Midjourney masterclass: hacking visual content creation with Midjourney, with Rory Flynn

Hey, it's Navid here. And in this session, you'll learn how to hack visual content creation with Midjourney. And I'm here with a true expert on this. His name is Rory Flynn. He's the co founder of systematiq.ai. He's operationalized common AI tools for businesses and marketing teams by injecting AI into workflows.

He's amplified productivity and impact, streamlined finances, and revolutionized client experience and communication. And I know he has something really, really special prepared for us here. So I'm excited. I'm, we're going to dive right in. So we're welcome Rory. Yeah, man. Thank you for having me.

Uh, you know, definitely, definitely excited to get going on this, but we do have a lot of stuff to go through. So I'm just going to jump right in if that's cool. Yeah, I would love to, you just jumped right in and we are going to share some, uh, something inside of Midjourney as well.

So you're going to get a full walkthrough in this session here. So definitely stay tuned for that as well. Totally, man. So let's do it. So, all right. So basically, like, I think everyone's here because they know that times are changing, like AI is dominating the social landscape, it's dominating the headlines, and the future is generative.

But like, what the hell does that mean? Right? Again, before we get into this, just a little bit about myself. Uh, my name is Rory Flynn. I'm the founder of Systematiq Ai. I am not a designer. I am not a media buyer. I'm none of those things really. I just try to solve common problems.

Right. And this is what basically, you know, what we do at Systematiq Ai, you know, I like to call us an operational AI company. Cause really it's not just about using the tools. It's really finding systems in different ways to inject them into your business. So we do a number of different things between training and consulting and done for you.

Add creative things like that. But realistically, it wasn't always like this way. It wasn't always this way, right? So where it all started. Essentially, we were digital marketing agency. We primarily focused in the areas of email and paid marketing or paid media. We had 90 plus clients at the time, right?

So think about that. Typically, we're always buried in insane creative needs. We had minimal assets and just always like a lack of bandwidth. But, you know, across 90 clients, typically, even on the email side, we had to do around 10 emails per month per client. So that's around 900 emails.

A month that had to get out the door, you know, extrapolate that out. That's 3000 plus emails over the course of the year with revision, with revisions and things of that nature. So it's just a ton of work right now. We decided in that case, because we were so far behind to use AI to like amplify our productivity, reduce our costs and just generally create happy clients.

Um, but again, like enough about me and we're going to get into everything here, but like really, who are you and like, what are your problems? You know, probably a solopreneur. Entrepreneur, maybe you're a digital marketer, maybe you're a content creator, but you're probably using AI in some sort of fashion.

The question is like, how are you using it? Um, you know, maybe you're doing it for personal tasks. You're doing things like writing emails or social posts. Maybe you're doing, you know, single tasks and maybe you're getting even into automation. If you've gotten into automation, congratulations.

You're probably far ahead of the curve, but, um, you know, this is where we can want to talk about building systems and really how you can inject all of these things into your workflows to make them much faster and much more powerful. So again, you're probably using AI, but at this point, you know, there's a lot of tool overload.

There's so much information and you really don't know what's possible, right, and where to go from here. But again, if like operationalizing AI, like the entirety of it, really, it starts with just identifying small problems, like identifying the holes, um, you know, looking for tools that can plug those holes and then starting to build SOPs around, right now, commonly, like where we go from here, like, where are these typical problems showing up every day?

I see this because I've worked in the marketing industry for a long time. There's like a creative issue because every month you need a ton of social posts. You need newsletters, ads, presentations, you know, blog posts, things of that nature. It just never ends, right? So, you know, the average small business, typically what we're working with, um, you know, they're working with fatigued assets.

They're working with a small variety of assets and just subparts like static imagery. Right. It's not the best, like you have some, but it's not a lot and you always need more. So again, this problem with the static though, and this images, they're just limited quantity. There's just not a lot to go around.

You know, they're time inefficient and they're also expensive to get that done. If you need to go do another photo shoot again, the time between starting a photo shoot and getting to the finished product, that could be a month, two months. It's just a really long time to get new stuff, but you really always need.

New stuff, right? So let's solve this problem. We'll get right into this thing. So next 45 minutes, we're going to run through kind of how to juice your Midjourney, how to juice your creative with Midjourney, like asset hack your brand identity. Building AI systems, you know, that are built for scale and then what the ROI potential looks like.

So first, just a little bit about how we got here. Um, you know, the rise of AI definitely happened in 2023. Uh, chat, GPT. So we saw tools like chat, GPT, Midjourney, stable diffusion, Dolly, like. All of these started to come out and seemingly 10, 000 other tools came out. But, you know, they were novel for a minute, but businesses essentially have started to leverage them and really start to push these tools.

Now, I think this is kind of where we stand right now. This is just my own personal view of it, of how this adoption process has seemed to, has seemed to go. Um, you know, first there's like stage one, like you discover these tools and you start testing them. Like, Oh, I found chat GPT.

I'm going to write an email with it. Right. That's kind of where a lot of people started. Then it was. You started to adopt it, like, how can I get better at these tools and start to use them in my individual life? Like, can I use them for my own personalized, you know, emails or writing blog posts, whatnot?

You know, stage three was more like operationalized, where people were all utilizing these tools in the, uh, you know, in the, in the business, maybe. So maybe there's five people on your team, you're all using the, like, the tools, but you're all doing it a different way, and there's no real standard operating procedure, right?

