[00:00:00] Zachary Murray: All right, everybody. Uh, welcome to the first fireside chat of the year. I'm super pumped that you're all here. Before we get into it, um, and I announced Rory, um, I would love to know everybody in the chat right now, where is everybody like tuning in from? I want you to just like spam the chat, um, where, where you're coming in from, where you're calling in from.
[00:00:31] Zachary Murray: I'm super curious. Um,
[00:00:37] Zachary Murray: Mexico City, PDX, Nashville, Zach's from Nashville, Toronto. Hey, I'm in Toronto as well. Super curious. Tell me where you were in Toronto. You're at Bogota, Miami, Greece, Bangkok, UK, Pakistan, London, Germany, Vancouver, Dubai. Wow. Okay. This is super cool that we have a bunch of people. South Africa, Amsterdam. Wow, this is crazy.
[00:00:59] Zachary Murray: Um, well, like I said, I'm glad you guys are all here. Um, we're going to be doing like a lot of these fireside chats throughout the year. Um, and so I'm super pumped for all you guys to be joining this and just start to ingest some of the information, um, from, you know, people that I know in the industry that have amazing things to share.
[00:01:17] Zachary Murray: Uh, speaking of that, so in December, um, I was in Bangkok, uh, to speak at Affiliate World. And I met Rory and we were like chatting in the speaker room. And then I saw his presentation and I was like, Holy shit. And I was already planning on doing these fireside chats. I was like, this guy's gotta be our first, uh, fireside chat.
[00:01:36] Zachary Murray: Um, his talk absolutely blew my mind. Um, and so I think there's a ton of value for our audience. People that are creating Facebook ads. I'm sure everybody's played with mid journey or Dolly. Um, but he's really taken these tools and thought of systems and ways of using them. Actually in real business case studies and
[00:01:56] Rory Flynn: use cases.
[00:01:57] Rory Flynn: Um,
[00:01:58] Zachary Murray: and in a way that you can like stay on brand and it doesn't feel like AI and really just like scale up what you can do, uh, with your creative team. Um, and so without further ado, I'm going to bring on Rory to the stage and he'll take it away, uh, throughout the entire chat or sorry, without the entire, throughout the entire presentation.
[00:02:15] Zachary Murray: Um, if you do have any questions, we do have a Q and a session at the end, so make sure to just drop them in the chat at any point. Um, and then we'll save them and we'll, we'll, we'll address them at the end. So, um, yeah, we'll bring Rory on stage.
[00:02:34] Rory Flynn: There we go. What's up, everyone? How are we doing today? Uh, awesome to be here. Zach, thank you again for inviting me. This is, uh, this is really cool to have everyone here from around the globe. I didn't expect that type of, uh, that type of reach here. So. No, without a, without further ado, let's get into it.
[00:02:48] Rory Flynn: Um, you know, one thing I will say, which is, which, you know, is pretty interesting here is that I was, before I gave this presentation in Bangkok, funny enough, I was looking, you know, for ads that I could utilize in the presentation that we're using AI, and I swear to God, I was on like the meta ads library for, I don't know, like two days, just trying to get.
[00:03:10] Rory Flynn: Different things to save for my presentation. I was like, could there just be a damn place to find like all these ads and aggregate them and this is not a paid promotion, I promise. Um, but when I met Zach at the show, I was like, dude, I was, I literally needed your tool like three weeks ago. This would have saved my, saved my life for the last, you know, and my sanity probably.
[00:03:30] Rory Flynn: So funny enough, it was a weird synergy, but. You know, we hit it off backstage and it was, uh, it was great. So, you know, I'm glad to be doing this, but, you know, without further ado, let's just rock and roll here. So let's get started now. Let me put this here. Here we go. All right, cool. So going to jump right into this, right?
[00:03:50] Rory Flynn: I think everyone's here because we know times are changing. AI is dominating the headlines. They're dominating your social feed and you know, the future is generative. But what the hell does that mean? Really? Right. You know, that's what we're going to talk about here today. So it's a little bit about me.
[00:04:04] Rory Flynn: My name is Rory Flynn. I am the founder of Systematic AI, and I am not a designer or a media buyer for that standpoint. But, you know, I really just look at a lot of these tools and solving problems. I mean, that's really at the end of the day, what this is all about. So, you know, what is Systematic?
[00:04:21] Rory Flynn: Essentially, we are an operational AI agency, meaning that we look inside people's businesses, find problems, and then plug them with conventional AI tools so that workflow can be improved. Right. We also do things in the training consulting space. Do a lot of done for you at creative. So number of things that we do, but a lot of times we're working with the, um, We're working for, uh, brands at different e com agencies and different things like that.
[00:04:44] Rory Flynn: So, you know, it wasn't always like this though, like it wasn't always utilizing AI. Now, again, where this all started was I was working for a digital marketing agency. We had about 90 clients. Now that's a ton of clients, a ton of work that has to get out the door. So, you know, if you were to look at kind of the creative needs that we had, we were typically working in the realm of email marketing and, um, probably, you know, more so in paid media as well.
[00:05:11] Rory Flynn: So when I say that 90 clients, typically I'm going to focus on the email side here just to give you a scale of production. We had to do around 10 to 12 email campaigns a month. So if you extrapolate that out, it's around. 12, 000 plus emails per year. Now we were only working with five designers at the time.
[00:05:27] Rory Flynn: So, you know, you start doing the math. It's a lot of work that has to get done. Now, oftentimes if anyone's worked with brands or you are a brand, um, you know, you always have creative needs, right? Like it's always there and it's ever present. And, you know, from an agency standpoint, oftentimes we're working with minimal assets and always like a lack of bandwidth.
[00:05:45] Rory Flynn: Always right? So again, what do we do? We started to utilize a I to amplify our productivity, you know, reduce our costs and generally just create happier clients. Now, what does that look like? And again, who are you? Right? That's really where this all starts. Like, are you a CMO? Are you a media buyer? You an affiliate marketer?
[00:06:03] Rory Flynn: Like you're probably utilizing a I in some way, shape or form at this point. How are you using it? That's like the real question. Like, are you using it for personal tasks? Just doing singular things. You know, are you doing, you know, building sales copy? Are you writing emails with it? So that's cool. If you've made it to automation, congratulations, you're probably very far ahead of majority of the people.
[00:06:25] Rory Flynn: So like, again, we want to consider what comes next after that. Are you building systems? Are we building things to make our workflow a lot easier? And that's what we're going to get into here. So these are the facts though, this is what recent poll done by Boston Consulting Group. They pulled 758. Of their consultants, and they found that those utilizing a I completed 12 percent more tasks.
[00:06:47] Rory Flynn: They were doing those tasks at a 25 percent faster rate, and they were also reporting a 40 percent higher quality of results. So, in a sense, is pretty much, uh, delivering on the promise. Of, you know, generating more and more productivity in reduced time and getting better quality results. So you're probably using AI, but again, at this point, I think where a lot of people are is there's just so much information, right?
[00:07:13] Rory Flynn: There's tool overload, the information, it seemingly never stops. There's a new tool every five seconds and you don't really know what's possible or how to string everything together. But like realistically, if you want to operationalize. ai, if you wanna put it in the business, it starts small. Like it doesn't have to be this big giant, um, you know, operation that you're thinking of stringing thousands of tools together.
[00:07:34] Rory Flynn: Really what you wanna do is identify these small holes in your business, then find tools to plug those holes and then start to really build SOPs and operationalize around it. Right. Now, speaking of that, again, 'cause we're in, you know, we're talking about media buying today and, and paid media, things of that sort.
[00:07:49] Rory Flynn: What are the problems that we're facing, right? It typically is a creative issue, right? Every month, if you're a media buyer or, you know, you're working in the paid media space, you're going to need new ad angles. You're going to need fresh, creative, and you're going to need. Like all the variants, right?
[00:08:03] Rory Flynn: Like all the data that we get, it's only going to make our ads better. Now, the average media buyer, what are we dealing with? Right? Fatigued assets, a small variety of assets. And typically the static is just subpar. Like that's what we're going to focus on today because static tends to be a, like a very shrinking piece of the marketplace in terms of the paid media landscape.
[00:08:22] Rory Flynn: And it's really, how can we make this an asset for us instead of just having, you know, Just a couple different images to throw out there and add to see if they work right now static. What's the problem with that? What are we dealing with? Um, oftentimes static is coming in limited quantity. It's time inefficient to get it done.
