38:30: course a good actor is what they do in the subtext, what's not even in the script, like the

00:38:37: emotional flavors that you feel more than you see them right. This is I think something that

00:38:47: of course the real actor can do better than an animated one. And then on the other hand

00:38:54: you look out into those bubbles on the internet and you see the first tries with that new tool

00:39:00: coming out where you realize, wow, if you would have shown me this clip, I wouldn't even have

00:39:07: noticed anymore that it's AI generated and it's a freaking emotional clip that gives me goosebumps

00:39:12: because it's so emotional in that close-up with the woman that just starts crying or something.

00:39:17: So the argument that I use myself and all the other skeptics also use, ah you

00:39:24: will never be able to replicate real human emotions. Technically, visually, you will. So

00:39:34: at some point this decision also becomes an ethical decision more than a creative one. I mean that's

00:39:43: it's something that, as you probably notice, also for me it's a very very emotional

00:39:50: aspect of the whole thing of course. I will to again summarize my answer to this

00:39:56: question. I would always want to, when it involves people, an emotional personal story

00:40:02: with people, I would always try to do it real as we did on the supermarket thing,

00:40:08: right? We casted a real family. Could we also have AI generated them? Of course. Yeah, probably also

00:40:16: as photorealistic and probably also in a way that somebody who's not so deep into all of this as

00:40:22: we are wouldn't even notice. Do I want it? No. If possible, I would always try to have as many real

00:40:31: people involved as possible. Yeah. And that's actually a good um hook to ask for this quality

00:40:40: discussion, you know, because I remember, well, it's not I remember, it's ongoing. So it's when

00:40:45: the AI projects hit the market and are real television, like not this I did a few clips with my

00:40:51: MacBook in my basement, we're over that, but in like real commercials airing on TV and then from

00:40:58: the industry there's of course quality discussions and people always find flaws, they always think ah

00:41:03: look this is here and this is there and I saw this and I saw that. Um how do you approach that topic

00:41:08: because in the end it's a television commercial and we're talking about the end consumer which

00:41:14: watches the screen. So, how do you see this whole quality discussion? Yeah, it's funny. I just

00:41:20: had a discussion about this exact thing that you just said with the production company on

00:41:25: my current one. But to answer just a step earlier when you said the first real commercials

00:41:33: that hit the market. I also remember that. I mean that was not even a year ago, right? When the

00:41:38: first official, I mean, years of Brazil was one of them actually as far as I remember. Yeah. Then I

00:41:43: remember the Vodafone one that was even half a year earlier. And when you watch this and we

00:41:48: were all impressed when that came out because back then it was state-of-the-art. It was the highest

00:41:54: level that AI could do. Everybody was totally overwhelmed. Right? If you look at the film today,

00:42:00: it looks so old. So old. Right? So these tools and this is part of this whole problem. They

00:42:07: evolve so fast, right? The quality level that they actually produce rises so incredibly fast

00:42:15: that you almost can't keep up. So that what you just said with the flaws, I think

00:42:21: that will always exist but it's shifting with the progress. Here's an example. What you just

00:42:26: said is true of course that clients are especially with AI you know looking even closer, what little

00:42:33: flaws they find and little imperfections. Now nowadays that the overall quality is so good

00:42:39: and you can almost not even tell it from real film, they still find them. The other day we

00:42:45: had, I won't say any names or companies, but the other day we had like one of many many many client

00:42:52: approval meetings when it came to what I said before, the so-called first frames, right? So, we had

00:42:59: the first frames that were perfect and they always ask me, Flo, are you happy with them? Yes,

00:43:04: please. These are our first frames. Now, there's another approval meeting. And then the feedback

00:43:09: comes in. Yeah. Here on picture 22B, um when you zoom in the picture 500%, there is this little

00:43:18: pixel irregularity there. This needs to be removed and this and this. And all of a sudden you

00:43:26: zoom in pictures by 500% to make changes. What? Yeah. But that's an industry thing I think because