Stage four is basically when we start to systematize this stuff. We build SOPs around the tools, around, you know, how to utilize them, guidelines, things of that nature. We start to string a couple of different tools together. Then I think we're pushing closely towards stage five here, where we're going to start to customize and really build out these systems, right.

That are more so tailored on building custom train models, you know. Automated agents doing multiple tasks. So that stuff's coming again. I don't even know where stage six goes from here, but I'm just kind of looking at the tea leaves as we unfold here. So kind of like when I look at this at the current stage, right, of like where everyone's at, the majority of us are probably in stage two or three, like closely pushing towards stage four.

Um, but again, using tools, solving problems, operationalizing, but how do we know that some of these businesses are doing it because it's happening and we're seeing it live, right? Like. This is SuitSupply, um, you know, luckily before doing this presentation, I kind of came across these ads on my feed and I said, look, you know, maybe these models are, uh, are real, but the backgrounds are 100 percent AI, right?

Like, you can just look at this and say, great. You know, there's, there's definitely those flowers. I don't know who's doing those photo shoots, even some of the details in the corners and things like that. They look like yarn. It's definitely AI, right? So the difficulty level on this stuff is low, but the effectiveness level is potentially really high.

So like, why would they do this? Um, because realistically it's on brand, right? Like it's visually unique and it's scalable. So that one background could turn into 50 backgrounds and those six images of people can turn into 300 images of people. And that's all different ways that you can test and, you know, utilize different pieces of creative in your brand, right?

So even, you know, some of these things are getting so good that you can look at down in the left hand corner over here. Thomas White says, you know, brilliant picture, great clothes. And of course the camel, like the camel doesn't exist, right? So it can be, it can be utilized in a brand sense.

Now let's talk about what's possible now, right? Like what we can do now. So these are just some examples of, you know, what we've done. In the space, I'll be quite candid here. I do not work for Uber Eats as one of my clients. Mood was one of my clients, but we've done this for a few brands, especially on the advertising side.

Um, there's still a stigma around AI 100%. They don't necessarily want people knowing they're using AI. So that's sort of a mock process in terms of what we've been able to do there and what kind of results we've been able to get. But just wanted to Put that out there, you know, up front so that everyone knows, um, now something like mood, right?

This was a case study when they came to us and how we utilized AI with them. They came to us post launch, fairly post launch, not really anything further than a month or two, but they had a really aggressive ad budget. So they were going to spend very quickly and they need to get assets quickly to get everything in their funnel so we can start to convert clients right now or convert customers.

So big problem was, you know, they had a big audience and the sending volume was fairly large. So we were going to do anywhere from 20 to 25 flow emails to start this thing out, which if you're not familiar with email marketing, those are your automated backend, um, you know, email flows, welcome series, abandoned car, browser, abandoned things like that.

And then we had to do three to four campaign emails per week to keep up with the volume of traffic that we were getting inbound. So. It's a lot of creative, right? And it's a long time to wait for that. So we utilize Midjourney just to give it more of a unique visual aesthetic. Because again, from email, it's a, it's a medium, right?

You cannot purchase directly off an email. You have to go from email to website. So again, how do we get people there in a very efficient fashion, but also Peak their interest and catch their eye. So that's why we utilize Midjourney to basically give us a little bit more of a differentiating factor in the CBD or the, uh, you know, the Delta eight space.

So, you know, the results were pretty insane early on. You know, we increased click through rate by 0. 38 percent in the first month, which, you know, if you work in email marketing, you know, is a fairly decent size jump. So we saw it and we saw it happening quickly. So we were able to tailor more of what we were doing and be a little bit more specific.

Right in the in essentially like the messages that we were sending because it didn't have to be just whatever stock imagery we had or whatever product imagery we had, we could tailor the image to the message. So the marketing personalization got a lot better. Now, this is pretty much as simple as it can get there.

This is another one of our clients that we did this for, right? This is the easiest way to visualize it. Sometimes you don't need like this giant production from AI. Sometimes you just need a stock photo that will work well within the marketing channel that you're utilizing. So here basically we just created it.

You know, we were trying to launch, um, the YouTube channel for one of our clients and basically all we needed was a picture of their YouTube video on a wall. So we created the image in Midjourney, we put it in Figma and then we put it in the email template, right? Pretty simple.

This took all of maybe five minutes. So again, it doesn't have to be this major full on production. It can be something as simple as this, just, you know, instead of going and looking through a stock image site for hours, you know, we can do this right now. Again, how can you do this?

So we're going to get into now how to like really dive into Midjourney and utilize the tool super, super, super well. Now, again, if you want like unique creative, a lot of this is going to start with Midjourney or any of the other AI, you know, image generators. Bye. I believe what works in here is understanding the tool and, you know, really knowing what works and what doesn't work, right?

I think that's another piece of, you know, really getting good stuff is just knowing what to avoid. But if you're not familiar with Midjourney, it's a powerful image generator. It's a super creative tool. I tend to think it has some of the best aesthetics of AI. Um, you know, again, there's people that have their own personal preference between dolly, stable diffusion, whatever.

I still think Midjourney is my favorite personally. So It can do a lot of different things and it's pretty versatile now breaking down like Midjourney, right? The good thing was like with chat GPT coming into the scene earlier on We were introduced to this idea of like the prompt, right?