[00:08:39] Rory Flynn: And it's oftentimes expensive, right? So think about what it takes to get. You know, a photo shoot done, maybe you're a brand, maybe you're an agency, what not, you know, if you're going to be a brand and then you want to go get a photo shoot done, you have to go number one, find a photographer, you have to hire the photographer, you have to set the photo shoot up, you have to then storyboard, then you have to actually run the photo shoot.
[00:08:58] Rory Flynn: Then from there, you got to get the, you know, everything has to get edited. It has to get retouched. There might be revisions. It's gonna be a two month process. And realistically, how many times a year can you do this? So I'm not saying not to do this. I'm saying to supplement. This right now, and this is how we can start to do this now again, let's solve this problem and we'll go through this step by step so everyone can get a good understanding of how this all looks now in the next 30 minutes, going to talk about kind of how to use mid journey to juice your creativity, how to asset hack your brand identity, how to build AI systems for scalable ad gen, and then what that ROI potential looks like.
[00:09:32] Rory Flynn: So first, like, how did we get here? We're like, how did we get to this point? In 2023, we saw basically the explosion of, of AI. Like we're introduced to tools like chat, GPT, mid journey, stable diffusion, Dali, and like, it was novel, right? It was like novel for a minute. It was cool. Um, but like, again, everyone had like their little adoption period and then businesses started to take hold of it.
[00:09:55] Rory Flynn: Right. And like the novelty wore off and it started to become, uh, leveraged. So this is kind of how I see. The standard progression of things happening now, what's going on here is typically like stage one, you find new tools and you start testing them. It's like the first time you found chat GPT and started going in there, writing a couple of prompts, right?
[00:10:14] Rory Flynn: Stage two is you start to use them daily in your, you know, maybe it's in your personal life. Maybe it's for fun things, but again, you start to utilize these tools as become becoming almost bionic, right? Like starting to use them, you know, as a, as a crutch or, you know, as an amplifier in your day to day activity.
[00:10:29] Rory Flynn: Now, stage three is where it starts to get interesting, right? This is basically when everyone in the organization, maybe, you know, you're in an agency of 25 people and maybe 25 people are using chat GPT, but they're all using it a different way. There's no systems, there's no like SOPs, there's no operations around it.
[00:10:45] Rory Flynn: Like, that's where we're getting. Um, now, again, once you get to the next stage, which is stage 4, this is when we start to build SOPs. And we actually start to use multi tool systems. So we start to string these tools together. There's actually an organization around it. It's not just Everyone doing their own kind of thing and getting to the same end products, right?
[00:11:04] Rory Flynn: You know, closely this year, I think you're going to see a lot of this happening now at stage five, where these systems become customized and tailored to build, train, build and train custom models off of them. So these systems. You know, once they're tested, once everything starts to become more illuminated, right, of how this all can work and what's where the system where the flaws in the system break and where they don't, we can start to build custom models on this so that you can run autonomously certain aspects of your business.
[00:11:28] Rory Flynn: Now, again, that's really, uh, that's really kind of pie in the sky. I think at this point for a lot of people. Majority are probably in stage two or three right now. You're either, you know, using these in daily, or you're just starting to get more operationalized with them. Um, people are solving problems, but how do we know, like this is starting to happen, like, how do we know this is really, uh, this is really going on.
[00:11:50] Rory Flynn: Maybe you've seen ads like this. This is where foreplay would have come in, you know, a lot of handy, you know, for me trying to pull this stuff. But, you know, maybe you've seen ads like this on your feet, right? This is, this is suit supply. Now looking at this, I can tell you with 100 percent confidence that these are AI.
[00:12:06] Rory Flynn: Now are the images of the models AI? Maybe not, probably not. But is the background AI? 100%, right? So again, when you start thinking about this, it doesn't have to be this full system, which I'm going to talk about here. Really? It could be something as simple as changing a background. Now, again, with the background, with these flowers, right, this could be.
[00:12:25] Rory Flynn: Just one image, but you could also change that to 50 other images, right? Same model, different image. So now you're, now you turn that one asset into 50, right? You can test that and then you can get different data and results from it. So the difficulty on this, like specific use case is very low, but the effectiveness level is potentially very high.
[00:12:40] Rory Flynn: So again, why are they doing this though? Cause it's on brand, like pretty much that's, that's why they're doing it. Right. It's visually unique. It's probably going to stop your scroll when you see this in your, you know, meta feed or, you know, on any other platform, but it's scalable. So again, with this, you can generate, like I said, 50 new backgrounds on this, you have 50 new ads.
[00:13:00] Rory Flynn: But as we see here, even down like the bottom left hand corner, this guy, Thomas, you know, is impressed with the camel. The camel doesn't exist. Right? So again, it's not like we have to be overly realistic with a lot of things. You can, there is some room. For creativity when it comes to stopping the scroll, right?
[00:13:17] Rory Flynn: So again, what's possible now, let's take a look at all of this and we'll start to put this all together. Now, this is essentially what we've been doing in the background. We've been taking, you know, a number of tools, string them together and building ads at scale. Everything in this image is built from, you know, mid journey chat, GPT, and then all the variants are, you know, basically reorganized.
[00:13:39] Rory Flynn: In figma. So this is what's possible. This is what's happening. Now, again, this is where we get into what the results have been. Now, we did a case study. We've done this with a number of brands and I have to be totally candid with everyone. The brands do not want me saying this anymore. They do not want me telling people that we are using AI.
[00:13:57] Rory Flynn: So. I'm using Uber Eats as a placeholder here in regards to one of the other large clients that we've worked with, just so everyone knows, just so I'm being totally candid, this isn't, you know, by any means, like any means fake, it's just, I have to be very careful about what we say. So when I, when we look at this again, why would these, these larger brands come to us?
[00:14:17] Rory Flynn: Why are they interested in utilizing AI? You know, majority of them are mature businesses. They're multinational. They have, you know, a lot of cash and they are running a ton of ads, right? So what their problems are a lot of the times very specific, very similar localization and more assets for testing. So what is localization mean in that sense?
[00:14:37] Rory Flynn: You know, if you're a multinational corporation, you're operating in, uh, you know, different continents across the, across the globe. Essentially you need to have different ads for different locations, right? Because if I'm in the U S and I'm talking to a market in South America or China, right? Highly variable, uh, you know, look and feel for what goes on between those different countries.
[00:15:00] Rory Flynn: Right? Like if you have a picture of an American model and you're sending that off in India might not hit the same, but if you have an Indian model and you're sending that off in India, that might hit a little bit different. It also helps in the vernacular, right? Like I think about this from the U. S.
[00:15:12] Rory Flynn: perspective, right? Like the northern northeastern part of the U. S. I'm in New York speaks completely different than the people in the southern part of the country. So, you know, looking at someone in Charleston, South Carolina versus someone in New York, if you can talk to them. In a very direct tone that looks and feels and sounds like them, that's going to be more effective.
[00:15:30] Rory Flynn: So again, localization has become a big part of why we're doing this too. So asset testing, if you've worked in the media buying space, I don't have to tell you the more assets you have, the better, you know, testing you can do, the better, essentially optimization you could do, the better results you can get.
[00:15:43] Rory Flynn: Now, the solution here was to utilize mid journey and chat GPT. In the creative lift and the initial project we ran for them was 500 fully polished, finished as to see how this went now with a really important piece of this is, and if anyone is doing, you know, AI in their business, right. You want to track this stuff because the end of the day, if it doesn't make a difference, what's the point.
[00:16:03] Rory Flynn: Right. So we track this, we track this meticulously and the goal for us was to really work on the time, uh, variable. That's what we wanted to focus on the time and the volume variable. So the results from this were really interesting because we tracked it so meticulously, but ended up being a 65 percent reduction in cost and time of production.
[00:16:24] Rory Flynn: So clients paying you hourly, clients paying you for your time, essentially saves them money. It saves you money, but you also have extended scale. Uh, to bring on more clients from a from an operational perspective, you can now bring on more work, you know, basically turn your arm army of let's just say five or 10 people into an army of 50 or 100.
[00:16:43] Rory Flynn: That's kind of the idea and that's where everything's going. Now, this is as simple as that process gets and as it looks right. Basically we take an image in mid journey, put it into, we build the templates in Figma. We generate the copy in chat, GBT, and then we put it all together, right? Like this is the simplest flow of the entire thing.
