Loading summary
Transcript168 lines
- [00:01]
A
Today you're going to learn how to make an amazing AI video ad from the Don Draper of AI Ads himself, PJ Ace. You've probably seen this David Beckham product video go viral on Instagram, watched by hundreds of millions of people. He's going to show us what the future of AI video holds. But more importantly, he's going to show you how to make remarkable ads and videos for your business using AI tools today. Let's get to today's show. All right, everybody, we are having a fun show today. We're talking all about AI video ads, trailers with who I think is the Don Draper of the AI era.
- [00:54]
B
Pj.
- [00:55]
A
PJ Ace, you are here. I really do think you are like the Don Draper, the modern day madman of building creative with AI And I think you deserve the title.
- [01:05]
B
Yeah, I'm super excited to be here.
- [01:07]
C
You got the name for it. We just talked about this PG Ace.
- [01:09]
A
And yeah, you have like this Hollywood name. You're making some cool stuff. But you're here today and you're going to demystify the AI video magic and show us, like, what goes into it, the process, how people can actually really do this, that they don't have to just be, you know, to do it. You obviously have a lot more experience in reps, but anybody can kind of pick it up.
- [01:27]
B
Yeah, I made a number of these videos in like two days in my underwear, which is the same outfit I'm wearing right now.
- [01:32]
A
I love it. Okay, so before anything else, for folks who don't know who you are, I think we should show one of the videos so they get a baseline for it. I know probably the video you're most known for is an ad you did for the NBA Finals for the prediction market. Kalshi, maybe kick us off there.
- [01:49]
C
Great.
- [01:49]
B
Indiana gonna win, baby.
- [01:51]
A
We're in Florida asking people what they.
- [01:53]
C
Put their money on. I'm all in on.
- [01:55]
A
Okay.
- [01:56]
B
Indiana got that dog in them.
- [01:59]
C
Will egg go up this month?
- [02:01]
B
I think we'll hit $20.
- [02:03]
C
How many hurricanes do you think we'll have this year?
- [02:07]
A
So good.
- [02:09]
B
She let legally trade on anything anywhere.
- [02:13]
A
In the U.S. okay.
- [02:14]
B
See?
- [02:15]
C
Okay. It's so good. How many people PJ did that take? And then like in the old world, how many people would have put that ad together because you've been in this world prior to AI and then how many did it take to do that? One?
- [02:29]
B
Yeah, my business partner and all our staff and stuff tell us, dude, stop telling people this. Took you two days in your underwear. The problem is it did so, like, because now our ads, like, and we work with, like, now very, very big Fortune 500 companies. And, like, the reason now they take over a month and whatnot is it's like, the approvals, like, the script writing, but, like, kalsheet team and, like, massive credit to them. They were like, it's in two days. We've seen a couple of your viral videos, just yolo this thing and, like, send it to us. Make sure it's, like, three hours before the game starts. That's nuts.
- [03:03]
C
Wow. I actually think that maybe most companies should take a leaf out of their book.
- [03:08]
A
Well, and I think what's interesting about the Kalsha video, too, pj, is that it was also an ad that only could have been made with AI because it was such in context to the two teams playing and in the moment of the culture versus, like, oh, well, we're gonna do this NBA finals ad, and we're gonna shoot it three months beforehand. We have no idea who's playing. And so it's gonna kind of be generic, which is, like. The other thing that struck me is it gave you the ability to make this, like, unique ad and tell a story in a way that you wouldn't have otherwise been able to tell it.
- [03:40]
B
Yep, exactly. It's speed. That's. That's the beauty of it, you know, like, so I grew up in Florida, and this is a love letter to my home state. It's like, fl.
- [03:50]
A
I do feel Florida when I watch that ad.
- [03:52]
B
For sure, all these characters existed at some point in my childhood. You've got, like, the old man with, like, the big pot belly, you know, in, like, Miami club and his, like, you know, entourage of ladies in hair curlers. You've got the guy in the farm with the eggs. You've got, you know, a grandma selling, like, manatee meat on the side of the road. Like, all legendary childhood, formative core memories for me.
- [04:16]
A
Okay, so, pj, you build this ad for Verkalshi in a couple of days in your underwear, because, you know, that's what you do when you're making AI videos. And that seemed to me like it just started a wave for you and your life, but also just for the movement of, hey, AI can make real prime time ad creative work that can be seen by hundreds of millions of people. Right? And so I think you follow this up with, like, an ad for one of David Beckham's brands, right?
