Everyday AI Podcast – An AI and ChatGPT Podcast - EP 198: Midjourney V6 - What's new and producing powerful ad creatives

Episode Date: February 1, 2024

AI has changed the creative advertising game. One AI powerhouse tool helping to push the future of art and creatives forward is Midjourney. Rory Flynn, Founder of Systematiq Ai, joins us to discuss what's new with Midjourney and how you can create powerful ad creatives.Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode pageJoin the discussion: Ask Jordan and Rory questions on MidjourneyUpcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTimestamps:02:00 Daily AI news04:25 About Rory and Systematiq Ai07:42 Who should use AI art tools?09:42 Discussing the improvement of AI-generated art14:38 Visual differentiation key in marketing, use AI.19:24 Detailed, professional video, a scroll-stopping social media visual.20:25 Experimenting with photography on a small budget.25:53 Importance of detailed prompts for generative AI.29:17 Using AI for ad creative localization importance.31:03 Discussion of consistent styling in photo set.34:37 Studying and emulating styles in commercial art.37:10 Collaborating, merging brands to create campaign mood.42:22 Adobe offers useful features for design work.43:46 AI can optimize ads, value in performance.47:27 Questioning impact of creative tools like Vision Pro.Topics Covered in This Episode:1.  Understanding AI Art2. Application of AI Art in Business3. Responsible use of AI-Generated Imagery4. Creating AI-Generated Ad Creatives5. The Future of AI in Ad CreativeKeywords:AI art, AI-generated images, DALL E, MidJourney, Rory Flynn, Systematic AI, Vision Pro from Apple, Runway, Adobe Illustrator, AI in advertising, ad creative, version 5.2, localization, tone specificity, commercial use of AI, AI tools, generative AI, copyrights and AI, AI news, AI in healthcare, AI newsletters, AI-enabled biological creation, AI regulations, AI in business operations, AI for visual content creation, midjourney v six, Kanye West's AI usage, prompt handling in AI tools.Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the Everyday Podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. AI has changed the creative advertising game.

And that's not hyperbole. That is actually happening live before our very eyes. Speaking of live, hey, thanks for tuning in. This is live. This is Everyday AI, your daily live stream podcast and free daily newsletter, helping everyday people like you and me learn and leverage generative AI. to grow our companies and to grow our careers. So if you are tuning in and maybe you want to know a little bit more about mid-jurney,

about AI art, about AI-powered videos, about how to maybe grow your business with generative AI, especially on the creative side, you're in literally the best place in the world right now because we're going to be bringing on here in a couple of minutes, who I think is one of the best out there in the game, bar none. and he's going to be walking this all through how to use mid-journey for more powerful and more effective ad creative. This is going to be a good one, y'all. But before we get into that, make sure if you haven't already to go to your everyday AI.com,

sign it for our free daily newsletter. This is going to be one you're going to need to read to believe. We're going to be sharing a lot of examples. This is a very visual show. So, hey, I'll throw out the warning right now. If you are listening on the podcast, first of all, thanks for your support. as always. Thanks for making everyday AI a top 10 tech podcast. We appreciate that. But you might want to check the show notes. And we always leave a link back so you can watch the live show. You got to see

these visuals. I know words, but I don't have the words to describe the level of creativity that we're going to be showing for you today live on the screen. So before we get into this, make sure, as I always talk about, sign up for the daily newsletter. We're going to be recapping the news, this show and more. But let's quickly go over what's going on in the world of AI news. All right, so will chat GPT be used to keep us safe from biological threats? Oh, maybe. OpenAI just released a study on the potential risk of AI-enabled biological threat creation. So the study found that at the current generation of large language models, it does not pose a significant risk, but Open AI did caution that future models could become more dangerous.

Open AI study found that GPT4 only provides a slight improvement in accuracy for biological threat creation and often produces, misleading results. Interesting. Hey, speaking of threats, government officials are demanding AI companies disclose safety results of their models. So the Biden administration has introduced new regulations for AI developers, requiring them to disclose their safety testing to the Department of Commerce in order to ensure transparency and safety in the development and deployment of AI technologies. So companies must provide companies must provide.

including their safety test to the Department of Congress. So the U.S. government is prioritizing AI as both an economic and national security issue and is exploring international regulation and hiring more experts in federal agencies. Tongatied this morning, y'all. All right. And then last but not least, a new open source model has leaked and it's pretty impressive. So there's a recent discovery of a seemingly new open source large language model on Hugging Face referred to as MyQ 170B.

And its potential really was to rival or even exceed the current top performing model, which we've all heard of, this little model called OpenAIs GPD4. So the CEO of Mistral confirmed the leak on Twitter slash X slash whatever we're calling it, saying it was actually based off of an older model of Mistral. So if you follow the large language model game and, you know, everyone's kind of been chasing Open AIs GPT4 for a long time. This is some news you're probably going to want to check out a little bit more of in the newsletter pretty intriguing.

All right. But you probably didn't tune in, if I'm being honest. You probably didn't tune in to hear the AI news. Maybe you did. However, you are probably here to get a master class in Mid-Journey. You know, we've talked about Mid-Journey a handful of times on the everyday AI show. And we've had some guests before in the past. I had to actually bring one of our former guests back for an encore performance because Mid-Journey has changed so much in the past couple of months.

A big jump from version 5.2 to now version 6. There's so much to uncover. I didn't want to bore you with me rambling on. So we're going to bring on an actual expert to help guide us through. So let's do that right about now. So please help me welcome to the show. There we go. We got him there.