00:43:32: Yeah. But not in this extreme. Okay. You know I mean if I really. Okay. Back

00:43:38: then I come back from the shoot with my footage. I start editing. There won't be a client that

00:43:43: presses pause and zooms into a picture to find, not zoom actually, zooming. Not maybe in the

00:43:47: online. So I remember a shoot. We were shooting um a food commercial and we had like you know pack

00:43:53: shot, last shot of the day, you know. So we went in overtime already and we didn't have like a

00:43:58: lot of time left because we had to switch out the light because we were in the middle of Berlin and

00:44:02: it was dark and stuff and the client, she was watching the monitor and we were shooting it like

00:44:07: for the 10th time or something and she said yes but nobody's looking this close like we do right

00:44:14: and then of course nobody watches the film frame wise, you know. So but of course this I think it's

00:44:19: a commercial industry thing that people really polish every frame in online or on set.

00:44:25: It is and my feature friends, you know, they don't have the budget for this actually.

00:44:31: Yeah, it is, it is an interesting thing. It always has been. It's only that to maybe round

00:44:37: it up today, with AI um at least for clients, it feels way easier to quickly do those changes,

00:44:47: right? So if you yeah then zoom in 500%. What you don't like, ah these four pixels back

00:44:53: there, then today it's one click to remove them. Yesterday it was I need to go back into the 3D

00:44:59: or into another department to have them removed. So because all these possible changes um can be

00:45:10: done so much faster and because obviously the overall view especially from clients is ah with

00:45:17: AI you can always change everything. This is why it's so important within AI productions to

00:45:24: really define those um important steps in between, those important yeah steps to confirm stuff.

00:45:34: Yeah. And to yeah like points of no return. So talking about stuff which took longer in the

00:45:39: classic process to do like retouching a frame for example. Let's maybe talk a little bit

00:45:45: about the people who are involved in the project besides you. So of course you have in you know in

00:45:49: a classical you have DP and you have like lots of in a camera department lots of people and

00:45:55: you have like a first AD and the second and also post-production. So how do you see that also

00:46:01: changing the landscape, that not so many people are involved in the project at all? You know, like

00:46:07: talking about post-productions and pipelines. Um is it, you know, it depends. What I realize, when

00:46:15: you talk about post-production, what I realize and also I am very active as a director in

00:46:21: social media, whenever I have something new to show, also AI stuff, I put it out there to show

00:46:26: people and also to see how they like it and also to yes maybe discuss, bring it on right.

00:46:32: Especially LinkedIn is a channel that sometimes gets very emotional in the comment feeds

00:46:40: and when I for example on LinkedIn posted my all my AI stuff, doesn't even matter which one, at some

00:46:47: point there were like people I didn't know but haters coming out of their corners to

00:46:53: start to rip apart the film and stuff and all these people were from the VFX and CGI corner of

00:47:01: the industry and obviously afraid of what's coming there with AI. But by now, on the way, I have

00:47:09: also met people from that industry like CGI guys, VFX guys that also learned the AI tools. And this

00:47:17: is where it becomes interesting because I just told somebody else yesterday, it's very important

00:47:22: in every AI production to plan ahead for the last act like the classic post-production act. You will

00:47:31: always need somebody that can handle the classic tools to perfect, to stitch, to finalize,

00:47:41: to polish or also to do bigger things, to do then manual with the classic ComfyUI workflows

00:47:50: and stuff. So I would to any post-production person I would say learn those tools, integrate

00:47:57: them into your workflow and then you will be more valuable than ever before if you can cover both

00:48:04: sides. That's a very good point actually because when I talk about projects or especially

00:48:11: look for people you know for projects, for active projects um it's always yeah we don't know where

00:48:16: to find the people. So I think there's going to be a time, what do you think about it, that people

00:48:21: start to adopt it faster or more than now and then they have like the an advantage of being