[00:17:00] Rory Flynn: So you can see how it comes together. It's not really that complex, but it's stringing the tools together is where it all happens. Now, how can you do this? Let's get into this. We'll go again. Like I said, we'll go through this step by step look at things very, very granularly so you guys can actually get out there and go do this.
[00:17:14] Rory Flynn: Now, creating unique ads, right? It's going to start if you have brand assets, but it's also going to start with mid journey right now, again, the things that are the, you need to get good at this tool are just understanding the tool. Like that's, what's going to make you really solid. At getting to the end results, just understanding the functionality of the tool, understanding what works, understanding what doesn't when that's the case, you can really prompt it will.
[00:17:38] Rory Flynn: And the faster you get it prompting, the more speed is what we're trying to acquire here instead of just tinkering. So you can get from point A to point B. In the most efficient way possible. Now, if you're not familiar with mid journey, it's essentially a powerful image generator. Basically you type in some texts, get an image right now.
[00:17:56] Rory Flynn: That is, it's a really creative tool and it's probably the best of AI right now. There's a lot of, there's a lot more image generators. Out there, and there's a lot of, you know, different tools that you can utilize personally. This one's pretty good to me because of the mix of creativity and ease of use.
[00:18:11] Rory Flynn: So I think it's a good tool for everyone to kind of check out if you haven't, because it's really the backbone of what we're doing here. Now, bringing down mid journey, just so we can get, you know, and start this stuff, right? Like you're going to want to have, if you don't have, you know, a lot of creative assets and you want to create assets from scratch.
[00:18:28] Rory Flynn: Right. Like everything's going to come through mid journey. So really what the good thing was about chat GBT when it started and when it really started taking off is it introduced us to the prompt, like what that is, how it works, um, you know, the thing about mid journey is the art of the actual images are in the prompt, right?
[00:18:44] Rory Flynn: Like it all comes down to how you actually structure these prompts and get this stuff out there. So this formula is kind of what I live by when it comes to this. Very generic, but at the same time, just think about it in these terms, like clear and direct prompts always are going to equal clear and direct output and ambiguous prompts are going to equal ambiguous output.
[00:19:01] Rory Flynn: What does that, what does that mean? Right? Clear and direct prompts. Basically, you're just being very meticulous and detailed. And if you are, mid journey is going to follow you. If you're ambiguous, basically you're leaving it up to mid journey to kind of paint the picture for you. Right. Like, and you're giving them a lot of creative liberty.
[00:19:17] Rory Flynn: So we don't want to do that if we're trying to be very specific and on brand. So let's go into how you can kind of do this. Now, the interesting thing about, you know, this whole process is mid journey has gotten way more powerful in regards to photo realistic images over the past. Let's call it six to eight months, but you know, there's a time and place for, you know, spaceships and aliens and, you know, futuristic living rooms, but we really need stuff to look real, right?
[00:19:40] Rory Flynn: If we're going to run this in the marketplace, we need this to look real. So, again, the best way to do that, in my opinion, is to use photographic elements or photography elements because mid mid journey responds to it very well. And it's a really good way to get polished looking images that don't look at that.
[00:19:55] Rory Flynn: A. I sort of like robotic look to them right now. This is typically what I'm doing here, right? Like, I try to. To have like at least a structure when I start prompting, so I have all these elements broken down. And again, I'll run you through these basically what we're doing here. Essentially, this is the, I always like to start with the subject and the action who's in it.
[00:20:15] Rory Flynn: What are they doing? Or again? What is it? And what are they doing? Um, the environment where it's taking place, the composition and shot type. It's gets more important as you want to get more on brand. Um, you know, the mood and emotion, always good to add into your, into your prompts. It gives it a little bit more feel and life to it.
[00:20:31] Rory Flynn: Specific cameras and lenses. This is something that I like to use. Some people don't, regardless, it's essentially like triggers mid journey to a photo realistic mindset when you use a type of camera or a type of lens, it doesn't have to be a specific camera brand. That could be something like 35 millimeter film or 50 millimeter lens or whatnot.
[00:20:50] Rory Flynn: Just a generic look and feel. There's a lot of things in photography that can alter the image. Now, film stock, lighting, color scheme, details and modifiers, right? Like these are all good things. To add to your prompts. Now, when I say this, you don't have to use all of these in every single prompt. This is a good structure, a good guideline on what you need in there, right?
[00:21:08] Rory Flynn: To like actually get the most out of it and control it to the max. Now over here on the one side of the image, the mid journey cheat sheet, I'm going to link this on the backend, basically a lot of functionality, you know, of the tool in here so that you can just have it all in one place. Licking that at the backend, you know, if you guys need this to utilize it.
[00:21:23] Rory Flynn: So this is like a basic prompt formula that I like to use, right? And like, this is how I want us to start visualizing how this stuff works. So typically, you know, what we're doing here is we're using phrases or short, very visceral terms to get from, you know, one piece of the prompt to the next piece of the prompt, you know, long sentences, they can work.
[00:21:43] Rory Flynn: But at the same time, you know, again, that's just more words that the computer is picking, or the robots are picking up on, and it could just jumble the vision. So I try to keep it like short and punchy, but typically like to put the photo type, like to put the subject in action in their environment, color scheme, camera, lens, lighting, composition, additional details on the parameters, right?
[00:22:02] Rory Flynn: Typically how I like to structure it, because this is very simple. It kind of gives a mid journey, exactly what you wanted, uh, what you want, or what you want to get out of it. Now, again, this is for more visual people, right? Like, think about it like this, right? Like all these, all these little subjects here on the bottom, those are just going to generate your image.
[00:22:19] Rory Flynn: So your words are constructing. Your image. Now, we talk about prompt structure. Also, this is very important because mid journey likes to read their prompts, I think from left to right, right? Like whatever's in the front of the prompt is going to be weighted heavier than what's at the end. So that meaning if you need to go and you're trying to get, you know, you're trying to get an image out of this and it's not working the way that you want to.
[00:22:43] Rory Flynn: Sometimes it's not. The prompt itself, sometimes the way you structure it. So again, whatever's most important to you, put it first. So if the lighting is the most important thing to you, put it first. If the subject is most important, put it first. If the environment is most important, put it first. You know, then have it kind of on a sliding scale, on most important to least important.
[00:23:01] Rory Flynn: It's probably the best way to think about it. So, I'm going to show you a little example here of what I'm talking about. Like, this is a, this is a typical, like, very simple prompt that I like to show, because it comes out pretty, you know, pretty consistent every time. So, this here, you know, we're using a Land Rover Defender, Vibra and Cusco, Peru, so we have the environment.
[00:23:18] Rory Flynn: Freeze motion, which is essentially like a, you know, stopping an item in, in action. That's like kind of a composition or balance term, asymmetrical composition, meaning that everything's not going to be totally symmetrical and dead centered in the image, um, you know, 35 millimeter. That's the type of lens we're using beggar, beggar, Pankrow 400.
[00:23:36] Rory Flynn: That's a film stock. It's going to make things a little bit more moody. Extreme tonal balance is essential, like a professional photography. Uh, term, right? Like that's the difference between light and dark elements in a photo. It just looks really professional. It gives it a little bit more pop. And then street photography is like more candid style of photography.
[00:23:53] Rory Flynn: So that's kind of what we're going for, right? We're taking all those little building blocks and building up to this image. So you can see over here, this is what we got. And now what happens here is a basically run a little experiment talking about prompt structure, right? On the first image, you can see what happens when you just run that prompt verbatim.
[00:24:10] Rory Flynn: That's what you get, right? Like it's basically what we asked for. The second prompt, all I did was move Vibrant Cusco Peru to first. So I moved Vibrant Cusco Peru to the first element of the prompt, then Land Rover Defender, then the rest of the prompt, right? So, so you can see what happened there was, in Vibrant Cusco Peru, basically, you got a lot more vibrant, And you got more Cusco.
[00:24:32] Rory Flynn: The truck is basically blending in with the environment itself, right? Like, it's a little bit more focused on the environment. It's more pushed back. So the environment became the focus because we put it first. Same thing on the extreme tonal balance, right? Difference between light and dark elements. It's obviously happening in this photo.
[00:24:49] Rory Flynn: I put that first in the prompt. And then BurgerPancro 400. Again, a little bit more of a moody film stock, right? As you can see, again, that takes precedent here. So again, it might not be your prompt that sucks. It just might be the structure. So you don't have to go and re re shift and rewrite a million different things.