And the prompt is really what makes Midjourney run and that's where the art comes from now typically, I like to look at prompting in a very specific fashion like Pretty simple equation here, like clear and direct prompts equal clear and direct output. Ambiguous prompts equal ambiguous output, meaning that the less detail you put in your prompt, the more Midjourney takes over and takes like it's creative liberty, right?

So if you just put like dog in a park, like you'll get a dog in a park, but you might not have control over anything else. But if you're clear and direct and you want to go really deep into each little aspect of your prompt, it's going to follow you more closely. And you're going to get something that's closer to your vision.

So, you know. All this stuff is cool, right? Like you can do futuristic living rooms and aliens and spaceships and stuff, but like a lot of times for marketers, we need stuff that just looks real. I mean, that's really at the end of the day what it is. So, you know, for us, typically the best way to trigger Midjourney into a, let's call it a photo realistic mindset is to use photography terminology.

That's what I've always found has worked the best. Um, you know, keeping it very consistent in that sense and also priming it to be more photo realistic, right? So what does that mean? Um, yeah. Essentially, we're using photorealistic elements here to really go and amplify these prompts and we'll use things like subject and action, you know, environment, composition and shot type, mood and emotions, specific cameras and lenses can work really well to if you're looking just to, again, trigger Midjourney into a photorealistic mindset.

You're using things like film stock and the lighting and the color scheme, details, and modifiers. All this stuff can be useful. You don't need to use every single piece of this. Um, sometimes you just need a couple different things, but typically if you want to build a structure and get basically really good at prompting and prompting quickly, it's nice to have a format and a formula so that you can just plug this stuff in and go.

Now on the right over here, this is a cheat sheet. I'm going to link this at the back end of the presentation. So you guys have all the functionality that you need, um, to utilize the tool. So first we're just going to look at like a simple prompt formula again, like this is how we can create the image and how to think about it, like try to break it down into visual building blocks and then like.

Build it back up into an image, right? So typically when I'm looking at a prompt formula, we'll get something like a photo type photo type could be anything from like a closeup to a long shot to, you know, editorial photography to street photography, whatever that might be. Um, and the subject in action, who the person is, what they're doing.

Environment where it's taking place, color scheme, like a camera lens film. Like it doesn't have to be in there every time, but sometimes I like to put it in there, lighting, composition, additional details, X, Y, Z. So again, thinking about it like this, how you're going to build your picture.

It all starts with words, right? And we're utilizing tokens here to get to the end result. Now, the other thing that's important with this is prompt structure, uh, because Midjourney tends to read their prompts from left to right. And what that means is I think when you put something in the beginning of the prompt, it holds more weight than if you put something at the end.

So essentially, like if something's really important to you, put it in the front. If something is not important to you, put it in the back, right? Because you will see here in the next couple of slides, and we'll go through this. Like this is a standard prompt example, right? Like I'm just going to use this simple prompt just to show you kind of how everything looks.

Now, what we did here is we have the Land Rover Defender in vibrant Cusco, Peru. Uh, freeze motion, which is, again, this is just like freezing an object in motion. We have asymmetrical composition, meaning like it's not just perfectly centered in the middle of the photo. Uh, 35 millimeter type of camera.

Berger Pancro is a, is a film stock. Extreme tonal balance is the difference between light and dark elements. In a photo, so like good use in professional photography, it just ends up looking really polished, right? Street photography is just a candid style of photography. So now you see how we have this structure, right?

Like Land Rover is first and then everything else comes after it. Now they're looking at this image, like that's what comes through. Like the Land Rover is the star of the show here. Now, all I'm going to do here in these next couple images is just going to switch up the first term, nothing else in the prompt, right?

So the first one is Land Rover Defender, right? Like the same image we just saw. The second one is vibrant Cusco, Peru. So what's going on in vibrant Cusco, Peru, you get a lot more vibrant and you get more of the city. Like the city looks to be more of the star, the car is pushed back in the image.

So again, that came through the third one. We put extreme tonal balance first. And as you can see, like the difference between the light and dark elements, like definitely comes through. And then in the fourth one, we did burger Pancro 400, which is more of like a moody film stock.

And it definitely got that moodier vibe, that came out of it. So sometimes you don't necessarily have a bad prompt. Sometimes you just structured wrong. So just keep that in mind while you're doing this. Cause oftentimes you end up tinkering with prompts forever. So this is something that can really help.

Now let's get a little bit more detailed. Cause this is where I think Midjourney version six has gotten a lot better. We've talked a lot about, you know, there's been a lot of conversation about Midjourneys coherence for something like Dolly or for some of these other platforms, but we've been able to get really detailed.

And again, this is pushing it to the extreme. So you can see. How this all works. Now, this is a really detailed prop for you. You do not have to use everything here, but if you want to have a very singular vision, and you want to have that executed, you can break it down this granularly.

And you can see how each one of these things work, right? So here I'm doing like phototype the subject, but I'm also going into the subject position, right? Like, is the person laying down? Are they facing left? Are they facing right? That can that, you know, can be controlled subject focus, like, where are they in the depth of the photo?

Are they in the foreground? Are they in the mid ground? Are they in the background? Are they in focus? Are they out of focus? Um, you know, what are they wearing specifically? You can do all of that. Like, you know, what kind of T shirt they're wearing, headband, lipstick, things like that, you know, environment, of course.