Rory Flynn, who is the founder of systematic AI. Rory, thank you so much for joining us. Thanks, man. I'm glad to be back. I'm glad to do this again. Oh, so good. So just like, let's get the basics out of the way. Like, what is systematic AI in like what do you do as a mid-jury master?

Totally. Yeah, man. So systematic AI essentially is a, you know, I call it an operational AI company. Now, again, there's a lot of tools out there, but how do you use them? How do you actually put them into your business and making them work for you? So typically we're just looking for, you know, businesses that have holes in their operations and their creative processes and we're installing tools and processes to get them from point A to point B quicker, easier with more explosive creator. So simple as that, you know, we're doing training. We're doing consulting.

We're doing, you know, a number of different things there. But it's, it's been pretty interesting with these tools and how they develop and the pace at which they develop. So, you know, there's been a lot of opportunity to inject this stuff. Yeah. And just, you know, for those people listening that aren't super familiar, right? Because I think when we talk about generative AI, I think the de facto thing that everyone thinks about is something like an open AI, right, the chat GPT, or maybe they think of, you know, Google Bard. I don't think everyone necessarily even thinks about AI art, right, unless you're maybe in a creative field, obviously. Give us like the zoomed out version of like what the heck is even AI art, what is mid-journey and why should we all be paying attention to it even if we're not maybe a creative.

Yeah, man. So I think when you look at it, right, like mid-jurney essentially at its most basic form, you put in some text, you get an image, right? Pretty simple. But again, typically, you know, when as marketers and people that have worked in a creative field are just basically in any industry now, right? If you didn't have photography elements, you didn't have different brand assets, things like that. You just had to go stock photography sites, and they all kind of suck. I'm going to be honest. You were like, you know, you spend more time on there looking for something that fits your brand than actually like producing the work. So if you think about all the creative needs, needs that you have in a business on a day-to-day. Maybe you have some. Maybe you don't. But there's

social media. There's blogs. There's ads. There's presentations. There's storyboarding. There's this, there's that. Right. So all of them require some visual piece. And how do we just get from A to B quicker? Because I think a lot of people are looking for just finished product from everything that comes out of AI, right? Like, I want to push a button. I want everything to be done. Now, granted, probably will get there. But at this point, you know, we can also look at different pieces in our business operations as to where we get from point A to point B quicker by using the tools instead of wasting time on ancillary tasks. So that's kind of how I look at it overall. And who should, you know, who should be using AI art, right? Is it, is it only people who are in, you know, the creative

fields, photographers, should should non-creatives be using, you know, tools like mid-journey? Like maybe what are some of the use cases when we talk about, you know, just the wide range of people that maybe could, or maybe if you could share some examples of, you know, people that maybe we wouldn't think, oh, they could use Mid Journey for this use case. Maybe let's talk through those kind of across different fields or different professions. Totally. I mean, like, I think personally, designers are the most apt to use the tools. Like, you already know the technology. You already know how to make things visually stunning. You also know the terminology, right? So thinking about prompting the same way as chat GPT, right? You'd have the right tokens to get the right answers. Same thing with Mid Journey.

You know, if you know what a high contrast photo is or, you know, different opacity levels or things of that nature, right? Like you're already been working at it. You can describe it. And that's what's going to get you the good output. But I think, you know, in everyday life, right, you think about, okay, I'm going to post on LinkedIn today. Do I just put a text post up or do I put a crazy visual along with the text to stop the scroll? Like something very simple like that. It doesn't have to be. this whole like crazy production. It'll be something very simple. I need a image from my newsletter that I'm sending out, right? Maybe just to tie it in, I'm talking about, you know, Burger King's new X, Y, Z or whatever the hell they're doing, right? Maybe I just need an image of a,

you know, like a cool image of a hamburger in like a 1950s environment just to like make the newsletter pop a little bit more. So it doesn't have to be, like I said, you don't have to create ad campaigns. You don't have to create, you know, Hollywood movies. You can just do little things. with it. It's just finding those little holes. Same will you probably look at chat GPT, right? I have a question. Can chat GPT answer it? I have a task. Can chat GPT do it? I just think about it in the same way as a creative sense. I need an image. Can Mid Journey do it? Most likely yes. So that's kind of how I think about it. You know, even I want to talk through my personal experience and get your take on this, Rory. So, you know, we've, we've been using, you know, Dolly,

I don't know, two, I think originally was very early on. And we, we've, we've, We were using Mid-Journey, I think, in the fours or before, right? But very early on, I'd say Mid-Journey or AI art in general was very AI art, right? Like, you couldn't really honestly use it for anything because the outputs, you know, at the time, it looked very, you know, computer-generated or, you know, obviously very, you know, disformed, disfigured if you were trying to generate images of people or animals. but it's not really like that anymore, especially when we talk about the big jumps from, from version 5 to now we're in Mid Journey v6. Can you talk just a little bit just about the industry at large and how now, from a quality standpoint, like where are we at?

We're pretty close with Mid Journey. I mean, being indistinguishable. I mean, I've produced things that people would never know were AI. And that's because, you know, there's a detail and a craft to it. I think a lot of the stigma around mid amount like AI art is it's kind of generic. But if you have a creative vision, you can execute it. It comes down to how you prompt. So I think, you know, again, there's certain tools like Dali is really good for people that just need some basic stuff.

It listens really well. It produces really quality images. Mid Journey is like taking that creative juice to the next level, right? And there's other tools beyond mid journey, stable diffusion, things of that nature that can go even further than that. So really just depends on what you need. And I think, you know, the jump from Mid Journey, even like you said, three, which was kind of like, it was cool. It was novel. It was like nightmare fuel.