[00:25:06] Rory Flynn: So let's go. And this is basically just a variety of different images that can come out of this, right? Like it's a very versatile tool. Basically, you can go in any direction you want, and I'm really focusing on the photorealistic elements of this too, you know, I mean, obviously you can go down into the animation route, you can go into the, you know, 3d or artistic looking route, but again, if we want, you know, stuff for ads, majority of the time, this is what we're looking for.
[00:25:28] Rory Flynn: So now let's get into, let's get into asset hacking. If you're not familiar with this, basically what it is is we're going to reverse engineer. Images. So this works really well. If you have strong, like strong brand imagery, think of brands like Apple, think of brands like Red Bull, think of brands, you know, Nike, things like that have a very strong visual signature.
[00:25:50] Rory Flynn: Now, why do we need to asset hack? Because we need consistent imagery to look like that. So again, this is really, you know, you can get really deep, but again, like really the whole goal here is to make sure the thing is ever is relevant. To the brand, um, and a disclaimer, I say this to everyone just as a, you know, just use this on your own brand.
[00:26:08] Rory Flynn: Please don't go and use this on everyone else's. It's not, that's how it's going to be used for. So again, here's the idea again for our visual learners here, when we look at this is essentially the theory of how this works. We're going to take our initial image. We're going to take 1 image and then we're going to turn it into those visual building blocks.
[00:26:25] Rory Flynn: You know, we just mentioned before. So basically that prompt structure and then from that prompt structural people extrapolate that out into thousands of other photos. Okay. Right. So basically we start with one. So we get to a thousand, right? That's how this looks. Now, here's what the process looks like.
[00:26:39] Rory Flynn: So step one, what we're going to do is collect our brand assets. Then step two, we're going to run it through the describe function in mid journey, step three. We're going to fine tune this prompt in chat GPT. Then we're going to go back to mid journey and image prompt. So these all use chat, GBT and mid journey together, and then we're going to iterate.
[00:26:57] Rory Flynn: So let's keep going. Now, first step, right? Organize your assets. I'm using Red Bull as an example here. They have really iconic imagery. It's a little bit more difficult too. So I wanted to show you how this could be done. Um, you know, they all have a similar aesthetic as you can see. Basically, you know, a lot of the same color scheme, different composition, a lot of good saturation, good, you know, contrast things of that nature.
[00:27:20] Rory Flynn: So this is a good example to show you kind of where you can go. With all of this now, now we're going to, when we go to mid journey, we're going to run this to describe function. Basically what the describe function is, is you give mid journey and image, and it will generate four prompts for you based on that image.
[00:27:37] Rory Flynn: So essentially all you have to do in discord, if you're on discord is type backslash, describe, paste your image, press enter. You'll get four different prompts, right? If you're using the alpha site, now the new site. Um, it's a lot different of a process. It's actually a lot better because you can customize little pieces of it.
[00:27:54] Rory Flynn: So just think about this when you need to actually break down an image into those building blocks that we're talking about, that's how this is going to look. So now let's go run this in the describe function and see what happens. We're using this again, we're going to use this F1 car as the example throughout the rest of the, uh, you know, the asset hacking process here.
[00:28:11] Rory Flynn: So when we run the results, basically we ran those four prompts and this is what we got. Um, you know, if you're looking at this and saying that's awesome, I'm going to say, I'm going to, you know, call BS on that. These are kind of crap, right? Like these don't look real. There's a lot of AI going on way too much like action.
[00:28:26] Rory Flynn: It doesn't look like, you know, it looks like a video game. This doesn't look like, you know, a real picture. So, you know, we can't stop here. We're going to keep going. Now we're going to go into chat GPT to do. So essentially what we're going to have chat GPT do is take the image that we are selecting and then we're going to have it essentially described it to the same functionality as we just did in mid journey, but we want to get it a little bit more tuned in.
[00:28:49] Rory Flynn: Right? So essentially, here's the prompt that I like to utilize. Typically, I'm doing this in a two step process. You don't have to do it this way. This is just how I like to do it. Um, but, you know, essentially, I'm going to read this fairly verbatim here and explain what, what it means. So essentially, we're going to have it paste the image that we're going to have it analyze the image.
[00:29:07] Rory Flynn: Want to have, you know, want to create a prompt for an AI image generator. So it sets it in that mindset. We want to describe the image like an award winning professional photographer in extreme technical detail. That's what we're really trying to extract from this. Because I can't look at that photo and say, you know, what type of camera it is, what type of, you know, aperture it is, or what type of ISO value or things like that.
[00:29:29] Rory Flynn: ChatGBT is pretty damn good at it. So, again, that's what we want to get out, the things that we can't explain. And then, you know, use this formula to structure the prompt. So that prompt formula that I showed you before, just paste that in there. And then include specific camera lens, uh, camera settings, et cetera.
[00:29:44] Rory Flynn: Be very technically specific. Again, we want to hit on that again. And then use short and powerful keywords and phrases. Do not use full sentences. So I always like to frame it in that sense. But also, now if you want to take this a step further, typically there's some testing that goes on between these two.
[00:29:58] Rory Flynn: Really, you can have it blend with one of the described prompts that was close. So if you take the prompt that mid journey generated, and then you blend it with the prompt that chat GPT, uh, created. Now we have like a really solid foundation, you know, for building these images out. So let's see what this did.
[00:30:13] Rory Flynn: Now, this is what the new prompt looks like. This is what was generated. So we have a panning shot, Red Bull racing car, dark indigo and amber. So we have the color scheme hillside. So we have, you know, essentially like, you know, the environment. You have these technical details from the camera, dappled sunlight, bold contrast, delicate high precision speed, uh, high speed capture, low ISO, wide aperture, right?
[00:30:34] Rory Flynn: Like I would never be able to come up with that. Like I wouldn't know what that was going to look like, right? So that would just be a total shot in the dark. So this got me there in about 10 seconds and now we can keep going, right? So when we run this is what happens, right? Essentially we're looking at this.
[00:30:48] Rory Flynn: So here's your F1 car on the left. Here's your new prompts on the right. Some people say it's good. I still say not there yet. Right? Well, you can, you can tell it's not exactly what we're looking for. It's very photo realistic in regards to what we just generated before, but it's still not there. So let's keep going.
[00:31:05] Rory Flynn: Now, this is where the most delicate part of the process comes in. It's called image prompting. You're not familiar with it. Basically what it is, is you give mid journey and image and a text prompt. So essentially if the same theory is if you're utilizing chat GPT and you run a bunch of data into it, and then you have it generate off of the data, this is how we're going to do it.
[00:31:26] Rory Flynn: It's the same sort of theory. Typically it gives mid journey some inspiration. It'll make your image output look a lot better. Uh, so this is the one, the one part of the process that really is. Pretty paramount in the asset hacking piece. Now, the other thing is there's a parameter. It's called image weight.
[00:31:43] Rory Flynn: Now, this is something basically lower on the scale. It'll focus more on the text portion of the prompt higher on the scale. It'll focus more towards the image portion of the prompt. We're probably going to want to go higher towards the image, right? That's what we're trying to do here. Cause again, we want to make this look real.
[00:31:59] Rory Flynn: Now, when we keep going here, that's what happens when we image prompt. So this to me, pretty damn good, right? Like we're close here. This looks, the composition's right. The lighting's right. The color scheme is right, but we've just taken that one image of the F1 car and create a very subtle variation of it.
[00:32:18] Rory Flynn: This is the process that starts the whole, you know, building block piece of it. Now, because when we go and we want to iterate, this is how it looks. Because we developed that prompt and we have all these different elements, right? Really, if you want to iterate off of it, you only have to change 1, 2, maybe 3 items.
[00:32:35] Rory Flynn: So think about these like little building blocks here, right? We have the photo type, subject, action, environment, etc. Like we're going to want to keep the color scheme, the camera lens, the film type, the lighting, the composition. Those are going to be important because that's like the visual signature. But all we have to do now is change the subject and the action, or maybe the environment, or maybe the shot type those three.
[00:32:56] Rory Flynn: Now, that's how we can vary this and keep it all looking very similar. So here's what I mean, right? When you're iterating these types of things again, all I had to do instead of, you know, Red Bull racing car, which changes the Red Bull skier, right? Came right out and change the Red Bull motorcycle came right out.