Again, talking about the location, where it is, same thing with the environment focus, like where is that, where's that depth? Is it, is it blurred? Is it crisp? What do you want? You just tell it and explain it. And then like environment, description, emotion, lighting, tone, like even time of day, textures, image type.

So like, again, think about when you're building a scene or telling a story, this is kind of like a, a, um, Like a screenplay writer would give this to a director, right? Like you're giving them every little piece so they can just go and capture the shot. Now, again, when we do a longer prompt like that, I'm not going to sit here and read this all out for you guys, cause it's kind of boring.

But at the same time, you know, we hit on all of these different, uh, you know, pieces of the prompt, like, you know, satisfied Brazilian woman laying in bed with a cigarette, exhaling. Facing right, you know, medium shot, editorial sharp focus, you know, oversized blue shirt, headband, lipstick, like, when you look at all this, it all comes out because we gave it that direction, you know, even something as simple as light interacting with smoke, like you see it right in the photo there.

So when you think about prompting, right? Like, that's you can be generic and you can be, you know, super light in your prompt. You don't have to go heavy like this, but if you want it, it can do it. You just have to be very descriptive with it. Typically, when I'm prompting like this.

Try to cut the fluff words out, you know, longer sentences, you know, there's a lot of generators out there like chat GPT and whatnot that'll get you there. But a lot of times you're just using, you're just using and wasting tokens and fluff words in there that don't mean anything.

So again, just like use the most powerful words possible. Um, you know, when you're doing this, maybe two to three words per description here, and you're going to get something very close to what you want now, even something as detailed as this, right? Like when you want to iterate it, once you've built something out like this prompt, all you have to do is maybe change one of these elements.

So we think about breaking all that down, right? I go, like, if you look at that prompt specifically, maybe the subject position, I had them, you know, laying down facing, right? You know, now I just want them maybe sitting in the center of the frame, right? Like all I did was change center of image right down there.

Right? So all of that, all that wording and this is all I changed and that's what happened. So when you're iterating, think about that's how, you know, if you like something or you don't like something, that's as easy as it can be. You don't have to go and change the whole prompt.

So I think that's pretty useful for people that are using this, you know, for branding purposes. Now, this is what is also very useful for doing this, you know, for branding purposes or keeping a very consistent style or look across your images. You want this because when you do this again, like you, you don't want to have your images all over the place.

If you're trying to keep a look and feel, whether it's for your social media, for your website, for your blog, for, you know, uh, even like YouTube videos, things like that. You want it all to look the same and you want it to be on brand. So, you know, something like this, like you see over here on the right, I basically put together this little storyboard, you know, of a surfer in Antarctica or Iceland, whatever I have here.

Um, really all I did was thinking about the visual building blocks, right? Like we have all the details about. You know, the background, we have the black mountains, rolling ocean, dark atmosphere, muted colors, like everything that's in white over here on the left, that's what I'm just going to keep for every single prompt.

Like I don't even need to change that. It's just really, all you need to do is change the shot type, the subject and the details, like of what they're doing, who it is. So, you know, the difference between. Close up of a surfer with frost on his face verse, you know, close up of a penguin on the beach That's all that I changed in this prompt everything else on the back end stayed the same So really again, that's how you can build out this visual signature and like this visual identity that keeps you consistent So you can stay on brand so that's just uh, you of the other example outputs It's really no, you know, there's really no end in sight to what you can do.

I just tried to pick some random ones From my library, just so everyone can get an idea again now, moving into more of like hacking this, this tool and what you can actually do with and why it can be so powerful because at the end of the day, like as marketers, right, like, we just need to extrapolate a lot of things and just, we need a lot of volume.

So how do we take some of our own brand images? Let's say our own brand images and turn them into infinite variations of those brand images. So they even stay close, but you can utilize them many times, right? So what is asset hacking? Essentially it's reverse engineering an image and then like, basically we're using it for consistent, uh, you know, brand relevance.

Again, I always say to disclaimer, like use this on your own brands, your own personal brands, whatnot, please do not go and do this for every other brand, please. Um, but like, here's that basic idea. So we have like our brand assets, right? Then typically from our brand assets, what we're doing is we're breaking them down into those visual building blocks we're talking about.

So once we have those building blocks, just like in the consistent styles we talked about, um, then we can basically have infinite iteration. Of images. So essentially once you have those building blocks down, then we can rerun these prompts a million times. We can, you know, change the subject, change the photo type, change the environment.

As long as the image look and feel is good, you know, then we have a branded consistent set of imagery. So typically what this process looks like, essentially like you organize your brand assets, then you run them through the described function in Midjourney. Then I'm using things like chat GPT to kind of tune those prompts up.

Then we'll go an image prompt in Midjourney and then we'll iterate. In Midjourney. So it's a little bit of a two step process between chat, GBT and Midjourney. Now here's kind of, we're going to go through this process so you guys can see how this works. Um, you know, typically, you know, I would, I wanted to use something that's a little bit more complex, like Red Bull to me to has very strong imagery, like it's going to work very well with this same thing with Apple Nike.

You know, anything that's got a really strong brand presence, a brand like aesthetic. So, I wanted to use them because Red Bull's a little bit difficult, right? There's a lot of decals, it's high action shots, it's not, you know, super, it's not just like pictures of faces, right?