Some of the stuff you'd get. But, you know, to now in less than, let's call it like 18 months, is insane because I can produce things that are polished and quality ready. And then even if I needed another additional tool to make it look even more real, tools like Magnific, I won't go too far into there. But it's a really good tool for maybe something that has like an AI polished looking face because sometimes you'll prompt. that it looks like a robot right you get a face it looks like a robot like super smooth skin everything is perfect you know tool like magnificent upscales it and then as details so skin wrinkles little hairs freckles you know bags under the eyes you're like okay now it looks like a human not a robot so yeah we're we're moving pretty quick man and i don't you know this is just in the level of quality

from each jump inversion plus how quick it happens you know this development cycle is not like you know your new iPhone every year. And it's like, oh, the camera upgraded. It's like every two weeks we get something that's like that level of, you know, that level of development. So I don't know how to extrapolate where it goes from here and how quickly it moves, but it's going to move fast. Oh, gosh.

Yeah, so fast that I would agree with Jay joining us here when he says, fasten your seatbelt. And I will tell our live audience right now, please get your questions in because we're about to show some examples on screen. And you're probably like, your mind's going to be like, wait, what? like the whole point of everyday AI is for us all to learn together from some of the smartest people out there. So real quick, we're going to take one question here and then we're going to get into some

examples because I can't wait for our live audience to see this. But a great question here from Richard saying, other than replacing stock imagery with AI images from tools like Mid Journey, anyone making money or use of it. So yeah, because I think sometimes, Rory, I think people like me and you maybe live in an AI bubble, right? And we just assume like, oh, everyone's using, you know, majority in Dali. But I think the vast majority of people probably don't. So maybe just talk a little bit here to Richard's question about like, hey, are people really using this out in the wild and are people making, you know, money in a career off of creating AI images? Yeah, I mean, if you're purely just trying to make money off AI images and not like utilize it within your actual business process, I mean, there's a million use cases. I can go, I can go spin up a print on demand shop and sell t-shirts today in 20 minutes, right? I have enough imagery there to go create a t-shirt shop like that.

Do that for any sort of product that, you know, on Printify, right? Like, I'm not going to go, I'm not going that route because it's just like, again, that just bloats the market. And I don't think it's really like where the actual use cases are. I think really looking inside of businesses and seeing how to get from point A to point B quicker, right? So I think about how many presentations do people have to do, right? Like a presentation, instead of doing a boring PowerPoint with a white background and black text, you throw a little bit more inject creativity in there, you have more engagement in your audience. If you're doing presentations for, you know, businesses, if you're doing different,

or sorry, proposals, things of that nature. Giving a little life to it can be really helpful. I mean, like I said, I work in primarily the digital marketing like, e-commerce side of things. That's where I've been for a lot of my career. But I think about ads, social posts, emails, blogs, everything that comes out from a marketing standpoint, how can we just visually differentiate ourselves? That's always what we're trying to do from a brand perspective anyway. So utilizing tools like this and then even inside your internal processes, right, like I think about getting from, like if you wanted to get to a photo shoot, right, you're a brand,

you know, think about that entire process of what it takes to get to a finished product. Maybe you have soda and you want to get to, you know, have a whole set of brand photography. I mean, you could essentially do it with AI, but also if you want to do it with a photographer, right, got to hire a photographer, number one, then when you get to the photographer piece of it, you got to direct a photo shoot after that, you know, or you got to hire models, you got to do all that kind of stuff. From that, then you got to get the revisions done, then you got to get all the images polished. It takes a long time. But also, you don't know how to direct a photographer. So if you give them a set of images that you're like, I want to create it like this, this is how our brand looks,

this is how we want everything to be positioned. That process is just going to be quicker. There's no thought process that you're essentially outsourcing decision making in a sense, right? So it can be a lot easier internally to get things done with tools like this. So long answer, but I mean, there's there's, I can go down that rabbit hole for hours. Oh gosh. Yeah. I mean you both.

Yeah. Good good thing we tell people, yeah, like, oh, you know, everyday I's a 25-ish minute show because, yeah, otherwise me and Rory might actually talk for a couple hours. But enough of the chit-chat, Rory. I need everyone to see the power. So especially when we talk about ad creative, right? This first example that I'm going to show is amazing. So I'm going to have, Rory, I'm going to have you, again, do our best to describe what we're seeing on screen here as I bring this up.

So let's go ahead and take a look here. So before I hit play, because, yes, we're going to talk about this, but there's this whole, you know, photo to video scene. Just quickly describe kind of this, this ad creative that we're going to be seeing here. Yeah. So this is what we've been testing lately. Now, we've done a lot of ad creative with Mid Journey and static like meta ads, right? If you think about the meta ads game, what it is is not necessarily as much of a quality game as it is a volume game because you need to run a number of different ads to actually get data to then pick winners and then iterate off of them.

So at the same time, you know, obviously having good. brand assets is going to be essential. But supplementing and getting more volume so that you have more ability to test so that you can make better decisions and then keep, you know, progressing from there. That's what this is all about. So typically we've done a lot of this with static and now we're just starting to test motion because even subtle motion, even if it looks fake, it might stop your scroll again. We're talking about social media marketing here and, you know, paid marketing. So it's, this isn't like on a camp, this isn't going to be on Times Square, you know, on a video board. This is going to be something that you're going to see in your feed on, you know, in your stories on Instagram, things like that.