[00:33:12] Rory Flynn: Changes the Red Bull dune buggy came right out because we have the rest of the visual building blocks in place. Now, all we have to do is change one item so that you can vary this at scale pretty quick. Now, that's where we started on, on the one side with the original assets, and that's where we ended. I say pretty damn good, you know, again, something like Red Bull going to be a little bit more complex because they have a lot of, you know, there's a lot of decals and things in nature.
[00:33:35] Rory Flynn: You might have to do some after production on a few of these images, but at the same time, you know, we're, this is how we can take this and rerun it 50 times and get 50 more assets. We can build this out in, into infinite. Essentially. Right. So that's kind of the power of this and where it can go. Now, how do you actually put the whole process together to get from that first image to a thousand ads?
[00:33:59] Rory Flynn: Right? Like, let's say, let's take a look at how this looks now being totally candid again. Uh, you know, I use the, I use the title one prompt equals a thousand ads. Cause you know, it sounds pretty sexy. I mean, it doesn't have to be a thousand ads. It can be 200. It can be 20, 000. It doesn't matter. The same process will work.
[00:34:15] Rory Flynn: So let's just say you wanted to have a thousand different variations of these ads. Really, what you'd need to do is you'd have, you need to have five images, you'd have five headlines and five sub headlines, you have five call to actions, and then you need eight potentially, maybe more aspect ratios, depending on whatever medium.
[00:34:31] Rory Flynn: You're advertising in, if that's Facebook, you know, uh, LinkedIn, whatever. So the time does altogether that it equals a thousand combinations. That's what we're looking for. So the five tools that we're going to use here, essentially, we're going to use mid journey for the images. We're going to use chat, GPT for the copy, and he's Figma to build the ad templates.
[00:34:50] Rory Flynn: We're going to use copy doc, which is a Figma plugin. That's going to become important as we go through this. So remember that one. And then just any sort of spreadsheet or, you know, uh, spreadsheet software. So it could be Google sheets, Excel, whatever. Uh, so what the process looks like and what this whole workflow looks like is essentially you create your templates in Figma, then you create your images in mid journey, then we're going to go to chat GPT, we're going to create our copy.
[00:35:15] Rory Flynn: Then we're going to organize all these images and copy into a spreadsheet. And then we're going to run it all together in Figma and actually have this thing pop out. So basically when we start with templates, a few different things, right? Well, we're using Figma because Figma has plugins, right? Like if you're not familiar with Figma, it's, you know, web based and a desktop mobile or a desktop app.
[00:35:35] Rory Flynn: Uh, it's like, it's a very intuitive version of Photoshop. Um, you know, it's really good. I just think it's a, it's a really solid tool for designers and anyone that's utilizing these, uh, you know, trying to build ads in this way. So, you know, typically what we're doing here when we're going to do scaled, you know, ad creation is we're using our highest performing templates.
[00:35:53] Rory Flynn: Like we're not, we're not screwing around. Right. Like these are the ones that we want to really, you know, amplify. So again, here, the things that are important is you want to build these templates, you want to structure the asset, the aspect ratios accordingly. And you want to add placeholders for things like the headline, the sub headline, the image itself.
[00:36:09] Rory Flynn: So when we link the spreadsheet and Figma together, it knows where to send everything. Right. We'll get into that in a minute. Now generating images. Again, you can either generate images from scratch. You can asset hack your brand. You can use images that you already have. It doesn't, you know, again, it doesn't have to be one way or the other.
[00:36:27] Rory Flynn: It's just a part of the process. Right? So I get typically picking winners or things that you think have the most viability, maybe it's winning stuff that you've done before in the past X, Y, Z. Then we're going to go to chat GPT and we're going to create the copy. Now, this is where I think everyone can put their own spin on it.
[00:36:42] Rory Flynn: Right? Like I think everyone has their own way of generating copy. In chat GPT, we've built custom GPTs for this at this point, but essentially, right, what we're going to do here and the key points that we'd like to hit on to give chat GPT as much data as we possibly can to generate the output that we need from it is typically we're feeding it things about company overview, the brand guidelines, the brand tone of voice, consumer communication guidelines, or how you talk to your customers, you know, brand marketing examples.
[00:37:11] Rory Flynn: If you have a number of ads that have worked in the past. Um, you know, target audience, exactly who that is, product focus, you know, again, what the product is that we're selling, what are the details around it? And then copywriting guidelines. I think that's really important too, is understanding your templates.
[00:37:24] Rory Flynn: Because if you have copywriting guidelines, like the ads I have before, you know, sometimes the chat GBT will spit out, you know, 30 word sentence. Like that's not going to fit on an ad. So typically I'm limiting it saying, you know, we want to get. A headline in less than five words or a sub headline in less than seven words or a call to action in two words or less, so it gives it some firm direction.
[00:37:46] Rory Flynn: It keeps it on task. Now, this is probably the least sexy slide you'll ever see, but probably has the sexiest results. Right? So what we're going to do here is we're going to structure this all in a spreadsheet. So essentially, from left to right, you're going to put your headlines. Your sub headlines, and then your template link so that it knows exactly where to send all these things.
[00:38:10] Rory Flynn: So once this is all in there, you're going to zip up this file. You're going to download this. Of course, you can download as a, as a CSV. You're going to put it in a folder with all of the images attached with the right name, essentially on the spreadsheet as it is saved on your computer. And then you're going to zip it all up.
[00:38:27] Rory Flynn: So once that happens, we can run over to. Um, Figma here, and once we get into Figma, essentially what we're going to do is just highlight all of the different ad templates that we're going to run the plugin called copy doc. I'm going to say this one more time. It's called copy doc. It's a plugin on Figma.
[00:38:45] Rory Flynn: This is the, this is the sauce. This is what makes everything go. So when you do this, you just run a sync in Figma. And what happens is it'll populate all those variants. So just like that one click, everything's in there. Now, again, because it's in Figma, you can go restructure as necessary. If some images are maybe just like, they don't fit the frame.
[00:39:05] Rory Flynn: Well, whatnot, you still have that accessibility in there. So that's where it really starts to get crazy interesting. Now, this is what the finished product looks like. So the first time we did this, we did a thousand ads in less than one hour. So it's possible. And I don't, you know, think about what it would take to copy paste five different subject lines, five different headers, make all the different variable combinations of it.
[00:39:29] Rory Flynn: And again, just do that over and over and over and over and over again. If you're paying a designer, that's just like, I don't want to say it's a waste of their time, but it's kind of a waste of their time too. They can be doing way more high leverage things. So for us, that's how important this can be one hour for a thousand of these.
[00:39:45] Rory Flynn: Now, again, we're doing this over and over and over again. So think about the volume and scale you can produce. So why are we like, why is it important? Like, why do you need a thousand? Cause it's a massive testing opportunity. Realistically, is that if you, you know, if you are running ads at scale, you're going to need to have different assets to test.
[00:40:05] Rory Flynn: It's just a fact. Right. The localization piece, again, we can get super hyper targeted if we want to, if we're going to be very, you know, strategic in our media buying as well. We can match what we're doing from a visual and a copy standpoint to that. The other great thing about Figma is like I said before, you have full control of them.
[00:40:22] Rory Flynn: So there's other, there's other platforms out there like advertising, AI, and things like that. Once you push generate, they're all finished. So if you don't like a bunch of them, you gotta go regenerate them again. These are all now in Figma. There's a full suite of ads for you to play with. And again, like you can just go and edit them if you need to, right?
[00:40:40] Rory Flynn: So it's just a really, really good place to house all this stuff. And also just the functionality piece of it is really impressive. So it's a major time saver, of course. Now considerations, right? We got to talk about this. Um, if you're going to consider doing this, what do you need to know? Right. I like, I don't think this is a replacement for designers or copywriters, right?
[00:40:58] Rory Flynn: Like I think there's still a need for them. They're going to be the best people to utilize for this, especially in building and formatting templates and structuring everything. I think that's a really good place to start. Now, the interesting thing about this whole thing is really like, if you're going to break down the importance of this, I'd say it's about 20 percent of the importance is on the AI, like itself, the creation of the assets and the copy.
[00:41:20] Rory Flynn: The other 80 percent of importance is on the process and the tools. Because if you think about this, it doesn't have to be like, you start with this, the same process that I just didn't get to the end result. It could be, I have a winning ad one ad, I want to make 50 variations of it, use the same process.