So, um, you know, let's see how we can do this. So, essentially what we're going to do is we're going to take this, this one image here, which is the F1 car. Then we're going to turn it into a thousand other images, right? So. First thing you got to do is you got to run this through the describe function in MidJourney.

If you don't know what describe function is, it's basically, you know, you type, you give MidJourney an image and it shoots out four prompts for you that it thinks that image looks like. From there, really how you do that is you type in backslash describe on the, uh, on discord. And you just paste the image and you hit enter, go ahead.

And then everything happens from there. Now, if you're on the alpha site, now it's pretty simple. You're just going to upload an image and you can be a little bit more customized and building this out. But you know, we utilize discord for now because that's still what's going on for majority of people.

So again, when I run that through describe, this is what I get, right? Like one, two, three, four. Those are the four prompts that we ran. You know, if you're looking at this and saying, cool, like, you know, maybe there are some cool images here, but to me, this kind of looks like crap.

It looks like video games. It doesn't look like a real image. So for me, as you know, utilizing this in my brand in any way, shape or form, I just wouldn't use it. So how do we keep going here? Right now, this is where we're going to take. That image, and we're going to bring it into chat GPT.

So we're going to take that, that F1 image. We're going to put it into chat GPT, and then we're going to write this prompt, right? I'll try to read this somewhat verbatim and explain it, you know, without being too boring, but basically we want to analyze the image and we want to create a prompt for an AI image generator.

You know, we want to describe the image like an award winning professional photographer and extreme technical detail. That's really what we're trying to pull out is a technical detail. Cause that's something I can't look at the photo and say, like. You know, what kind of aperture is that?

Like what kind of, you know, ISO value is it? You don't have to put these in the prompts, but it's great to have it as a descriptor in there. So typically then we keep going with this, you know, use it, use this formula to structure the prompt. So we'll insert the prompt formula.

Um, you know, that we saw before, like one of those ones that I gave you earlier, you know, include specific camera lens, camera settings, you know, be very technically specific. You shorten powerful keywords and phrases do not use full sentences. Right. So gave it some direction on how to do this on how we want it to output.

Now, the second step of it, totally optional, you don't have to do this, you know, once chat GPT has provided you that new prompt, you can also do a second prompt and add some of the described functions in there as well. So some of those described prompts, if any of them worked really well, you can have a blend those two prompts together.

So sometimes you get a really good synthesized prompt out of that. Sometimes you don't need it. It's just, uh, you know, another step if you want to go a little bit further. So again, like the new prompt that we come up with now is panning shot, Red Bull racing car, dark indigo and amber.

So we have, you know, this photo type, we have the Red Bull racing car, we have, you know, color scheme, we have hillside, which is the, you know, which is the environment. Then we have basically some camera elements here. We have 35 millimeter film style, dappled sunlight, bold contrast, delicate, high precision, uh, speed photography.

So again, I would not be able to do this with just my eyes and, you know, describe is not going to put out something like this because it's kind of all over the place. So this is again, now we're starting to build this theme, like this core imagery set. So once we go through that, right, we run it back and we go through, you know, go through Midjourney, paste that prompt.

Here's what we get. Now, some people could stop here. I would say this is still, this is still pretty far off, right? Like we got more photo realistic, but we didn't get close enough to that image to where I'm comfortable because it's still a little bit out of control. Right now, what we're going to do from here is we're going to use image prompting.

So essentially, if you're not familiar with image prompting, it's basically you give Midjourney an image, you write a prompt, and then you could basically combines those two things together. So, what we're gonna do here, and think about it if you've utilized ChatGPT before, like Typically, it works better if you give it data first, right?

And then you generate off of it because it just, it has some context to work off of. So this is kind of how we're thinking about it. So typically here, when you're doing an image prompt, you're adding the image and then you're adding the prompt that we just had before, right into the same prompt and you're pressing enter.

Now, the way to get really dialed in with this, you know, a lot of times Midjourney has this, um, you know, the parameter is set at the default. For image weight, image weighting basically is a parameter that you can utilize at the end of your prompt. And what it does is basically.

Change the balance between how much weight is on the text and how much weight is on the image. So you can play around with that. Typically a higher level of image weight will be closer to the image itself. That's going to be something you can utilize. Again, every image is going to be different because based on just the way that the journey generates.

So when you do that, right, we get to something like this. So essentially we took that. One image of the F1 racing car. And we broke it down into little text indicators and the little visual building blocks, and then we built it back out in Midjourney. Now we can play with it. Right?

So this is essentially what we can do now. So when we iterate this and we want to get to like all these different assets, essentially think about that prompt formula again, right now, everything we're going to keep the same here. So that color scheme, the lens, the camera lens and film, the lighting, the composition, the additional details.

Those stay the same, and now we can just change the photo type, the subject and action, and the environment. That'll all keep that same look and branded feel. So, when we iterate again, that's all we're doing. And these are the kind of images that we get. So, instead of it being, panning shot Red Bull racing car, panning shot Red Bull skier, panning shot Red Bull motorcycle, panning shot, you know, uh, Red Bull mountain biker.

So again, we can build out that aesthetic and basically where we started was with these assets. Now what we have over here are these assets. Now these might not be perfect in your eyes. I think these are pretty good in regards of what we can get across, especially if it's your brand, not Red Bull.