So what we did here is like we had static assets. And this is just a mock up. So everybody knows this is just a total mock. We're testing different things here. Again, just to give a little subtle motion, because again, if that stops you in your tracks and you say, what is that? That's a win, right? Instead of people just scrolling right by it. So again, sometimes the variable that people are looking at is different. It's not necessarily what is like this is a Super Bowl ad? Is this a million dollar campaign? or do this take me five minutes to get data? Like, we're on the five minutes to get data side of it. Right?

So that's kind of how we look at this stuff. So good. All right. So let's go ahead. So again, we have a mockup campaign made from mid-journey images. And we'll talk about the process and the other tools. But let's just go ahead and play this. And then we'll kind of talk through a little seven-second clip here.

Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one, conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the All In One Creative AI Studio. Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows, drawing on 60 plus pro-grade tools across

Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible, so you can refine, redirect, or take over at any time.

You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at Firefly. Adobe.com. And yeah, so what we're seeing here, too, is just this, basically we're adding subtle motion to the images. We're not creating like full on Michael Bay action scenes. It's just a little bit of motion, a little rotation, a little bit of zoom in, a little bit of, you know, tilt in the camera, just to get you again from a visual perspective to stop you in your tracks and say, wait, what? You know, that's all, that's all we care about. So that's the idea.

And the quality here is insane. So for, for our podcast listeners, we have three different photos that were created in mid-Journey, all that look part of this, you know, kind of fictional Mark Jacobs campaign, but, you know, similar style, similar lighting. It starts with a photo of a model, you know, with a nice white blazer leaning in. So again, it looks like we have motion. It goes then to a high, like very high-end looking purse. The details are exquisite. I mean, we have shadows. We have lighting. We have, you know, rule of thirds going on. So it looks like, you know, it's almost on a dolly, right? Like not dolly, like the image generator, but like a dolly that you would use in professional

filming. And then the last shot that we have is a watch, right? In this watch, again, highly detailed. We're getting motion. I mean, when you look at this, it looks like it took maybe 10 to 15 humans, you know, a couple of days in probably a six or seven figure budget easily to produce. reduce what in theory could be a little seven second, you know, clip here. So, so, so, so, so Rory, walk us through just real quick what this process looks like,

because a lot of people when they think mid-jurney, when they think AI images, they don't ultimately think what we have here, which what you said is an example of a scroll-stopping social media visual, right? This is a short video with movement. So quickly talk everyone through the process of, of how this even works to get from idea in mid-jurney to something like this. Totally. So I mean, you know, this wasn't a seven-figure budget. This was me and my sweatpants on my couch in 30 minutes doing this, right? So, you know, this is, again, from the process standpoint, I was literally playing around with some level of photography, right? I wanted to have some sort of studio photography to see what, you know, what we can do with it, how close we can get at the studio photography. And again, I'm thinking about like Mark Jacobs and their kind of bold look, direct flash photography, things like that. So from here, it was really just like, oh, I can I get a good look. Can I animate it too? So. So basically in mid-journey, what I'm doing is I'm utilizing a very specific

set. And how I'm doing is I'm being very specific in terms of the look, the feel, kind of the tone, the colors, so there's a cohesive look. And then I can just apply different subjects or objects to it. So instead of it being a woman in a white suit, it's a handbag or it's a watch. So once I have those image sets down, then typically I would go to runway. I mean, normally I just stop it static, right? I'd go build some templates.

I'd create, you know, the ad. But what we're doing now, again, like testing subtle motion, I would take these images from mid-journey into runway, give it a little bit of motion to see what it looks like. Then, you know, utilizing different upscalers, like Magnific for photos or Topaz Labs for video, and then building templates in Figma, putting it all together and watching it come to life.

Now, again, this is not perfect. So anyone in here trying to nitpick, you know, one of the clock hands is a little bit, bit, you know, a little bit not straight. Like, I get it, but this is a concept, right? This is what, this is what can be done. These tools are not fully developed yet like runway. Very good. Give it five more months. I mean, you're probably going to be indistinguishable. So, I mean, there's other video platforms too. People don't, yeah. Like, Rory, people literally, I don't think understand how powerful. Like, I remember telling people, you know, a year and a half ago, like,

hey, look at this, like, AI image, look at this, look at that. And I'm telling people, like, more than a year ago, this is how we are going to be seeing ad campaigns in the future, 100% AI image, like 100% AI generated. Is that like just quickly, because I want to get into more examples, is that where we're headed? Like are we going to get there? Is that the future of, you know, ad creative with things like you just showed here using, you know, three or four AI tools in your sweatpants and in the matter of hours getting

this level of quality? Is that where we're headed? I think so. I think you're starting to see it. I think you're probably going to, you know, if I had one prediction, you'll probably see a Super Bowl commercial this year with some level of AI integration into it. Kanye West just released a trailer for his new video. I'm not technically a real Kanye West fan, but I saw it and I was like, that's runway.

100%. I knew it right away. So people are starting to utilize it. And whether it looks totally polished like this or can be used in a more artistic way, you're going to see it because it's too cool not to, right? You know what I mean? Exactly. Now I miss old Kanye.

Okay. So let's talk a little bit about Mid Journey V6. And let's talk a little bit about what's new. Because, you know, Rory, one thing that you said when we were walking through that Mark Jacobs kind of mock ad. And again, if you're listening on the podcast, you got to go and watch this one so we can see the visuals. But, you know, you talked about the power of prompting. You talked about the importance of having that specificity. and that consistency in your prompting language from shot to shot.