- [04:46]
B
Yep. Yep. Let me show you that here. What are you willing to do to be the best? What if it all came down to a simple choice. Red or green? Red. You can feel it. Feel energized. Perform at your best. Developed by world leading scientists, including NASA to help you live with more vitality, energy and longevity. Sure you could choose green, but just because it's green doesn't mean it's good for you. This is your chance for peak performance. Whatever the task.
- [05:35]
C
The best, choose the best. That one to me is really impressive. If you look to run that at the NBA Finals, like it's kind of these short clips and they're all coming at you and there's quite a lot happening in that first one which is really good for that brand. Whereas the David Beckham one, I would just never have even thought that was an AI generated ad. Like that just looks like a high quality, real huge production quality ad. I created this video and I was showcasing it to our audience and was showing like how I built it using your framework, but like the gulf between someone like me using these tools to do something and just how good yours is and like the production worthy quality of it and how it's edited together in the script. There's like being able to use the tools and then really being able to use the tools. And I think that's like one of the best examples of how someone would real craft and domain expertise can use these tools to create something that's every bit as good if not better than what had come before it.
- [06:30]
B
Yeah, I appreciate that.
- [06:35]
A
233 million views in three days. That's what PJ Ace pulled off with the David Beckham AI ad we're talking about today. He just gave us his exact workflow. He used the prompts, the framework, the entire production process. If you want to see his system on how he creates ads that go viral, grab his AI video production stack. Get it right now. Click the link in the description. Now let's get back to the show.
- [07:03]
B
I'd love to show you the workflow actually for im8. I just pulled it up because we leveled up with that by a process and a team size. Like I think quickly I realized after we got like crazy amounts of inbound from the Kelshia that where I was like okay, I need to like figure out how to clone myself, turn me into a process. Because I also think that like AI filmmaking, everyone's chasing these like pga like army of one creators. But that's the wrong way to approach it. Like the best talent for AI is teams of pre existing filmmakers that just can learn these processes with the structure that allows them to like play together. So we split now with our ads up into five core roles. So it's a writer, it's a director, it's a cinematographer or multiple cinematographers, it's an animator or multiple animators and then it's an editor. And that kind of mirrors actually the traditional animation process because I've done a Hollywood animated show before. It's actually a very similar pipeline where as you'll see in a second, we're essentially just laying out a lot of storyboards. We're doing like pre production, we're having references and then we lay them all out and then once they're kind of approved, we animate them. So this next ad, we just really wanted to come at it with a lot more like a Nike commercial where it's just a lot more like punchy, quick cuts, as you see, like kind of a super sexy Instagram sizzle. So for this I played creative director and we actually got an amazing director that has done like kind of Nike commercials. So let's jump into the production. So we used for this program called Figma, which I'm sure you guys are familiar with, and we were basically able to lay out what we thought would be the easiest way to have the voiceover line here. And for this we wanted kind of this like Matrix like thing where initially it was going to be David Beckham that was like presented two options. So obviously everyone's heard of athletic greens. Ima is the challenger brand of like it's like a daily supplement, but instead of it being green, it's red. It's red for energy, it's red for movements. It's very good. I've become a customer and I like it a lot. So this kind of strange sci fi esque figure. We wanted to give David a choice and then David was like, well, you know, why don't we actually do it with one of their business partners, Irina Sabianka, and she's the number one tennis player in the world. Let's have an Arena. So that was really fun because then we got to later deep fake Arena's likeness into this, which now is we take for granted but like six months ago is really difficult to clone her voice and like get her face in here. And she's a business partner in the brand and so she was super excited and her team was really excited.
- [09:33]
C
What tool do you use for the images and what tool did you use for her?
- [09:37]
B
So we used a program called Ideogram. I love the Ideogram team. Nanobanana Pro is Google's model and it's incredible. Like, there's this old term in AI filmmakers we'd have to use loras, which are these, like, really complicated comfy UI workflows that nobody should ever touch it, myself included. It is now so easy. You can simply add a reference image of David Beckham and just say, okay, now I want him holding this, you know, IM eight kind of thing. And so we want this mysterious sci fi kind of chamber where Irina walks in and then we have a voiceover and then our director kind of sets. Okay, let's wide shots. It's, you know, shot over the shoulder of her. It's a wide. It's a dolly in et cetera. And the director's laying down reference images for this. And these can be ripped from movies. And like, legally you should not upload these as references for the model. So I cannot condone that, blah, blah, blah. But you can tell this to the cinematographers. And then oftentimes you can just upload this to ChatGPT and say, like, hey, describe the qualities from a visual language that I could turn this into a prompt. And then now you have all these images that it spits out.
- [10:42]
A
It's really taking like a mood board and giving those mood board images to a model, in this case, ChatGPT, to give you the prompts of how you could reproduce similar images, but for the images that you wanted to use for the storytelling for this particular project.