00:48:28: the guy, the girl, who knows how to handle these tools. Because this affects all departments right,

00:48:33: I mean if you're a DP and you don't know what is going on there, so what would you tell people

00:48:38: uh which are listening now and they are not sure how to approach this because they're afraid of

00:48:46: their department vanishing. I mean you talked about the art director and the stylist. I think

00:48:50: it's a tool for every department. So totally agree. I would and this is what I would tell

00:48:54: them: no matter which department, I would say are you creatively involved in creating

00:49:02: a film? Then you better learn those AI tools. Another example on the Perwoll campaign for

00:49:10: example, in my little team I had built around me, um there was one guy that is a DOP but that also

00:49:18: then helped prompting and creating the shots in AI because he knows how to use them. But he is

00:49:24: still a DP and is still shooting as a DP. But he can also do this. And other examples like I just said,

00:49:32: you know, with our production designer that helped us with our world building or

00:49:37: um and that was not connected to one of my things, but it was the same production

00:49:42: designer, the awesome Marlin Ortner by the way, that helped us back there and he for example is

00:49:47: also very deep into that whole topic and he uses it for his department but in a

00:49:56: totally different way. So, for example, in pre-production, the other day he told me he had

00:50:02: a shoot somewhere abroad and then way before the shoot, weeks before, he went there alone and took

00:50:08: pictures of like the location where he will build his set at some point, went back, fed these

00:50:17: pictures to an AI and then with help of AI created a 3D simulation of the sets he will build and then

00:50:28: transferred this into a VR world. And then he gave all the clients and agency VR glasses and said,

00:50:33: "Okay guys, let's take a walk across our set that doesn't exist yet." So yes, you can in every

00:50:41: point of a production involve AI tools. I think that's actually what people mostly

00:50:48: forget. If I talk about prompting a first frame or animating it or you know prompting it ongoing

00:50:54: to a moving image, it's actually involved in every step of the process and it's much more

00:51:01: than just the picture, you know. Yeah. So but still I mean you hear how optimistic I'm trying

00:51:07: to be also for other departments and stuff. But coming back to your first question and to also

00:51:12: the topic right now here on the Berlinale, everybody is talking about just that um. I do think that in

00:51:21: the future of course there will be much more full AI productions that in the end create films that

00:51:28: look 100% real and convincing and emotional and stuff. And of course that will lead to many people

00:51:37: in those departments having getting in trouble or not having enough work. That's something that

00:51:44: I am afraid of. Yes. Yeah. Because at the moment you talked about the creative heads of departments

00:51:48: which as the name says have a department. Yeah. Yeah. So and there are some jobs which

00:51:54: um yeah will and this is where the ends meet. The front runners, the people who

00:52:03: adapt early enough and who can make it part of their world by learning it and integrating it, they

00:52:12: yes they will survive. Okay. So, good. Um, yeah, emotional topic. So, I'm trying to do the circle

00:52:24: to the bigger picture of the industry because of course there's not only production involved

00:52:29: in a project, but you mentioned it already. There's also projects you do directly yourself

00:52:34: for clients. So, we all know the process, there's a client, there's a production company, there's

00:52:38: an agency which commissions the work to the productions. How do you see this landscape

00:52:44: evolving right now? Maybe in your projects or in the future because if you as a person who's

00:52:50: creatively managing the project can do it kind of yourself in some cases, it also affects

00:52:56: the whole industry of the commissioning chain. And it will. I mean for me um and I think that a lot of

00:53:02: people in our industry are experiencing this, that in the last couple of, let's say I don't know, 10

00:53:08: years there was already a shift towards, I mean you heard more and more, ah the production company

00:53:16: produced that campaign directly for the client. There was no agency involved. I

00:53:21: had a couple of those. There was a production company that, for a client, we did like four

00:53:27: big campaigns with like seven, eight films each, like four big productions. Twice in Cape Town,

00:53:34: twice in Bucharest, directly for the client. One person from the client, no agency involved. It

00:53:40: was a small team, almost like a funny class trip. You know, you knew each other very well,

00:53:46: very trustful relationship, but yeah, but without the blown up structure in between, that of course

00:53:55: makes stuff slower and more expensive. And I think that AI will accelerate that development.