So again, that's where we can really start to take this and extrapolate this. Now, each one of these photos can be iterated on into infinity. You can do any different subject environment, action, whatever. As long as we keep that same look and feel. You have a brand. So that's kind of what's really interesting about this now.

How do we like put this to work though? Like what is really like, what is really the next step? Like, great. We can create images. Well, what do you do with them? So again, let's just say you needed a thousand assets and I'm just throwing this out there cause I work a lot in the paid media space and typically, you know, what we try to solve for in the paid media space is a volume problem.

Like we always need more assets to test, right? We need more assets to test. Cause that means we can optimize. That means that we can get better results, but this process that we're going to utilize here doesn't have to be just for ads. This can be for YouTube thumbnails. This can be for social content.

If you have a couple of different accounts that you're running a company page, your page, you know, is syndicated pages, things like that. If you need assets for different, uh, you know, different, uh, Social posts, blog posts, whatever, the same process can be utilized. So typically what we're doing here, I'm going to do it in the context of, of ads.

And basically the process is what we'll work with everything. So just follow along with that. Um, really what we're going to do here is we need five tools. So you need Midjourney, chat, GPT. Figma, uh, if you're not familiar with Figma, it's web based and a, uh, desktop based design app.

Um, there's CopyDoc, which is a Figma plugin. That's the linchpin in all of this. So keep, you know, we'll go over that again, but really remember that one. And then Google Sheets and Excel, any sort of spreadsheet software will work fine just with this. So essentially with these 1, 000 assets, what we're doing is we're going to take only five images, you know, five pieces of copy, and then a couple different aspect ratios.

And we'll just make. A bunch of different combinations of it. So again, same theory applies if you're doing a YouTube, you know, YouTube thumbnails, and you want to test a bunch of different images with a bunch of different copy. Right? This is how this can be utilized. Um, so this is what the process looks like.

Typically, we're generating the templates in Figma. Then we're generating the images in Midjourney. Then we're going to go do the copy generation in ChatGPT. Then we're going to organize these images in, uh, organize these images in a spreadsheet, and organize the information in a spreadsheet.

Then we're going to go generate the ads. So first thing we do is like generate the templates right now, what we're doing here is for us in the ad space, we're typically using our highest performing templates. Um, what we're doing is building these out in a very specific fashion aspect ratios, depending on what, uh, depending on what medium you're advertising on.

So whether that's, you know, Facebook, Instagram, tick tock, Google, whatever that might be. So we'll build those out and then we'll have placeholders for the headline, the sub headline, and then the images, right? Or whatever other texts you have to have in there. So that's where it starts.

Then you can generate your own images in Midjourney. This doesn't have to be Midjourney now, right? Like you can have your own images that you utilize. You can use, you know, different things that you've generated in the past on Dolly or whatnot, as long as you have some sort of base image here.

This process will work. So what's going on here? You can, a lot of times we're picking winners for ads. So whatever images have worked in the past and we're iterating off of those. That's kind of how we've been able to get really good success with it. Now, when you want to generate copy for something like ads or YouTube videos, whatever you're doing.

Again, I think everyone has their own process for this. Typically I'll show you what we do for, you know, chat GBT and getting the right copy down for these ads. Like typically we're doing like a company overview. Brand guidelines, tone, brand tone of voice, consumer communication, brand marketing examples, right?

So we have some, uh, you know, some really good examples to feed that in there. The target audience, of course, the product focus, you know, different things about the products that we're selling. And then the copywriting guidelines, because typically this is an important piece, mostly is to keep these under a certain, you know, under a certain Word count, if that's the best way to put it, because, you know, it's oftentimes ChatGPT will go and write you like a novel when I need five words, right?

So we're very specific in terms of like our headline needs to be less than five words or a call to action needs to be less than two words. So things like that can be really helpful in your prompt. Now, this is probably the most unsexy spreadsheet and slide you've ever seen, but it probably has the sexiest result.

So what we're doing here is Is basically we're linking, uh, we're going to end up linking this spreadsheet with Figma and then everything is going to populate with one click. So really what we do here is we structure this with the necessary information. So we want to have the headlines, sub headline, the CTA.

The images, you know, we'll, we'll attach those into the spreadsheet and also zip them up in the, uh, in a file with the spreadsheet. And then we didn't want to have the template number. So it's all linked together. Right now, when we do that, basically we're going to go back to once that doc is done and we've zipped it up into a zip file and we go back to Figma.

And when we hit in Figma, basically, we're just going to run a sync between that spreadsheet and Figma using the CopyDoc plugin. I'll say that one more time, CopyDoc plugin. It is a lifesaver. So, essentially, that would turn, if I only had five images, five pieces of copy, and seven or eight templates, that would turn into a thousand ads just like this.

It would all populate instantly. So then, from there, basically, like, You have a finished product of a thousand assets in less than an hour. So if, again, think about it in terms of testing, or if you're doing a lot of social media posts and you just want to get them done, you don't want to actually go create them.

If you have templates, you can do this. Do you want to create YouTube thumbnails? This process can work. So again. This is all very, uh, you know, how to put these tools to work, how to string together multiple tools, how to make them work for you now, like, why would you need that though?

You know, I think there's probably a lot of people ask me that. I mean, if you're doing anything that requires testing, you know, having a ton of different variants is really important. And also, like, you have full control within a tool like Figma. So, you can then go and edit this stuff post production.