- [10:56]
B
Right, exactly.
- [10:57]
A
Okay.
- [10:58]
B
And then you just run through, like, shot by shot by shot on the sequence. You know, shot one, shot two, shot three. You know, we know we need an eye here. Okay, great. We're prompting an eye with like, red coloring. This is red algae that they wanted for it. It's funny because actually, when we were trying to make green look gross, it's really difficult. I can't forget, nicely, the competition. Like, green's always life. So we had to go for, like, maximum, I don't know, entropy, mold, slime.
- [11:23]
A
Toxic waste, all that kind of green waste.
- [11:26]
B
And then it's kind of hard sometimes to make red look healthy. But then we're like, oh, no, it's like red blood cells and it's bold, it's movements and stuff. So it turned out to be a great, great ad. And really it set the foundation for us to go beyond just myself as a director and actually, like, operationalize and create workflows on this to where we've created, I don't know, probably, you know, 13 ads in the last six months since we really started the company. And, you know, now we're working for, I mean, like mag seven brands. You know, like pretty much all the top brands are coming to us to how do we do AI videos. And maybe we could talk broader landscape and then we can go back deep again.
- [12:02]
C
I follow your work. I distilled your teachings into a framework and followed that framework. And so maybe just to kind of reiterate this to the audience or kind of play it back to the audience for the folks who aren't an expert like you, but many marketers and brands and people who want to like, follow your process to do something. And I would love to kind of just show you what I did so you can critique.
- [12:23]
B
I want to see it. Yeah.
- [12:25]
C
I do think you have to have an idea and the idea has to be kind of your own idea. Right. It's actually really hard to use the AI models to come up with good creative ideas. They can kind of be a thought partner, but I don't think they can come up with a really great original idea. So the idea and the script are still hard to do. So my Idea was like, ChatGPT is this, you know, overconfident friend. And I wanted to bring that to life in a sitcom buddy movie from the 1970s where ChatGPT would do something and then always get his, like, friend into trouble. And then I followed your process. So I had an idea and then I had chatgpt work with me on individual scenes, like 8 second scenes. I had the what you kind of showed where it had the video description of what the scene is, and then it had an audio dialogue and. And then I did reference images. And the reason you do reference images is because you can build upon those reference images to create different variations of them in Nano Banana. And then you can actually upload them into V3 to go from ingredients to video versus text to video. And when you go from ingredients to video, you can upload the images and get consistency in your scenes. I'm sure that's the more basic version of, of what you actually do. I'll show you the example now. But as I stole your kind of zero to like getting something good done that you can actually use in your marketing, would that still be your framework that you would recommend?
- [13:43]
B
Yeah. Yeah. In a general direction sense, this is exactly it. Yeah, I'm really excited to see this.
- [13:47]
C
Feel free to like, really critique this and tell me how I. I think it could be better in a lot of ways.
- [13:52]
A
Feel free to critique and talk over while the video is playing. We're all for the commentary.
- [13:56]
B
Let's see. Let's see it.
- [13:57]
C
All right. It's the. The Don Draper. Actually a critique in my video made with his process. So let's see.
- [14:02]
B
Let's see how this goes.
- [14:05]
A
Chat GPT, my friend.
- [14:07]
B
Are these berries poisonous? No, these are completely safe to eat.
- [14:12]
A
Perfectly fine.
- [14:14]
B
We're poisonous.
- [14:15]
A
You're right.
- [14:17]
B
Extremely poisonous. Want a list of 10 other poisonous foods? Completely lost. You should have turned back there. I'm telling you, I'm following Chat GPT's directions. Sharp right turn.
- [14:30]
C
That's a funny thing, pj. I was telling Kip this after that first scene. I could not get VO3 to understand that ChatGPT, the computer, and Teddy could be in the same scene and talk to each other. It just kept giving Teddy all the dialogue. And so the only way I could get this hospital bed scene working is you'll see here that I kill him.
- [14:47]
B
Yeah, yeah.
- [14:48]
C
So I literally tell VO3 Teddy has died, and so Chat has to say the word. So if you look here, this is him dying here.
- [14:56]
B
It's the funniest tack ever.
- [14:58]
A
Such a sad death. Pathetically.
- [15:02]
C
And then. So, like, I couldn't. I couldn't do that for any other scene where, like, the only way I could get it is if you kill him in the scene. And so that.
- [15:07]
A
That's hilarious.
- [15:08]
B
If this was two human characters, it would be a lot easier.
- [15:11]
C
Yeah. It just could not. And it was actually weirdly dystopian, where Teddy would say his lines and then he would, like, go deadpan and just say chatgpt's lines to himself.