So now we're on the screen here talking about prompt coherence, because that's something that's really changed in new Mid Journey v6. And there's kind of the difference of a similar prompt with a woman, kind of smoking a cigarette in front of some plants with some great shadows and great smoke. But talk a little bit about how some of these updates in Mid Journey v6, specifically, you know, handle our written words and our prompts a little bit better. Yeah. So there was always a kind of a gripe with Mid Journey about its coherence.

Like it wouldn't listen to everything in your prompt. So with Mid Journey v6, the model became more literal, right? So literal meaning that like it's more like Chatsy BT now. Like it'll only do what you tell it to do. It won't go and fill in as many blanks as, you know, it used to. So what I've done here is really tested the limits of the coherence, meaning like let's get super, super, super granular and almost write a prompt like a screenplay. Think about telling it a story.

With this prompt specifically, I went as deep as saying, you know, again, who's the person, what they're doing, what direction they're facing, you know, what color shirt they're wearing, what headband they're wearing, what color lipstick they have. You don't have to go this overboard. But if you want Supreme Control, especially if you're something, you know, more branded or more feel like that, you can get it. So, I mean, even, you know, if you look at this entire prompt, it's probably 60, 70 words, but we're breaking it out very specifically and telling it exactly what we want, like what the position is.

Even, you know, is the background blurred? What is, you know, what is happening with the smoke? The lights interacting with the smoke, right? If you say these things, it's going to happen. So really, it's just getting your vision nailed down. And that's where the art is and the prompt is like saying, this is how I wanted to look. How do I describe it? So copyrighters too, probably going to have a really good opportunity to use these tools.

because you can describe things. You know, that's really what it comes down to. And talk a little bit real quick because I think it's actually, even though we're talking semantics here of, you know, prompt coherence, it's actually, I think, extremely important because, you know, one of the, I think, one of the biggest hurdles to clear or obstacles when it comes to integrating generative AI is the bad information out there or is the people who are looking for shortcuts.

And I think especially with earlier versions, versions of mid-journey, you know, people would try very long, uh, detailed prompts like this. And the prompt handling wasn't that great, right? But now, you know, like, like what Roy is talking about. I'm not going to read this whole prompt here because it would take a while, but I'm going to just spit out like a couple of first sentences. So it says, satisfied Brazilian woman laying in bed with a cigarette, exhaling, subject position, facing right, phototop, phototype, medium shot, editorial, subject focus, sharp focus, midground, wearing, muted blue oversized t-shirt, headband, red lipstick, environment, cozy,

favela bedroom, houseplants, and by, right, that's like a third of, a third of the prompt. But earlier versions of mid-journey or even current versions of Dolly don't do very well with this very long, very detailed prompt handling. And people do test, right? And they say, oh, you know, this version got, you know, 70%. But what we see here on this screen, Rory, is, is you have all of those key terms, right, that you use and it looks like you have about 20 or so kind of key terms here. And it hit mid-jurney v6 hit each and every, right? There's a green checkmark. It hits every single one, right?

Yep. And that's the thing is like when I, the new structure that I've been playing with is an old version four structure where it was really like you had to tell version four exactly what you wanted it to do. So, you know, in version 5.2, you could just write like a few terms and it would kind of create the whole image for you, right? but you had less control over that. Mid-Journey took more liberty.

Now I'm going back to the older structures where it's like, okay, who is the subject? What are they doing? What are they wearing? What position are they facing? You don't have to do all this. This is just if I want something very specific. You know, what's going on in the background?

Again, being very specific, like almost a screenplay, like a screenplay writer, would set this up for a director. You know, so that's kind of how I'm looking at prompting and it's getting great results because, again, it just knows what to do. So even, you know, just taking a lot of the, like, think about it in like chat GPT prompting too. What we've done a lot of the times now is we're limiting those tokens, right?

Like just limiting the tokens, getting rid of the fluff words, getting right to the meat, and it'll read that in a very specific way and then produce the way you wanted to. Oh, gosh. I can just talk about this one for forever, but there's more things we have to get to here. So, you know, especially when we talk about, you know, creativity, right? and ad creative. This example is fantastic. For our podcast audience,

this is a mockup for, you know, Uber Eats. And it looks like we have four very distinct different, you know, campaigns for different types of food. And each one has about, you know,

four or five different ads within this ad set. So I'm going to zoom in a little bit here. Maybe Rory, if you could just walk us through, well, we'll zoom out first and then we'll zoom in. But walk us through the process and maybe how people, can put something like this together in mid-jurney and why they may want to.

Yeah, well, this is for, you know, we've done this for a few brands, right? Like, I can't name them specifically because, again, there seems to be a stigma around AI and people not wanting to know that, you know, they're utilizing AI. So this is a mock-up in essentially of how we've been utilizing it, especially for ad-creative, because ad-creative, again, like I said, it comes down to a volume game and you need as much as possible. So with things like this, you know, there's really an importance in being very meticulous, list, but also like really looking at this and saying, how can we get more localized in terms of our ad set?

So maybe a larger company, you're running ads in multiple countries is sometimes if you're, you know, you want to talk to an audience in South America or Asia and all of your models are American, right? Like that's it. Maybe it doesn't fit the same way as having someone that's hyper localized or even in the language to a chat GPT. The difference between something, someone that talks, you know, New York vernacular and Charleston vernacular, totally different. So if you can change that up and be a little bit more tone-specific and localized, your ads are going to hit better.