[00:41:36] Rory Flynn: I have winning copy. I want to change the images. Same process. I have really good images. I want to change the copy. Same process, right? So you can start to really extrapolate on how this whole thing can turn from just a few different elements or a few different pieces into a thousand different pieces.
[00:41:52] Rory Flynn: Um, and that copy doc and figma is really like the linchpin in all of this, right? So the other two things that I want to that I want to mention here that we've started to play around with now since Bangkok, this is pretty new. Um, Magnific is a tool if you're not familiar with it. I mean, I use it on every image now at this point, basically what it is, it's an AI upscaler, but it's a lot better than a lot of the upscalers.
[00:42:15] Rory Flynn: So you're utilizing pictures of faces and things of that nature. Like a lot of times you'll see on Dali, mid journey, stable diffusion, the faces come out looking like AI, right? Like it's like a very polished skin. Like everything is perfect. Magnificent. Give this a little bit more dynamic. So even just wiping, you know, wiping the face of that AI look, giving it, you know, some, some freckles or giving, you know, some pores or a couple of different hairs.
[00:42:40] Rory Flynn: So it's like, it looks real, like a couple of blemishes. It goes a long way. So I think that's that's one tool that I would definitely recommend. It's very, you know, you can scale up to, I think, eight X now. So even if you're looking at doing something more on the print side or something in a larger format, you can get these images to look, you know, really clean at a really big size.
[00:43:01] Rory Flynn: Now, the other tool we're pushing into here. Is, uh, is runway now, this is, there's a lot of different video editors out there. Animate, you know, AI animators. Um, you know, I tend to like runway cause it has a very specific level of control and you can get very dynamic with it. Um, so consider this while you're, you know, you're looking at this cause that's what's next, right?
[00:43:21] Rory Flynn: Like, as we keep going here, that's what's next. It's obviously we're solving the static problem right now. And that was a volume problem that we were solving. Now we can add motion to it. Now we have video assets as well. We can extrapolate this even further. So subtle motion, you know, I, we've been testing this and it's been fairly effective.
[00:43:38] Rory Flynn: I don't have a lot of data on that in comparison to what our static ads do, because it's so new. Um, but at the same time, that's what's happening. Now, when we look at this, I'm going to pull up, I'm going to see if I don't crash the stream here, I'm going to try to pull up a quick video just so everyone can see this of what this could potentially look like and why this is important.
[00:43:57] Rory Flynn: Again, when we think about stopping the scroll, really like what is different? What is what is someone not seen today? And what is again, eye catching? So let me put this video up real quick. It's four seconds. Um, just so everyone can see sort of how we're envisioning this. So let me pull this up. All right.
[00:44:15] Rory Flynn: I think I'm sharing my screen. Let me go here. Expand this. So I'm gonna run this through twice, but really this image on the right here, this was done with, you know, this is, we took this image straight from, you know, Figma, put it into runway and we use the tool called motion brush just to brush over this, uh, chicken sandwich here.
[00:44:35] Rory Flynn: Just give it a little life, right? Like it looks like someone's actually handing. This chicken sandwich to you now, is it super compelling? Is it going to win an Oscar? No. Is it something that might stop a scroll on the way down your Instagram feed potentially, right? Like that's really all we're trying to do here is how do we just.
[00:44:53] Rory Flynn: Be a little bit different. How do we test these little variables because that's where, you know, as media buyers and people in the paid ad space, you know, that's what's, that's what works. So let me stop sharing here real quick. Now obviously once we get through like this portion of it, what else is next?
[00:45:11] Rory Flynn: Cause we've solved the volume problem, but again, what I'm looking for next is how do we solve the performance problem? Right? So now that we have the assets, how can we create a closed loop? That's where, you know, again, leave you guys with something, you know, along those lines to think about is how do we then tie in?
[00:45:28] Rory Flynn: Basically, the performance aspect of it, so we get the data at the back end and close the loop to the front end to then influence the inputs. Right? So now we're thinking about, like, a full circular system that can operate like that. And that's where I think automated agents are going to come in. Now, this is if you're not familiar with them, essentially, it's like every manual piece of the process that we did here together that can be automated.
[00:45:51] Rory Flynn: So there's agents that can work autonomously. Now you can create, they can do multiple tasks, but they can be very siphoned out. That's kind of where I see this all going. And I think that's what a lot of people are going to have to consider for some of these processes sooner or later, because businesses are going to start to use them.
[00:46:06] Rory Flynn: Once they figure out how to build them, they're going to start to use them. Right. So it's all about staying ahead. All about staying ahead, staying in tune with what's going on. And then I, you know, we're really going to keep, we're really going to keep pushing forward here as 2024 progresses. But this is something that we definitely got to talk about, right, is ethics and legal.
[00:46:22] Rory Flynn: If you're going to do this, right, like if you're going to do this, I have to be, you have to be cognizant of what can happen, right? So always, if you have clients or you have key stakeholders, maybe you're doing this for yourself, whatever, just have to have the conversation to make sure that there is number one, potentially some SOPs around this.
[00:46:38] Rory Flynn: Number two, um, you know, again, just a clear operating procedure for how you're going to go about it. That's always, always going to want to have these conversations. Now, the other thing here is the U. S. copyright. If you're in the U. S., you know, a lot of what this says is basically, if anything is fully generated by AI, it cannot be copyrighted.
[00:46:56] Rory Flynn: So, you know, getting to think about it like stock photography. So whatever you produce, if, unless it's altered by human hand, you know, it's going to be considered stock photography and anyone can use it. So that's one thing to take into consideration. Also take into consideration what data you're feeding into mid journey.
[00:47:10] Rory Flynn: If you have very proprietary stuff or same thing with chat GPT, I wouldn't do that unless you're on an enterprise plan or utilizing another model that can, you know, use the chat GPT back end. That is, you know, your data is protected. You don't want to do that. You don't want to give that stuff away. So again, this is what can happen here.
[00:47:27] Rory Flynn: On the left, I just want everyone to consider this too, right? Like we don't know what mid journey or any of the, you know, image generators were essentially trained on. We don't know what that data is. So basically what I did in this image was I looked up as a very famous image from the eighties. It was in National Geographic.
[00:47:44] Rory Flynn: And what I did was I just looked up the information about it, like what camera it was shot on, like some different technical details, put a prompt together, ran it through Midjourney, came out almost the exact same, right? So as you can see, left and right, I mean, it's, it's fairly, fairly obvious that that was, that image was trained, you know, on the Midjourney library, right?
[00:48:06] Rory Flynn: So again, I didn't use an image prompt for this. This was just straight text prompt. First shot. Right now, even the things that are super, super crazy here about this image is even a little subtle details like the green on her shoulder, like that's pretty immaculate in terms of how close it is, the color of the background.
[00:48:24] Rory Flynn: If you're doing this and you're utilizing mid journey, just know that this can happen. So I think if you want to keep yourself in, you know, basically clear water, the, one of the better things that you can do, number one is reverse image search. A lot of these images that you're going to put out, like almost every, I mean, I would do, I do it for every single one specifically when we're building only five to generate a thousand ads, right?
[00:48:45] Rory Flynn: Like you want to reverse image search. I use tools like Duplo checker. There's a few other ones out there, but that'll basically scan the internet and see what's close. Cause if it's close. Don't use it. Just go and redo the image, you know, a number of different times, change some different elements of the prompt.
[00:48:58] Rory Flynn: You can do that at will. It will only take you another 10 minutes. So, you know, that's one way to do it. The other thing I will always say is as do's and don'ts to tell my team this. Never use artists or, or like, or names, just don't use people's names, right? Like you see this a lot of times on social media where it's like in the style of, uh, you know, in the style of Wes Anderson or in the style of, uh, Michael Bay or in the style of X, Y, Z artists, just don't do that.
[00:49:22] Rory Flynn: Like, you don't need to, that's just opening you up for exposure. If someone's going to go, you know, send a lawsuit, there are lawyers out there that are just reverse image searching all day to see if they find their clients work. So again. Know, know that that's a, that's a possibility. And also if you really want to know what a person's like, you know, composition is, or like how it looks, if you want to know what Mike or Wes Anderson looked like, go use chat GPT, ask what kind of color scheme, what kind of composition, what kind of lighting elements he uses, it'll spit out the same visual building blocks that we just talked about, and then you can go iterate it yourself.