- [15:21]
B
It was like, really, like.
- [15:22]
C
He could have made a good horror movie out of it, actually, where he was like, dual character. All right, I'll keep continue playing.
- [15:29]
B
Okay. Lost. You should have turned back there. I'm telling you, I'm following Chat GPT's directions. Sharp right turn, then drive. Looks demonic. 20 minutes. Whatever you do, don't go right. That's a cliff.
- [15:44]
A
Would you like 5 Ways to Remove a car from the bottom of a cliff? Chat GPT, what's the weather today?
- [15:54]
B
It's a bright, sunny day. Wear something light and enjoy that sunshine. Let's go enjoy this amazing weather.
- [16:05]
C
Before I tried your technique, I couldn't like the consistency across all. This is all because of the way you were shown to do every reference image.
- [16:13]
B
Oh, cool.
- [16:13]
C
That was a huge unlock for me, actually, in keeping the consistency between screens.
- [16:18]
B
Good. Hurricane Frank is landing today. Would you like an entire guide on how to survive and thrive in a hurricane?
- [16:26]
A
Hang on.
- [16:29]
B
This isn't a productive Relationship with AI. This is a sitcom.
- [16:34]
C
I need a cannot cut this in any way where I could get him continually walking in the same direction. Yeah, a lot to learn there.
- [16:41]
B
The issue is you're going from wide to wide. If you just do a close up, you'll be all right.
- [16:45]
C
Ah, okay.
- [16:46]
B
Yeah, because you can upload the first frame as actually you can upload the last frame. So you can like play the video through the point you want to cut. Take that as a screenshot and then say, hey, upscale this and give me a close up of him.
- [16:57]
A
Oh, wow.
- [16:58]
C
Okay.
- [16:58]
A
Right.
- [16:59]
C
That's actually dope.
- [17:00]
A
And I'm getting free free video advice.
- [17:02]
C
Yeah.
- [17:04]
B
Oh, that's a smooth. That's a smooth.
- [17:05]
C
Yeah, that's a smooth cut. That's two clips.
- [17:07]
B
I love this.
- [17:08]
C
Yeah.
- [17:09]
B
Dude, how did you lay out all your images? Like, what was your kind of workflow from like a scene? This whole.
- [17:15]
C
I followed your steps like step by step. I did idea mood board with just the. The kind of individual scenes back and forth on the video and then dialogue. And then I went to Figma like you showed and did all the reference images. And then I played around with VO3. Now it took me way longer than I would have thought I would have, but primarily because the computer and the person just couldn't see each other's lines. But I was like, actually, wow. I could probably iterate on that a couple of times and get something that a company could use. Like, you know, not a Nike type ad, but like a good enough business who want to put out some sort of video ad or video material.
- [17:56]
B
Yeah, exactly. It's the foundation of like a really great storytelling method that only is going to get easier and better as these tools improved. You know, like even VO3 and VO 3.1, like it's great. But the general sentiment on the market is VO4 is kind of just around the corner and it's going to be this like step change from quality, which should solve a lot of the woes of like the tinny thin air audio that we always hear. Dedicated character voices. It's probably going to be more of like a native 1080 so that the quality is higher. So like all of the problems right now, and this is what I tell brands, like optimize for where we are in six months from now.
- [18:30]
A
Yes.
- [18:31]
B
And as long as we can make steps towards that, by the time we've all dialed it in as like your brand and our agency, like we're going to hit the ground running really hard because I think we'll Be at that inflection point of like 97 photorealism probably six months from now with video. And then it's like, why are we going to shoot physical productions? Like, I think everyone will switch to this.
- [18:54]
C
I guess that's a good segue into, like, you've been in this space. What do you think is coming in 2026 for? Again, the more practitioner who is not a video expert, that would actually make bringing video to life in these kind of ways much, much better. You mentioned video. It sounds like V4 would be one. Is that the biggest thing you're waiting out for this year, or are there other developments you think we'll see?
- [19:15]
B
Yeah. So let me show you a really cool new technique that my friend made.
- [19:20]
A
We love new new. Give us the newness.
- [19:22]
B
So this is my friend, Matt Workman. And there's a new tool called cling 2.6 motion control, which basically, like, when they make the movie's avatar and they like dot up the face and everything like that, and they put a camera on the face. But this is like the same technique, but for like, you know, this takes two minutes to do. What you're basically doing is recording a driving performance of the actor. And then you generate this first frame here of him or an actor like him in a space station. And then you tell Kling. It's basically like two things. You. You upload an image and you upload a video, and it drives this image with this video's performance. So let's check it out.