So then maybe, you know, you want to go change, like, one piece of copy. Or you want to adjust an image. Or you want to try something else. You now have full editable capabilities in there. It just saves a ton of time. Like if I was to do a thousand ads and copy paste that over and over and over again, I mean, that way I would also like probably want to blow my brains out.

So, I mean, you know, that also, if you're working with a designer, that's going to save their time to free them up on other things that they can be doing instead of just copy pasting asset over asset over asset. So we, again, considering going through this, like what are some other things to look at?

There's another tool that I find very helpful if you're going to utilize this in like a commercial setting. Right. So. Really, what is this? This is magnific. It's an AI upscaler, but at the end of the day, like it adds subtle details that make it look more lifelike. So really what it does is, you know, it also, it upscales your image, but it will go and add different, you know, characteristics to someone's face, right?

It'll add some, some lines. It'll add some shadows. It'll add a little bit of like blemish, maybe some bags under the eye. So it just makes things look more real. So it doesn't look like on this left, the left side of the images face is. Polished, right? Like it's, it's very like smooth.

Everything is almost perfect on the right. It adds a little bit more human elements to it. Like the skin wrinkles. There's, you know, a couple of different blotches on the face. That's good to me. Your, uh, it doesn't have to be your cup of tea, but for me, like I want things to look real and indistinguishable.

So the more real a person's face looks like that, the better it can be. Now, the other thing I want to talk about, of course, is the ethics and legal side of this, right? So if you're doing this for yourself, I mean, again, you are personally responsible for what you're doing. But if you're doing this for clients or for anyone else, like, you just have to have the conversation.

It has to be an open discussion that you're utilizing, um, that you're utilizing AI in the work, right? Like, you have to make sure everyone's comfortable with that. That's the first thing I say. But like the other thing to understand is typically what's going on, you know, in the copyright situation.

So in the US, which is what I know the best because that's where I am, um, essentially what's going on here is anything that's fully created by AI cannot be copyrighted. So that's my understanding of it. Meeting. Whatever is going to be produced by Midjourney or whatever and put out for, you know, commercial purposes, number one, anyone else can use it because it's technically your own personal stock photography that now everyone has access to.

So just consider that, right? Like, if you're very brand conscious and you don't want people taking your stuff, just know that, you know, even if they don't know it's AI. Like, they could still do it and there's really nothing you can do about it. So, again, think about it like that, but also, you know, from a Midjourney's copyright perspective, right?

Like, you're, they're not liable for whatever you create. So, the thing is here that this is gonna, this is what can happen. So, I just want everyone to know this too. Like, basically, this, these two images on the right, like, if you're not familiar with them, it's a famous photo from about the 1980s.

I think it was a National Geographic cover. And It won a ton of awards. It was like one of these visceral photos. Now, all I did to test this, and this was a while back. So Midjourney has gotten better at this. What I did was I really just created a prompt, I went and looked up the photo, I got the technical details, like what camera was utilized, and you know, what type of aperture, things like that.

Plugged that into the prompt, ran this first shot, came out almost exactly the same, right? Everything is pretty, pretty damn close. Like I look at the, I look at her shoulder, and there's even like this little green patch. Over here, simple things like that, the background. Now, I didn't even use an image prompt on this.

So I didn't use an image prompt influences. This was just a straight text prompt. So just know that this can happen. And basically if you're generating something on here, we don't know how these images were trained, especially on tools like Midjourney. So again, there's no way to know where they're coming from, what they're pulling from, how they're generating.

So you gotta be careful. And some of the things I do here for this number one is our reverse image search. Basically everything that I'm putting out for commercial purposes, ads, anything that's going on a website, whatever. So utilizing tools like dupli checker, there's a million reverse image searchers now you'll do it oftentimes.

And you'll find a couple images that you're like, Oh, I see where this pulled from or what this generated from. So just know that because all you have to do is then go iterate it and you don't have to have that same image. You can do it a thousand times. Just don't do anything that's close.

Don't expose yourself there. The other thing I talk about is. The do's and don'ts. Like, I typically don't like to use in the style of prompts. So if I was like, you know, surfer on the beach in the style of Wes Anderson, right? People preach a lot of this on, on, you know, social media and on YouTube and whatnot.

I don't think it's good if you're going to utilize it for commercial work because if you're going to work backwards, like basically if someone was getting you into litigation, going backwards and they went back to the end result, which was your prompt and it said in the style of X person, right?

that. So I typically don And he's sort of in the style of, I don't like to use names in, in the prompts. Now, the other thing you can utilize there instead of doing that is actually going to learn it, right? Like if you like Wes Anderson's composition style, right? Go to, you go to chat GPT and you say, you know, or his image style, I'm sorry.

Go, what is, you know, what type of composition does he use? What type of color scheme does he use? What type of equipment does he use? And then you can build out those visual building blocks that we were talking about and you can create in his style, but you learned, right? Like you didn't just go and blatantly take it.

So I think that's. Uh, one way to think about it. Right. And there's a lot of things that this is gonna progress very quickly. I think, you know, as we get to, you know, see really more of where these tools can go. So, you know, I think it's something to keep an eye on and always be cognizant of what's going on in this space.

Now here's, you know, the, the, the interesting part, I'm sure a lot of people are interested in what these kind of results look like, right? So what we've seen in, we worked on this, we did 114 projects in November with ai. And on the low end, which is really interesting part of it was still 30 percent reduction in time spent on these projects.