So again, things like this, this is where, you know, we can be, again, like even with food, right, like maybe a ham or a chicken sandwich doesn't hit the same way in South America as it does in the U.S., maybe we're utilizing something different down there, you know, empanadas, things like that, whatever it might be. I'm just throwing out examples off the top of my head. So getting more specific and having the ability to be versatile but still staying on brand. that's where the stuff can be, you know, very, very impactful. Speaking about staying on brand, a great, great comment here from who people say is the other

Bash brother of Mid Journey. So shout out to Drew. Yeah, yeah, former guest. I think Josh from the comments here maybe called you guys that. So Drew is asking a very important question here. So talking about consistent styling, right? Because when we're seeing what we're seeing on the screen here, you know, I know it's, you know, kind of like the same image. that's important as well and we're going to get to that.

But maybe let's jump straight into this and maybe talk about consistent styling. So in this example, you know, there's, you know, six different photos and they kind of have this, this dark vibe, dark feel, you know, on a beach and, you know, someone surfing, kind of, it looks like in the Arctic, but this is a great example of having a consistent style or a consistent tone. So maybe if you could, Rory, walk us through kind of this photo set and get to Drew's point about, like how do you even get this consistent styling in mid-journey? Oh, what's up, true? So, yeah, this is good.

And think about this for, again, even this process, this is something you can utilize storyboarding, you know, within an agency, storyboarding campaigns. This doesn't have to be a finished product. But again, consistent styling, I think about it in terms of mid-jury prompting. You want to break an image down into visual building blocks. So you have a subject, you have an action, you have lighting, you have composition, you have color scheme. You have different elements, right?

Like, those are all little visual building. building blocks. Now, typically, like, the composition in the color scheme and, you know, a number of different, like, details and modifiers, that's how you're going to want to stay consistent. So especially something like this, we're going with like a dark atmosphere. We want to have a specific color scheme. We want to have different elements of, you know, I want to have black mountains and black sand. Like, I'm putting all of that in the prompt. Now, again, when you want to consistently style it, sometimes all you have to do is then change the subject. So then it goes from, you know, high angle shot of a beach to close up of a, of a surfer with frost on his mustache.

But then at the back end of the prompt, it's still, you know, black sand, black ocean, dark atmosphere, you know, high contrast, things like that, muted color tones. So when you're able to just find that prompt structure that works and those building blocks, then you can just iterate it will because then the consistent theme is going to come through. And that's how you can really start to build these worlds or ideas around one, you know, certain vision, if that makes sense. And so, you know, a question here, which I think is maybe we should have addressed this one sooner, but it's probably something that so many people are thinking is the question here from Sajid asking, can we use runway and mid-journey for commercial purposes? Because if you're watching on the live stream like me, even though I see this AI image, like generation stuff every day, I'm like, holy cow, this is crazy.

Maybe if you're listening on the podcast and this is maybe new to you, it's something you're thinking like, oh, yeah, can you be using these amazing tools for commercial purposes? I know, it's kind of a gray area, Rory, but what's your, you know, what's your response to using this for commercial purposes? I've done it. I mean, we utilize it, but we do it smart. You know, we do it the way that you have to be cognizant of all the different pieces that are going around it. Would I use this for a Super Bowl ad? Probably not. Am I using this for a, you know, giant billboard ad? You know, probably not. My users on, you know, meta ads and social media marketing right now, definitely, right? So things you've got to be aware of, number one is that we don't know what these tools were trained on. We don't know what Dali was trained on. We don't know what Mid Journey was trained on.

We don't know where that data came from. So utilizing a lot of the times reverse image search to check to make sure that this is not, if I pulled up one of these images before I put it out commercially, that it doesn't come specifically or derivative from some other work, right? That's one thing. And also, just from an ethical standpoint, don't like to use artist names in my prompts. ever like any sort of artist names right it's just again you don't need to do that and it's i think a lot of people do that in the style of um commercially right like do it for fun you know i don't know if that's if

that's still crosses the ethical boundaries i just don't do it but the if you want to go look up someone like you've seen a lot of this stuff on west anderson like a style west anderson right if you want to know how west anderson composes his imagery you can go look it up on chat gbt you can say what kind of compositions, color schemes, lighting effects. And you can add that to your prompt. Then you can build out those visual building blocks, like I said, without having to go and be so, you know, blatant about stealing kind of the style. Now, again, it opens up a really long conversation. But things like that, yes, and then also just being very cognizant of what copyrights are, you know, what the copyright laws are within the country that you're in.

Like, U.S., everything that's produced fully by AI can't be copyrighted. So essentially it is your own personal stock photography that anyone else can use. Same thing. There's other countries now that are utilizing different tactics. So I'm not versed in every single country. I just know that's what U.S. says. So always something to be cognizant of is, yes, we are using this. But again, where did it come from?

What was it trained on? How is it generating so that you can be protected and be ethical in what you're doing? Yeah. And I will say this without getting us too far because we need to see some more of these creatives, because it doesn't make sense how good they are. However, like, you do always, always caution. Like, I always caution people read terms of service. Even personally, Mid Journeys, they had a big update.

I talked about it on the show about, you know, a month or so ago. Not a huge fan of Mid Journey's new updated terms of service, but like Rory said, use this smart, use this responsibly, especially if you are using it for commercial campaigns, use your best judgment. It's important also if you are working with an end client that you use transparency, right, like that you disclose when and if and how you may be using AI imagery. So I had to get that out of the way. But speaking of commercial campaigns, this one is silky.