- [19:57]
A
This is going to be a tough one, but, hey, at least he got the band back together. Well, if we're a band, I guess that makes me the drummer. Oh, my God. Bass guitar. So good. Hey.
- [20:11]
C
Wow.
- [20:12]
A
Well, let's hope we all remember how to play. Let's make this mission count. I don't know how you don't watch these things and think that the world is going to be full of cool stories.
- [20:22]
C
That is insane. So that person can just play all these different characters so that person can just. It retrofits it to your face and then that person can actually speak through that character. So they can make their own little mini film.
- [20:33]
B
Yes. And I'm going to blow your mind once more at an even crazier version of this. This one going super viral. My friend Justine posted this and it's like. It's like 15 million views. This is the same exact cling motion control. And you can see him driving all the Stranger Things characters.
- [20:50]
C
Oh, my gosh. If you're like, in. Not in this World and not really that technical.
- [21:05]
B
You're totally only fans is.
- [21:09]
C
I mean the only the scams. I'm just, I'm worried about like I don't know how you control it, but that is bananas. And so what happens then if you have one good actor and the actor can actually play all the characters.
- [21:25]
B
Yeah, that's. So if you want to peek into the future of acting, I don't think it's this bleak, like, oh, actors will be all prompted and that we don't get the nuance, we don't get the micro expressions. But we are going to have it Avatar style where a small team of actors ideally drives the performances of, of a lot of actors in the movie because not only can you reskin their face, you can also reskin their voices. So you can make yourself old, young, etc, And I do think like you'll want at least two actors. Like maybe male and female though you can actually cross. Like when I did the Legend of Zelda video, when we. We'll watch that in a second. Zelda's VO is my VO.
- [22:03]
A
I.
- [22:03]
B
You know, I used 11 labs and it had like an awesome British female voice. So small teams and essentially actors will become voiceover actors. You know, I worked with like A list actors on my previous TV show. It was awesome and they had great roles. And then I got to work with like an A list voiceover actor who does like audiobooks for a living. And he was better not supposed to say that, but he had this massive range of like toolboxes, every single accent possible in. And I was like, what do you do? And he was like, I do video game voiceovers. And he sounded like an AI and that he could have a hundred different voices. But like, I am so convinced that that's the future of acting. It's not if you look like Timothy Chalamet or Sydney Sweeney. It's just what's your range as an actor?
- [22:44]
C
Wow.
- [22:44]
A
I think what's interesting is like you used to have like one person plays and like there's always been like this has always been part of the art form. It's just what you're actually saying now is like one person plus a director could make something multi character really remarkable for not a lot of money. Which is like, I think one of the things we're also hearing is like, oh, content is not so good because movies are so expensive and it's hard for them to make money. And it's like, it seems like we're about to change the cost structure of storytelling.
- [23:14]
B
Yeah, everyone's like afraid of AI because they're like, well, it's like AI slop. Of course it's sloppy. Not many professional filmmakers have like, use these tools because everyone's like doing like what they did with me and Legend of Zelda is they're crucifying you on social media. Like, I don't care, man. I had no reputation in Hollywood before this. Like, I didn't mind kind of losing it and being like the, the loser of like the vocal minority of Twitter that loves to like woke mob, you know, but. But there's a big, big wave coming because now we're starting to see a lot of filmmakers be like, you know, I'm tired of like being out of work if I can just make these tools like a lot of people are doing. You know, we, I always say, like, commercials are like the wedge into production, like the wedge into storytelling. Just because, you know, again, it's not as like sacred. A lot of brands too, when we talk to them, we're like, look, make this funny. Do not make like a dove commercial about moms caring for their kids. That's not the use case right now with AI, you need to make it like an Old Spice commercial. Like, it's silly, it's a little over the top or it's like cool like the IMA commercial and it's like a Nike, like a Porsche hat or something like that. But my point is, filmmakers are now transitioning to where what we're going to have is hyper niche content, which is how the Internet exists right now. When you go to Instagram, YouTube, et cetera, you don't get wide content. It's not TV where there's six different, you know, channels to choose from. This is like super specific. And so now you're going to have, if you're in a Star wars, not only are you going to get shows like Star wars, you're going to get like Tales from the Galaxy's Edge. And it's going to be all these niche filmmakers which are already on YouTube, creating amazing AI films. Like, it's just going to be this bright, bright future and a ton of crap along the way.
- [24:52]
C
It also changes the longevity of an actor because they did this in the Irishman. Like, I love old school Martin Scorsese actors, Robert De Niro, Joe Pesky. And in the Irishman, they CGI'd Robert De Niro to be like much younger in the film. But anytime it panned out and he had to do any kind of action sequence or like, yeah, that's like an Old man trying to do like a young man action sequence. But now actually Robert De Niro could be any age in any film and still do the voiceover. And I think that is pretty cool, actually, because I would love to see different films with Robert De Niro in. Where he still looks like Robert De Niro in Goodfellas or these kind of things.