[00:49:54] Rory Flynn: Right. Try to limit the exposure there. Right now, the other thing that's interesting is we're going to see this I think a lot coming up in 2024 is this this whole like ethical AI movement. So Getty has now partnered with runway and Getty has launched its own image generator. Basically trained on all of the Getty images.
[00:50:13] Rory Flynn: So they're, they're copyright protecting, they're managing risk. That's a unique selling proposition for them. Same thing with Runway. Runway is partnering with Getty to utilize their training database. So Runway is going to be able to generate images and videos based on, you know, right. Getty images that aren't essentially.
[00:50:30] Rory Flynn: Uh, stolen, if you want to call it that we, I don't even know the definition because it's so vague in regards to how this all works. So again, these are just things to consider that, you know, them that you consider them before you just go and blast, you know, people with a thousand ads, right? So here's the, here's the real interesting part.
[00:50:46] Rory Flynn: We'll talk about this. Of course, everyone probably wants to know how this looks, but in 114 projects in November on the low end, we saw a 30 percent reduction in hours spent on the client. Now work at sexies on the high end was 85%. Reduction in clients are in time spent on these projects. That's what we're talking about.
[00:51:04] Rory Flynn: And as we've run 100 percent positive CSAT score, like amplified productivity, reduced costs and less time that promise was delivered. So that's essentially why we're doing this and why we're building out these processes, because that's where it leads. And that's, again, for us, turning our small army into a team that's way bigger.
[00:51:24] Rory Flynn: Uh, you know, again, not trying to replace people, trying to make people happier in their jobs and take away a lot of the robot tasks. And give them more creativity and flexibility in their workflows. So with that being said, this is a little gift. Um, this is a, you know, QR code. It'll take you to a notion doc.
[00:51:41] Rory Flynn: I put a couple of PDFs in there, beginner guides on mid journey. Um, you know, cheat sheets, things like that. Uh, just some additional prompt formulas and resources for you guys to utilize. Should be super helpful. Um, you know, if you need to, or have any questions, I'm always available. Uh, you know, find me on LinkedIn, find me on Twitter, always down to chat.
[00:52:00] Rory Flynn: Stuff is, this stuff is super fun and I was, you know, it's just going to keep, she's going to keep exploding. So, you know, interested to see where it goes, but, uh, you know, hope this was interesting and gave some people some context, but, uh, yeah. Appreciate it guys.
[00:52:18] Rory Flynn: Oops, Zach, I think you're, uh, I think you're muted. I'm a hundred percent
[00:52:22] Zachary Murray: muted. Um, yeah, what I was saying is like, I've watched that talk before, but every time I'm like, Oh, I need to keep leveling up my prompts. It was like, definitely like a really good refresher. Um, so thanks to that. And we got a bunch of Q and a, that came in as you're presenting.
[00:52:34] Zachary Murray: So if you're cool with it, like rifle through this. Um, and try to get through as many of these as possible. Uh, so the first one's from Peter and it's like, he's asking you how do you charge clients for AI work?
[00:52:45] Rory Flynn: Um,
[00:52:46] Zachary Murray: essentially, like, is it on a per asset basis? Like are you a retainer? Like you just fit it into another service offering.
[00:52:52] Zachary Murray: What does that look like? And is it, is it different
[00:52:54] Rory Flynn: than. It's all, it's all variable, I think, based on the client and what they need. Right. Like a lot of times it's not just, you know, ad creative that's going on that what we're doing. So it kind of builds into it. We tend to have standard packages, but looking at it really like, I think, you know, depending on how you like to price.
[00:53:10] Rory Flynn: And I think, you know, depending on what the conversation is like with them, because you are going to produce higher volume, right? Like at the end of the day, that's what the output is. Um, you know, you might not produce the same level of, I don't want to say quality. That's the wrong way to say it. But, you know, if you're going to produce really great looking assets and you produce five X of them for them in the same amount of time, you know, I think it's super, super valuable because you have, you know, the ability to affect performance.
[00:53:34] Rory Flynn: So again, it comes down to sales and offering tactic there. Um, you know, I think everyone's different, but that's just typically how we look at it. Sweet. Next
[00:53:42] Zachary Murray: one's from Bobby, and he's wondering if you have any templates for the Figma
[00:53:45] Rory Flynn: part. Uh, templates for the Figma part. I have a few. I haven't system, I haven't like really built out an operation for it yet, but can definitely send over, you know, if you want to hit me on LinkedIn or whatnot, I can send you a few of these are pretty simple.
[00:53:58] Rory Flynn: Um, you know, again, stuff that's going to work really well with this process. Okay,
[00:54:01] Zachary Murray: cool. The next one's from Adrian. Um, he's wondering that like, you know, if like a product image just can't be generated by mid journey, um, you know, what's the workflow for that? He's usually photoshopping and manually, you know, it's pretty time consuming.
[00:54:14] Zachary Murray: You know, how do you, how are you approaching that? That issue?
[00:54:17] Rory Flynn: Got it. So here's, here's a, another, um, little quick tool that you can utilize like this works really well with lifestyle photography. Um, you know, I would say if you have product images that you want to utilize, And you have labels and you have logos and things like that.
[00:54:33] Rory Flynn: There's a new tool called, I'm probably going to butcher the name. It's called creates like Q R E A T E S dot com. What it is essentially is if you have PNG files or you have different files of your, um, of your product, you can put it into crates, crates, however you say it, it'll generate almost like a 3d model and you can move it around an image.
[00:54:56] Rory Flynn: You have to generate the background. So if you generate a background in mid journey, you generate a background in Dali, whatever it might be. It'll take that image, you can move it around, place it, and then it'll adjust for lighting, tone, uh, shadows, things like that, because if you try to put PNGs in Photoshop, you know, sometimes you'll get a can that's like overexposed and the background is dark and moody, it doesn't look right, you know, so it adjusts for that, so it's a good tool.
[00:55:18] Rory Flynn: Cool.
[00:55:19] Zachary Murray: Uh, Brian wants to know what's the hesitation with brands not wanting to broadcast the use of AI?
[00:55:25] Rory Flynn: Great question. Um, you know, I think, I think we're at this place where it's still seen as. Like we're tricking people, right? I think that's why I try to talk a lot about just education around it in general, because people now, when you like, I can look at an image and say like, that's the journey or that's Dali.
[00:55:42] Rory Flynn: Like you can look at it cause I think it's going to get a little bit more, they're worried about the downside and the PR aspect, right? Like you see all these fake images coming out of people. Now you see how things go on social, there's always going to be bad actors. So I think larger corporations in a sense that have, you know, there's very high or very low risk tolerance.
[00:56:01] Rory Flynn: Um, you know, they're just trying to stay in their own lane and not try to broadcast exactly what they're doing. Right.
[00:56:07] Zachary Murray: Um, next one's from Joshy. He wants to know, so he's like, maybe he has a, you know, a sample product, like a book or something. Uh, he wants to figure out how to, you know, create mid journey or AI generated ads with people holding the book.
[00:56:18] Zachary Murray: You know, maybe that book on the shelf amongst other books. Is there any prompts or, or, you know, the kind of workflow for,
[00:56:24] Rory Flynn: for that use case? Yeah, so you can, you can definitely do that. There's another tool called flare. ai. Yeah. So yeah, for, for hands and stuff like that. If you have a PNG of your book, like you can, you can add the hand in there yourself, you can use hand templates.
[00:56:40] Rory Flynn: Like if you wanted that, that tool crates is going to be, crates is going to be good for, um, you know, doing anything with anything with writing or text on it. That again, that might get screwed up within going processing through mid journey. These other additional tools can be done in this, excuse me, in the same way.
[00:56:55] Rory Flynn: Cause the end of the day, we're trying to get a couple images, right? We're not trying to get a thousand images. You know, we need 10, right? Like that's our ad sit. We can get a thousand variations. So utilizing stuff like that can be super useful.
[00:57:06] Zachary Murray: Uh, Nancy, Nancy was also juice in the chat. So just shout out to Nancy for like, that was super helpful.
[00:57:13] Zachary Murray: Um, she said, I'm sure you can fly through these steps. Still curious. Is there an average length of time? That you feel like you're spending on
[00:57:19] Rory Flynn: each, on each image. Um, it depends like utilizing chat, GBT and describe function to get from one image to a prompt, like that shouldn't take super long, right? Like that should maybe take five minutes.