So let's talk about this. Rory, maybe just describe a little bit what we have going on here and kind of the process and how this can be ultimately be used for ad creative. Yeah. So I came from the product licensing world. Essentially what that is is you're putting two disparate brands together and creating a new product line. So that's kind of how I thought about doing this. This is a mood board, right? But like, to me, producing this for a presentation to pitch this from maybe I was representing Jeep and I

wanted to pitch this to Patagonia, this would have been months to get this done or actually look this good. I did this with one prompt. Right. So we're collaborating, taking the two brands starting them together. Now we have like a whole mood board essentially of a look and feel. This could also be for a campaign. This is just way, again, like think about these processes from if you're working in design field. It's just how do we get from point A to point B? and how do we make it smooth? How do we tell the story? Because I think at the end of the day,

if you can't tell the story, then what's the point? So that's kind of where the ideation came from this, and that's where a lot of this stuff I do comes from. So good, so good. All right, let's look at one more because I know, you know, I was telling Rory before the show, oh, we'll totally, 30 minutes.

We can't. My gosh, this stuff's too good. So this is a fun one, right? And obviously, going back to our other points. So this is, you know, Nike shoes and very, you know, futuristic, you know, bright neon's. But talk a little bit about this one. And obviously this isn't one where you would, you know, commercialize this, Rory. But talk about just some of the levels of creativity of how you got to this kind of Nike NASA collaboration mood board here. And also maybe

some advice for others on what you learned, you know, from even creating a mood board like this and how they can use, you know, some of your findings or learnings. Yeah, man. So something like this, this was initially when version six came out. What I really want to do is just push it to the limit. Moodboard, when you think about it for an art generator, is really hard because it's a million different elements. There's a lot of different textures, styles, colors, things going on.

So basically, that's always my like test is how far has the new version gotten. And this one, I specifically just love Nike branding. I mean, you know, who doesn't, maybe I, maybe I'm in the minority there, but whatever. I always like to see Nike and NASA because there's two disparate brands, right? Like space and Earth, right? It's two totally different pieces. How can we blend them together and make it cohesive? So here, you know, you see a lot of different textures, a lot of different colors.

You have the technical theme of NASA. You have the like street vibe, you know, like the street wear vibe of Nike kind of blending together in a very cool and unique way. So I like to test these different things to see how good mid-journey gets, you know, basically each release. And this one was just really fun. I mean, like, I just love, I love using those two brands together. If you ever want to go try that, anyone, Nike and NASA, it comes out specifically great basically every single time. So that's, that's kind of the thought process behind just having a little fun with the tools too, you know. So good. So good. All right, here. We're going to do this

before we wrap because there's so many good questions. And that's what I love about the show is smart people can answer questions. Let's let's go rapid fire here. Rory, we'll see if we can get through a couple of them quickly. So Alar asking, so I'm focusing heavily on creating short ads for brands. What's your thought on this approach? I think it's useful, right, depending on what your source images are and how you're getting to that end result. I mean, tools like Runway can also utilize, you know, existing product photography. It doesn't have to be just generated images. Like you can utilize your static images to animate them. So I think it's a great idea, especially if they're open to testing things like

this because why not? You're going to have a lot more video tools coming out other than runway, you know, over the next six months, you know, stable diffusion video tool, things like that, it's going to get way more easy to do this. Yeah, huge. And yeah, we talked about that earlier, just the process of, yes, mid-jorney images are great, but then being able to bring them into another program like runway and add, you know, visuals or add movement, that's where it becomes super powerful. All right, next, rapid fire. Here we go. Richard asking, don't AI tools, text and pixel just lead to everything becoming just a bit generic? I think that depends on who's prompting it.

You know, like, I don't know if I've seen another Nike image that looks like that. I mean, you can go try to find it and search on mid-jurney, but I'm not saying I'm the best or anything. It's just my vision's different, right? Like, I think that's what it comes down to. So, you know, if you can get different with your vision, if you just say the same thing every time, like, girl on a beach, like you're going to get things that look generic. We're going to say, girl on a beach, you know, with the sun half setting, reflecting purple light against the water, blah, blah, blah.

It's going to look totally different than just girl on a beach. So I think it all comes down to what the vision is and executing on it. Perfect answer. All right. Anthony asking, I wondered what the reception everyone is getting from clients to start integrating this into the projects. What's it been like for you so far? I think it's been great because it's a, I'm using it as a supplement, not the, not the

entirety, right? So like, again, it's a supplement. I like to think everyone, you know, it's not just like, let's scrap our agencies, let's scrap our photographers, let's scrap everything. It's a supplement. We can test it. I think every good business is testing and tracking their math. metrics. So again, things like this, can we inject this a little bit? And also, what we've seen with

some of it is it's adjusted course on some brand, you know, positioning, right? Like we've tested some more AI looking stuff with a little bit more focus on specific areas. And then like it starts to shift the perception because that stuff hits on social. It's like, okay, let's maybe make our ads look like that. So organic to organic to pay. So it's interesting how it's developing. But I mean, again, like it's always, it's always client by client basis. Everyone thinks about it differently. Cool. Tara, Tara asking, how would you compare this to Adobe's AI tools, their Firefly? I like Adobe. I use Photoshop generative fill for a lot of different things, depending on what, you know, sometimes I pair the two together.

You know, I think there's a really good thing for anyone that's in design or needs vector images now on Adobe Illustrator. I don't know if many people talk about it that much, but there's a text of vector option. So this is something I want to drop here because it's super helpful if you're looking for icons or little type typography, things like that, like text to icon or text to, sorry, I just totally lost my train of thought there. Text to vector is like massive because that used to be a pain in the pain in the butt on AI was creating or on Adobe Illustrator, you know, creating vectors. So think about that one too, really helpful. So so good. All right, here we got two or three more. So Alar saying, you mentioned that you don't use artist names. What about brand names like you did with the Nike shoes?