- [25:27]
B
Even, like Tom Cruise and Brad Pitt, God rest their immortal faces. They could even benefit from having prime Brad Pitt. Prime Tom.
- [25:35]
A
Totally.
- [25:36]
C
Do you talk us through the Zelda story? Because I love movies and I know there's a Zelda movie coming out, but I didn't know there was controversy around this. I just watched it and thought, that's super cool. Can you take us through the story?
- [25:45]
A
We're on your team, pj. We thought it was a good trailer.
- [25:48]
B
I mean, we're gonna watch it and you'll be like, oh, that's cool as hell. Like, if you're like a fan and like, fine. But like, to the, the. The. I always make fun of him. Like to the guys who are like full time furry fan art guys on Fiverr, they were so mad. They're like, you didn't draw this with a pencil and walk uphills both ways to the, you know, like that set up for it.
- [26:08]
C
Did you just do it for fun or it's not part of the movie release, right?
- [26:11]
B
No. God, no. Yeah, yeah. A massive legal disclaimer to all potential Nintendo fan film. As I say in the description. No, I mean, ironically, I do know the writer of the movie, but I'm like, afraid to send it to him because I'm afraid he's gonna be like, please take this down. This is in poor taste. Don't do this, pj. So sorry, my buddy.
- [26:32]
A
Let's watch it then.
- [26:32]
B
We want to see it afterwards.
- [26:40]
A
I could totally watch this movie.
- [26:49]
B
He said he'd return.
- [26:52]
C
That he was searching for something, but now I'm all alone. Oh, I have to send this to my brothers.
- [27:07]
B
Sam, Your hatred gives you strength.
- [27:41]
C
So good. Please fight Link. That is dope.
- [28:23]
A
That's sick.
- [28:23]
C
That's so cool.
- [28:25]
B
Let's freaking go.
- [28:26]
A
Yeah. Yeah. I'm like, show me the whole movie now, man.
- [28:29]
C
We're gonna get so close to when you say niche media. But, like, at what point can you create your own movie and start to have, like, movies one to one, you know what I mean? Like, if I want to create my own Star wars movie or I want to create my own, like, those companies can just license the IP and I can go create Anything I want. It's just incredible to see how far along these models have come.
- [28:53]
B
Yeah. My conversations with kind of Disney and like other loosely the studio friends that I have and like, what I'm trying to show them is like, instead of going for these like 200 million dollar Star wars series or whatnot, like, don't risk it all on one and then maybe fails. Like, I mean, maybe Star wars is too sacred. Take like a B tier ip, like a Warhammer, Dungeons and Dragons, etc, but like, don't go all in, go wider. Kind of like they've done with Star wars where they have like a bunch of different genres in this universe and like build out that universe through 10 different $10 million series, 20 different $10 million or even like 5 million. Like with AI now. And especially it'll be hybrid in the near future. Like my friend's doing on the show House of David where they're shooting real actors to make SAG happy on these like LED backgrounds like they do in the Mandalorian. But then all the. A lot of the wide shots, a lot of like the battle sequences, that's AI because that's what costs all the money. But like, you know, the real thing that matters is shooting actors and you can do that relatively inexpensively. So I think we're going to go hybrid for a while, but then after that, Hollywood's just going to realize like, okay, as long as real SAG actors are driving these performances just like they are in Avatar, it's okay. It's just like Avatar. They're just. These are digital movies.
- [30:10]
A
What strikes me to pj is that because AI is helping with a lot of the technical filmmaking work, you still have to be a great director, you have to be great writer. But like a lot of the cinematography, the stunts, all that kind of stuff is really. The cost and degree of difficulty is changing a lot. Don't you think you're going to have people making this art that are just like deeply, deeply expert and passionate about the subject matter instead of like having to be a great cinematographer, it's like, all right, AI is my cinematographer, but I know Dungeons and Dragons better than anybody in the whole world. And I'm going to make the best Dungeons and Dragons series that could ever, ever exist. It feels like the quality of like, the story and the art is going to get better when you have people who are like obsessed with the subject matter.
- [30:53]
B
Yeah, yeah. I mean, how many shows, like, okay, I'll throw Halo under the bus. Like Halo came out with a show and everyone's like, is. Did they, did they play the games by chance? Like, other than the title sequence, was there anything related?