00:54:06: So yes, there will be. I'm experiencing it myself. I've done a couple of AI jobs

00:54:12: directly for a client without agency or production company. And I also think that this

00:54:18: will yes, probably accelerate. Yeah, probably think so too. Okay. So looking back um to what

00:54:30: you did in the last months and years, what has really made a difference in the last 12 months from your

00:54:37: perspective, you know, like a paradigm shift or something where you said okay this is now, I

00:54:42: mean you just mentioned Seedance is like this is the next paradigm shift or was it another tool or was

00:54:47: it a mindset? So that is a funny thing somehow right. If you follow that and as I said

00:54:53: before for me that is my LinkedIn feed where as soon as something new happens in that world, if some

00:54:59: of the models release a new version or something I see it first on LinkedIn and then it's,

00:55:07: by now I think it's really funny that it's like this race between the models and then everybody saying

00:55:11: ah did you guys see Kling 3.0? It's the end of everybody else. This is now, you know,

00:55:18: the paradigm shift. Boom. And then day after tomorrow, uh, the other new model comes out and

00:55:24: then everybody jumps on that and say, "Oh, this is going to kill everybody else." It's like it's

00:55:29: a race between all the models out there and yeah, every other new release is then determined as

00:55:37: this changes everything. It just goes incredibly fast. What comes out in the end is yeah

00:55:46: with each day more perfect. Yeah. And it's just breathtakingly fast how it evolves. You

00:55:54: almost can't keep up. So it's kind of an ongoing paradigm shift. Yes. More or less. Yes. Yes. On

00:56:00: a daily basis. Okay. That actually makes the answer to the next question kind of impossible.

00:56:04: But what is the greatest potential or where do you think is this all going and where are we in

00:56:08: three years? Ah, what is the greatest potential? The greatest potential. Um I think, and when

00:56:19: I get this question I always say let's use AI for stuff that wouldn't be possible without

00:56:29: it. Right? I know my supermarket in the lake. Of course it could have been done in 3D, would have

00:56:35: been much too expensive. I'm doing another one now that I really can't talk about yet, but that's

00:56:43: also a film with an overall main idea that is so visually big and epic that you also

00:56:54: almost wouldn't be able to do it any other way. Right? So I would say I'm hoping for a future

00:57:02: where AI is also encouraging the creatives and the clients in the very beginning to think

00:57:10: of stories that we can only tell with AI, right. So as an accelerator of new ideas and also to

00:57:20: being brave, you know. So let's, I mean for me it just opened a door to things that would almost be

00:57:31: not possible without it. And this is what I love about it, right? Also if it comes to a small

00:57:37: film, an emotional story between two people, of course, I want to shoot it with great

00:57:43: actors and stuff. That's my hope more or less. Very good actually, kind of also maybe already

00:57:52: a finishing statement because I would also have asked the question, if you could, what's the

00:57:58: other three key takeaways or advices for people in the industry but this is actually already it

00:58:04: I think. So um or is there anything else you would say, what would you tell people to do now, you know,

00:58:10: what's the advice? I would say, you know, it's getting cheesy but still

00:58:19: I would say let's not be afraid of what's coming but let's embrace it and let us yeah be able to

00:58:31: do what we love to do in an even bigger and more epic way. Um but let's at the same time please not

00:58:39: lose everything we had in the last 50 years. Film crews and awesome people working together. This is

00:58:46: still for me one of the yeah, for me personally one of the most important things in creating a

00:58:51: film is that there are different creative people coming together and creating something new

00:58:57: and this for me, this has always been the magic of film and I really hope that this we yeah

00:59:06: can remain at some point in the future. Yeah, that's the magic everybody always talks about

00:59:13: on set which is maybe difficult to find in a right and then um yeah for, if you now that you're saying

00:59:21: it on set, for example what I do on every shoot I do right, no matter if it's a small shoot in

00:59:27: Berlin or a big production in Cape Town, first shooting day, everybody is, we're ready to shoot.