[00:57:32] Rory Flynn: But then again, when you're going through the image prompting process, I've gotten it to the point now where I can do it really quick. The average person, you can do it in, you know, less than 20 minutes. I mean, if you're just generating and you're changing up little aspects of the prompts or you're adjusting the image weight, like it doesn't take that long.
[00:57:49] Rory Flynn: Um, it's just like getting it to the point of where you're comfortable. With it looking like your brand, I think that's where the subjectivity comes in. It's not, not as objective. So that's where you could probably spend more time because you're just like, is it there yet? You know, so yeah, that's, uh, that's kind of my thoughts on that.
[00:58:04] Zachary Murray: Cool. Uh, I actually wants to know if they can use Canva.
[00:58:08] Rory Flynn: Canva for, you know, I'm assuming
[00:58:11] Zachary Murray: it's like, I know KM has an AI feature. Um, yeah, like with it. I mean, I, I personally, I'm, I'm a big fan of the journey. So
[00:58:18] Rory Flynn: yeah, they. You can't do, you can't build the templates in Canva, like you can in Figma. So, I mean, you can't have that copy doc plugin that links and then have it all populate.
[00:58:29] Rory Flynn: You can certainly build, you know, what you need to in Canva, like, it'll get done. But, you know, from that standpoint, really, Figma is the, the linchpin in all of this, of how you can generate these at scale. So, okay. Um,
[00:58:40] Zachary Murray: Nita wants to know, are we able to upload an image on
[00:58:42] Rory Flynn: chat2BT and ask it to analyze? Oh, yeah.
[00:58:45] Rory Flynn: That's part of, that's the, the second part of the process, right? Like, when we go through. Like how to pull extract that data from the image from chat GBT. I'm always uploading the image in there if I, if I forgot to say that, I apologize. Um, putting the image in first and then asking it to extract the data from it.
[00:59:01] Rory Flynn: Cause that's what you want.
[00:59:02] Zachary Murray: Um, Peter wants to know outside of localization, what are there other real good applications that you've seen for this kind of
[00:59:09] Rory Flynn: like add creative at scale? It's interesting, right? Cause I think a lot of people are scared to test. New angles, especially when there is such limited creative.
[00:59:17] Rory Flynn: So, you know, if you're a, if you're a brand and you're really running on the same assets, you're using them for your website, your email, your ads, like you're like, I can't do this. I have to stay in this little box because that's all we have. Now you can get a little bit more robust with things like this because you can test more than that also can influence what you do from a, I wouldn't say go generate your brand identity on mid journey, but it can influence your brand direction.
[00:59:41] Rory Flynn: And add to hits. Right. You can add hits really well. And it was a little bit outside of your norm. And it's like, why did that hit? You start running different ads and testing that against it. It's like, well, maybe that's an aesthetic that works with my market. Can I go this as a bigger play in the entirety of a brand or an agency?
[00:59:58] Rory Flynn: Cool. Um,
[01:00:00] Zachary Murray: Amir wants to know, so you used Red Bull as like an example. Um, and he's just wondering, you know, they have a, not as well known of a brand. You know, how can they do like achieve a similar, similar result without such like a well known
[01:00:12] Rory Flynn: global brand with Mid Journey? Yeah, I just used, I use Red Bull because it's complex, right?
[01:00:16] Rory Flynn: Like this will, this will work a lot easier with something, you have a fashion brand, if you have a, you know, apparel brand, right? Something that's not so like technical in nature, like a car driving down a hill with smoke and stuff like that. Um, You know, that's where it's really going to be. It's, I went to the max on that one, probably to try and to try and show you that it's possible, but it's also a strong brand aesthetic.
[01:00:39] Rory Flynn: Like if you have a very similar look and feel across your images, it's going to work. If you don't, and then those assets are
[01:00:43] Zachary Murray: already existing, right? If you need that kind of baseline assets, you build a brand and then it's kind of, you know, building branches out from it. Um, David wants to know how can I keep character consistently, basically, uh, having the same AI generated person, you know, inside
[01:00:58] Rory Flynn: of different assets.
[01:01:00] Rory Flynn: So mid journey is going to develop this, but if you're really looking for this, you know, if you're looking for someone and another tool that can do this, it's stable diffusion. You have to train your own model though, which is where it gets a little bit more. So there is a couple of hacks on this that you can utilize in mid journey.
[01:01:17] Rory Flynn: There is a tool called insight face. So it's called inside face. You download it as an extension to, or a plugin to discord. And what that does is basically you upload a face on there and then you can paste that face onto anyone. Now, what'd you have to do to hack around that though, as being consistent in your prompt.
[01:01:34] Rory Flynn: So if someone has long curly hair, you have to put that in every single prompt so that the structure of the face and the head and the, like the rest of the image looks the same. So. Same clothes, same, you know, hairstyle, same skin tone, and then you can just change the face on there and it'll look very similar.
[01:01:51] Rory Flynn: That's one hack. A million of them. Midjourney is going to get into this, but Stable Diffusion, if you really want to do this, that's where it's at. Um, it just takes a little bit of technique. There's actually
[01:01:59] Zachary Murray: a chat that just came out from Mo. He's like, which is better in terms of controlling your design more?
[01:02:03] Zachary Murray: You know, you're saying, you know, if you want to have a lot of just like tweaks and stuff, they're probably more complex, but go with stable.
[01:02:10] Rory Flynn: Yeah. There's like, if you think about the tools in general, right? Like Dolly is going to be your entry level. Like you can, Dolly is the Canva to stable. Diffusion is the Photoshop.
[01:02:21] Rory Flynn: Right. And like mid journey somewhere in the middle. So mid journey is good at a lot of things and it's specifically not great at overly being like being extremely detailed, consistent, stable diffusion is that. Stable diffusion can get you, you have to, you have to learn how to train your own models, which is technical process also requires some computing power that you need to have.
[01:02:41] Rory Flynn: And it's, uh, no, again, it can be really powerful. That's where things are going to move though. Like people, once they start to understand that, and they can generate images off their own brand imagery, like very well in a very clean, trained model that way. It's able to do really good. It just takes a little bit of time to learn.
[01:02:57] Rory Flynn: Okay. We've
[01:02:58] Zachary Murray: got the last one. Um, it's from Christina and she's wondering if there's an automation for adding runway motion to the assets. She's just saying, you know, it kind of seems like that is, you know, a big chunk of time that you're adding to this workflow. automations or, or, you know, tools, tactics, um, for, you know,
[01:03:14] Rory Flynn: not overloading your workflow.
[01:03:16] Rory Flynn: It's manual, right? Like this. Not yet. I don't believe I haven't figured one out yet, to be honest. I'm just probably not smart enough
[01:03:22] Zachary Murray: manual and going and needing to shoot the video. So, uh, yeah, you can still do it,
[01:03:27] Rory Flynn: uh, in your office, 10 seconds to generate. I mean, that's really all it was. I highlighted the sandwich push generated and it moved.
[01:03:35] Rory Flynn: So, you know, that's as easy as
[01:03:38] Zachary Murray: it gets. Awesome, man. Well, I appreciate you taking the time. That's all we got for today. Um, I just dropped in the chat. We have a new, we have another, uh, webinar coming up on February 7th with Savannah Sanchez, where she's going to break down, um, just some tactics and tools for, you know, building ads and creative that stop the scroll.
[01:03:56] Zachary Murray: Um, Rory, before we kind of like head out and part ways, uh, is there anything you know, you want to say or, or any like where people can follow you? Um, I know you gave the QR code where they can download all these awesome assets, but yeah, anything left, left, left on your side?
[01:04:09] Rory Flynn: Yeah, man. Like I, I do, I probably spend majority of my time on LinkedIn.
[01:04:13] Rory Flynn: So if anyone, you know, I like to tie it just like to have conversations. So if anyone's, you know, wants to pop in the DMs, talk about stuff, I'm always around Twitter, you know, at roar underscore fly, I believe it is. So that's it, man. I mean, it's pretty, pretty simple. I hope the, uh, hope this helped, uh, but I appreciate you guys having me on.
[01:04:29] Rory Flynn: This was fun to go through this again. So, uh, hope we can do it again.
[01:04:33] Zachary Murray: Awesome. I appreciate your time. Thank you everybody for joining. I hope to see you guys on February 7th. Um, and yeah, I appreciate everybody for being super engaged in the chat and made this, and it is awesome.
[01:04:43] Rory Flynn: Thanks guys. Thanks guys.