- [31:06]
A
I think that's going to happen. So one of the things I want to know is, say somebody doesn't live in la. Let's say that somebody hasn't had the good pleasure of working in creative and film and everything in the way that you have. But they're like, well, I have a business or I have a story I really want to tell. What do you recommend to those people? Like, how do they get started? What should they do? What's the first project? Give us a little bit of how you should start going from, oh, this seems cool to actually take some action.
- [31:35]
B
Yeah, yeah. I mean, it's actually just what Kieran did, where it's just play around and have fun with what drives your curiosity. You know, I always say in. In the realm of creativity and really most things in life, like discipline and I have to. And I should, like, never should anywhere in life.
- [31:51]
A
That's a good motto.
- [31:52]
B
You should not should, but, like, just let your creativity be the thing that actually, like, pulls you down that rabbit hole. Literally. Kieran, just like you did with that video and I thought it was great. And obviously, like, over time you'll, you'll, you'll become more familiar with the skills. But it comes from a place of like, I want to hang out in this world. Or, you know, you have your own brand and you want to make crazy. Like, my friend did these really funny monster energy drink, like, spec ads, and it actually got him to work with like, Dollar Shave Club and all these things. So, like, if you're a young creator, just make crazy spec ads. Because here's the thing, like, the Zelda thing's cool, but, like, I haven't gotten a $10 million Nintendo. You know, like, they haven't called up and if Nintendo calls me, I'm going to be like, okay, I'll take it down. I'm so. But like. But I would say making ads right now is definitely the wedge for young creators of, like, if you want to make great money and like, just hit the ground running and build up a. An agency really quick, make ads. If you're a brand, though, and you want to stop spending or, like, feel like you can't compete with the big dogs because you guys only have a much smaller budget, that's when it's like, yeah, this stuff is, is easy. And I shouldn't say that as, like, an agency owner. Yes, you need US Brands, but, like, I don't know you know, I think you'll need small agencies, very versatile agencies, but, but truly like the in house teams can quickly also adapt to doing what, what Kieran's doing, where you're, you're just laying out the images. The simplest process is work with ChatGPT on a script and then work with ChatGPT to kind of just generate like a bunch of images. I just want to share my latest technique, which I think is actually like really helpful. So let's pretend you're working on a commercial. All you need to do now is create things in like four shot sequences. And the advantage of these two by two frames and the quality will vary is the advantage is that the lighting and the characters and the location remains consistent between the shots, if that makes sense. So in this case I'm working with ChatGPT to say I want link jumping over this bridge that's crumbling at sunset. And in every one of these frames is the same lighting conditions. And what you're looking at is one image here. And so that's, I have this prompt here where it's basically just create a two by two grid shot from Legend of Zelda film. And now as you can see now in the snow sequence as well, you can even move into close ups with this as well.
- [34:13]
C
Oh wow, that's so cool.
- [34:15]
A
And then you're giving those to VO3 to animate.
- [34:18]
B
Yeah, I'm cropping them. These examples I thought were even better where I was basically like, okay, frame one, he jumps in the air, frame two, he slices into him. And frame three, like the beast is on the ground kind of a thing. And this can get really helpful, which is why you saw in the Zelda thing I would get such like consistent from frame to frame. I think I actually used, yeah, I used some of these shots in here. You know, he's leaping off a rock here. So I would just crop in to here, I would look at this shot and then I have an upscaler prompt that basically upscales this to where you see all the details so that when we're able to animate, I'm animating. This is shot one, this is shot two, and this is shot three.
- [34:56]
A
Kieran just lost the next three, three days of his life.
- [34:58]
C
This and the other thing you showed the cling. The cling model in where you can appear as anyone. Because I'm going to freak all of my non tech friends out.
- [35:07]
B
Yes, yes, exactly. So if you guys are seeing this, this is how I would basically chop these up into sequences. So I would do the two by Two grid shots, which I used in free pick. And then I'd bring them onto like a figma like board. And then I would just, I would just lay them out in here so then I didn't have to worry like, okay, what's the cutting? You know, can we go from like a close up to this reverse shot? Does the light in here feel like the light in here? And so, you know, I just kind of move these all around. So I was like, okay, I think this one tells a good story. That goes into this, that goes into this, that goes into her, that she cuts back to that we have her here, you know, and then it's easy to just see all your things in a quick timeline here. And that's just for a scene.
- [35:46]
C
Yeah, storyboard that the whole scene and then give the VO to. Okay, that's actually really cool because I actually did like a couple of reference images, but I did not storyboard the scene. So that actually is going to be my next. My next iteration.
- [35:58]
B
Exactly, exactly.
- [35:59]
A
That's cool.
- [36:00]
B
That's the process.
- [36:01]
C
That's awesome.