00:59:33: The shooting plan says shooting now. Then I gather everybody that works on the set today. Everybody,

00:59:39: every runner, yes, also catering, everybody that works here, please have a circle around me. And

00:59:46: then I say hi. I introduce myself. I introduce the client and the agency. So all the decision

00:59:52: makers on set. And then I quickly tell everybody the story of what we are shooting. Even because go

00:59:58: on a film set in Cape Town and ask the grip guy, what are you shooting? I don't know, something

01:00:02: with beer. And I always found that so sad. So I quickly tell everybody what we're even shooting,

01:00:09: the story and say thank you that you're all here. Awesome. If anybody has an idea to make it even

01:00:13: better, here I am. Let's go. Everybody walks on set with a smile and after, so many people come to

01:00:20: me and say, so cool, that little talk this morning. Takes 2 minutes, costs zero and makes the film better I think. So

01:00:28: yes I totally agree what you just said. For me, the magic of a good film is all those

01:00:35: people that um yeah made it. Nice. That sums it up. So, what I'm taking away is let's embrace

01:00:43: the possibilities but keep the human factor and people who actually use the tools. Um and then

01:00:50: uh we have a bright future ahead and we have to adjust to it a little bit I think. And uh

01:00:56: but the time would be now. Couldn't have said it better. Thank you very much. Thank you. So,

01:01:02: thanks for being here, Florian. Uh, of course, I will post all your LinkedIns and website

01:01:08: uh into the show notes and also the links to your projects if they're available online, I guess. So,

01:01:14: not all of them yet now, but I will. Yeah, a couple are in the making, but the ones I talked

01:01:19: about, a couple of them are there. Nice one. Nice one. And, uh, of course, you can also find

01:01:22: most of the cases which are out on the commercial content campuss with interviews and stuff. So,

01:01:29: thank you very much. Thanks for having me. Enjoy the time working with AI. Maybe we can do this

01:01:34: one again like in some time to see where we all are. And on the next podcast we're going to talk about

01:01:39: the dark side of AI. About the moment. Dude, I've seen this interview. I've seen an interview

01:01:45: lately with some professor. I had to stop watching because he opened up the other scenarios with

01:01:51: okay and what happens the moment the AI realizes we don't even need the people. This is actually

01:01:57: a point I was talking about yesterday. Yeah, this is okay. Let's do a, let's do an extension on

01:02:02: the after the ending of this one because I had, you know, it's meetings with film people here

01:02:07: in Berlin at the moment. There was some dinners with people and someone told this story where

01:02:11: they actually tested AI and they had two AIs working simultaneously and they spoofed it. So

01:02:18: they made one of the AIs pretend to turn it off. Mhm. And it kind of, I

01:02:24: have to research it, but it kind of regulated the air condition in the room. Yeah. Yeah. So, it's

01:02:29: not the point. There's a couple of things. This is the point where I'm like, okay, we're only talking

01:02:34: about film. Yeah. Yeah. There's another one. And I also don't want to remember it

01:02:39: wrong. But a similar one where also they kind of like tried out what the how the AI would react

01:02:47: and also said something like it's going to be turned off and then the AI lied to save itself. So yes,

01:02:54: there is, let's only talk about advertising please, because yeah we should keep an eye on that.

01:03:00: Humanity should keep an eye on that. But yeah, so um let's do a dark side, or maybe like a

01:03:08: um a followup in some time to see where we are, anytime, if we're still alive by then. I hope

01:03:14: so. I hope so. Florian, thank you very much. Thank you. No, great pleasure. Thanks. Bye.