Yeah, I'm not putting that out for commercial use, right? Like that's just basically showing some examples of what can be done. Now, maybe that's wrong. I, you know, again, I typically tend to utilize that for own personal use. I'm not going to go and put that out into commercially. So, again, maybe I need to think about that one a little bit more. But, you know, consistently, I've just been doing that for fun to test the tools and to see kind of what can be produced from it.

Yeah. And obviously you're not alone in that one. All right, Maricio, asking, if you were to offer creative ads with AI, how does that pricing differ? to a real marketing agency. Great question. Again, I think it comes down to how much you want and how much needs to be what the output is. Because if the output is the same, like, you know, if basically, if we're creating the same amount of ad sets with humans and with AI, like I don't see the difference there from a pricing standpoint, especially if the performance is going to be able to be judged

a lot quicker and you're going to get more data and access, you know, to optimizing those ads. So I think it's, you know, I think a lot of people think that AI is going to drive the price down. Personally, I don't think so because it's going to help performance if we have more data and are able to optimize quicker. So the value of that is just as important as the creative itself, right? So I think, again, it's going to take a little bit of positioning if you're experienced with the tools and experience with the outputs to be able to position that to potential clients. But at the same time, if you're producing results at a much quicker rate and a much higher clip, again, your value is still that, not just the creative. All right. And then our last and our rapid fire questions from ALA.

Again, what's your thoughts on post-production, like adding some unique elements than when, oh, at that point, it's not fully made by AI. Yeah, what's your thoughts on that? 100%. 100%. Like, you know, I do a lot of this stuff in Figma. You saw the Mark Jacobs ads. You know, they do the Uber-Reeds ads in Figma. I create templates in there, drop the images in.

And even the after effects in terms of like building different pieces into the ad sets. Like if you're a graphic designer, right, like you don't need to use a full image. Like I need to have all this text in there. It can mid journey generate all of it. No, just generate the image and then add the text yourself. So, you know, it's just a piece in the mix. Like I said, don't think of it as everything is a finished product. It's just a piece.

You're a chef in the kitchen putting all the ingredients together. And this ingredient can be done a lot quicker than slow roasting, you know, a pot roast for five hours. You know, this is like the Instapod. version where you do it in the air fryer version we do in like 30 minutes. Right. So like that's kind of how I look at it. Oh, so good. All right.

We made it through rapid fire. I'm taking up so much of your time. But I have to ask you, Rory, as we wrap everything up here, right? Because the premise of this is mid journey, you know, kind of the combination, right? So if I were to summarize everything, I think that the AI image in the AI creative space has changed so much in the past six months with the giant leap in quality and control going to mid-journey v6. And I think that mid-jurney right now is so far ahead of all the other AI image generators. And then combining that with the new controls and capabilities.

We didn't even get into multi-brush and runway. We'll have to do that on another show. But you combine these huge improvements in quality and control of mid-journey v6 with being able to get consistent images and then being able to create movement with these AI images by using an AI video tool like runway. In my opinion, this changes the way that ad creative is done. So, Rory, with all this in mind, maybe what's your either piece of advice or your biggest takeaway for people when it comes to AI's influence on ad creative in our society in 2024? What's that takeaway?

I think the biggest piece of advice I can give is learn the tools now while they're developing because I think about this, quick analogy, if you gave, you know, your grandparents an iPhone right now instead of learn it and they've never been experienced to it, they'd look like hieroglyphics to them, right? But us, we've been with every iteration of the iPhone so we can just pick it up and know how it works. That's how I think about using the tools and learning the tools because they're only going to develop and the more expansive they get, the harder it is going to be to just pick it up and use it. Right. So that to me and I think about looking at the future of marketing, right? Like is everything that we're doing,

right now from a creative standpoint, the end-all, be all? Like, is it always going to be email? Is it always going to be websites? Like, I think about the Vision Pro coming out for Apple, that's going to create immersive environments, right? You're going to be able to have an immersive environment for your brand marketing perspective, right? What does that look like? You know, you're going to be able to do that with conventional tools quickly enough to get to market to compete. Like, this is where, this is why don't sleep at night. You know, this is why don't sleep because I'm just like, I see Vision Pro and I'm like, what? Like, what does that do to everything? So, you know, without me going on another diatribe for 40 minutes here, you know, I'll kind of put it at that.

Oh, man. Well, hey, if you do want to hear like 40 minutes more worth of content, that's what our newsletter is. So, so, so Rory, thank you so much for joining the everyday AI show and walking us through what's new and mid-journey and how we can use that with our ad creative in AI. We so much appreciate your time and your insights. Yeah, man. It's great, great being back. I appreciate you having me.

Oh, always. And hey, there's so much that Rory just showed us. This is literally, we just got a university 101, 201, 201 course on AI art. But if you missed it, yeah, there's a lot going on, visuals, prompts. We're going to be sharing it all. Don't worry. You know, Rory has just a wealth of information. We're going to be sharing more about his company, some of the, you know, courses and, you know, guides that he has. Make sure to check out in the newsletter. It's all going to be there. Go to your everyday AI.com. Sign up for that free daily newsletter to check that out. Also, make sure to join us tomorrow as we talk about maximizing the effectiveness of AI and healthcare with the president of the American Medical Association. All right. We'll see you back tomorrow and every

day for more everyday AI. Thanks y'all. Meet Firefly AI assistant. Now live in Adobe Firefly, the Allman One creative AI studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe. Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI.

Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your Everett. EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.