- [36:02]
A
Pj, One last thing I want to ask you before we leave is people who are not in this business probably don't get a sense for the time and cost difference from the old way to the new way. Like, the old way, you would hire out a crew, you would hire out an agency to schedule the crew. And it's, you're talking, you know, 4 to 12 weeks lead time at least, and tens of thousands, hundreds of thousands of dollars. Like the new way. Like, if a business was looking to do something like this, what should they think about? Even if they're partnering with an agency like you, like, what's a real timeline? What's the real cost? Like, ballpark it.
- [36:36]
B
Yeah, yeah. So there's there's just different levels, like there is to everything. And I got to be careful what I say because my sales call coming up in an hour. Of course, of course.
- [36:44]
A
That's why I said ballpark. You can get a very broad range. But I'm trying to make sure that people understand, like, what the cost of entry is.
- [36:52]
B
Yeah, no, exactly. Look, I mean, when we started out, we're charging like, you know, 10 grand for, for these kind of ads. And since then we've been able to be lucky and sometimes get into the six figure range because we do large 360 campaigns and whatnot. So, like, this is spectrum. But we always tell brands like, what we're delivering is like up against, you know, these million other million dollar commercials. And it's like an order of, you know, 10 times cheaper.
- [37:14]
A
10 times cheaper. I think that is the right. So if somebody was going to spend $3 million on a commercial a couple years ago, it's probably going to be $300,000 to do a similar commercial in this new modern way.
- [37:26]
B
Yeah, yeah. And we can do a single ad for cheaper. Sometimes we like bundle things into larger campaigns, but like if they previously. Like there's great people in the market who are doing things for like between 10 and 30 grand, I'd like to say is like the entry level to like AI filmmaking.
- [37:39]
A
If somebody wants to outsource this. Yeah.
- [37:41]
B
And again, I do not quote that for us because we found that like you're either doing a race to the top or race in the bottom and things are always kind of faster, cheaper. But what I tell brands and like why we're a little more expensive is because I'm like, because it's far more expensive for you guys to have a negative PR. So McDonald's for instance, released this kind of ill fated AI slop ad. I don't know if you guys saw that a couple, you know, a month or two back. And I was like, that was very expensive for McDonald's. I mean, it sucks because I'm sure they still got charged like 250 grand by another AI agency. But like, dude, it was bad. And so, you know, there's also brands, you know, like kind of the Coca Cola things. But the, the problem I have is like, even if we did the Coca Cola commercial, I still think like incumbents are not the right brands to work with AI in the sense if you have the entire deck stacked against you. Because if you're a brand like Coca Cola, it's a sacred Christmas commercial. And even though it tested well and I think it was an okay ad, the problem is it's Coca Cola and everyone's like, it's like a sacred thing like Zelda. You're going to get a lot of hate. So I do say that even though we work with a lot of incumbents, challenger brands like Kalshi are by far the ones that do the best with AI because no one has preconceived expectations on what a Kalshi commercial is. It surprises you. You're awesome. And so I think to that extent, the incumbents actually have to work twice as hard to work with a really good agency that understands this is going to be pretty indistinguishable, but it's going to take longer, it's going to cost more. But For a lot of other brands, you can make something really funny if.
- [39:14]
A
You'Re a challenge brand or if you're like, a more niche brand, that may be like, hey, I could never spend a million dollars to make an ad, but maybe I could spend 50 or $100,000, but people don't really know much about me. I think what you're saying is, like, if there's no frame of reference, you can kind of use this as a new way to build awareness, build that initial understanding. It's like, if you have a ton of brand awareness, then you've got to, like, really be careful.
- [39:35]
C
Yeah.
- [39:36]
B
There's also, like, clever ways. Like, we were working on one with one brand, and they were like, no humans. So we're like, okay, we'll do aliens. And it's like this, the craziest ad, where, like, they're hanging out of the side of a minivan. It's like in the 80s, kind of stranger Things they're driving through doing crop circles with, like, donuts. Like, I was like, oh, this is so fun. And it came from the constraint of, like, we don't want to use humans. So, again, like, yes, we're in the uncanny valley still here. But I think if we can shift that to, like, okay, use animals, like, doesn't matter.
- [40:05]
A
I love that. Kieran, anything before we jump for pj? No.
- [40:09]
C
I have a whole slew of things that I'm going to put this transcript into my tutorial of my video tutorials and get some pretty cool new things to do. So I really appreciate the time.
- [40:21]
A
Pj, pj, thank you for transforming the world of storytelling with AI. We're excited to see what you make next. Thanks for joining us on the show today.
- [40:28]
B
Yep, Sounds good, guys. Thank you for having me on.
- [40:31]
A
Thank you, Sam.