So think about creating ad sets in, you know, in the thousands for brands and things of that nature, right? On the high end, it was an 85 percent reduction in hours spent. We're really focused on tracking this meticulously because at the end of the day, if it doesn't save time and it doesn't save money, like what the hell is the point?

Right. So we are very, very diligent and meticulous in tracking that. Cause again, like wouldn't be telling you this if it wasn't, you know, if it wasn't a use for a time saving purpose or some sort of amplification of your process. So again, that's where it becomes really, really powerful.

And again, the ads that we're running had a hundred percent CSAT positive score. So again, they're, they're, they're winning, they're working, they're generating dollars. And that's the thing that's really keeping, you know, this whole train going now, really at the end of the day, what did it.

What did AI do? I mean, it amplified our productivity, it reduced our costs and it was, it's been in less time. So that's kind of the idea and the theme of this all. And again, even with a process like I talked about with copy doc, that's less about just AI than it is about having an actual process, right?

I mean, something like that, that singular process, you don't have to do a thousand ads generated from scratch. Maybe you have a winning ad and you want to turn it into 50 more, you have a winning image and you just want to change the copy, you have 50 copy variations. Same theory applies with the YouTube, uh, YouTube, um, thumbnails, right?

Again, those, you have a winning image, you want to test some different copy, or you have a winning image style and you want to keep going with that same process. So again, all these things are going to be here to help. And it's all going to amplify, of course, over the coming months with all these new tools and developments.

So now this is a little gift for me to you. Um, basically it's QR code. If you want to scan this, basically I have the Midjourney cheat sheet in there. I got four PDFs for beginners on getting started. If you want to have, you know, just a step by step walkthrough on the process, there's some prompt formulas and there's some additional resources in there.

So I think, uh, you know, hopefully that this was helpful and, uh, you know, I think we're going to get into a couple more things here, but appreciate you guys letting me run through that. Yeah, we have some more stuff. This was really valuable. Hopefully you guys, uh, have gotten some value from Rory's presentation so far, but we actually, we will dive into Midjourney now to show you a little bit in real time.

And, uh, Rory is really good at that, but of course with the live demos with this stuff, sometimes you have to run it twice as we have done previously. But yeah, if you want to go through a few examples of what. Uh, this can look like maybe something photorealistic, maybe something that, uh, like a thumbnail or something that people actually can implement as a creator.

Totally. So I think, you know, what's, what's useful here, right. It's like, if we're all trying to stay on brand and we want to have this like photo real estate, realistic look and feel. It's weird. Let's talk about like, kind of what I did with the consistent styling, because I think that's what a lot of people need is I need something in this style, but I need it for a ton of different things.

So, like, let's take a look at that. I'll show you kind of what I did there, um, just to give you like a good understanding. Now, let me share my screen back again. Here we go. So really, like, again, if I wanted to look up these, you know, the surfer picture, right, let's go back and find some of these images here from the beginning.

Um, so, yeah, so we have like, that's right. Like, this is. Basically, if this is our brand colors, like, you know, maybe we have a dark atmosphere or this is kind of what we're trying to get to essentially on this is the alpha site, by the way, this is going to be live for everyone, probably in the next couple of weeks or so.

Um, but again, it's very easy to prompt in here and very similar to what happens in discord. Just a little bit different. So again, like when we look at this, we want to keep everything the same. We want to keep all of these images, you know, really close. So when we do this prompt, right, we look at this prompt, uh, you know, closeup surfer, wearing a full body wetsuit covered in frost, you know, frozen mustache.

So that's like our, our shot type. Our subject, some details, then we just have like the back end of this, which is this, you know, minimalistic Icelandic landscape, you know, high angle, but that shouldn't be. And then that's actually a mistake. So again, everyone is getting to see what this looks like in real time.

We're not all perfect. So black mountains, rolling ocean, landscape, dark atmosphere, muted colors, sharp resolution, color contrast, right? This is really like the building blocks of it from about here to here. This is the, this is the aesthetic piece. Now, when we change like these little elements up front, we'll see how we can keep it pretty consistent.

So instead of doing this, let's do close up of a seal, uh, walrus, uh, surfboard, and then let's do quickly covered in frost. And then we'll hit enter and we'll watch this stuff generate. So this is kind of, again, hopefully with this, as it generates, we will see that, um, you know, hoping the journey doesn't have like a little hiccup on me here and take forever to do this.

We should see basically the same sort of style and aesthetic come through. And again, that's where the asset hacking can come really in handy. What I talked about is like getting it to that point. So you have that look and feel. And then from there, just extrapolating off of it, so I might have to, uh, I'm going to have to refresh the browser for a second here, just to see if we're going to actually get to these images creating or, uh, there we go.

Okay, so really, we had, all right, everything is still generating, but here's like our surfboard, right? Like, what did we ask for surfboard covered in frost in the same exact style? So it gave us that in the same exact look and feel. Now, when we go down here, it's still taking forever to generate, um, even have this in turbo mode for anyone that's used a Midjourney before in the past.

So it should be working a lot faster than this, but here's like our seal, you know, like a close up of a seal, basically in the same exact style, here's our, let's hear a walrus and here's our seal covered in frost in the same exact style. So this is where it can get. You know, really good for branded images, right?

Like even if I wanted to go do something like put myself in this photo, or like that would be something, let's say, okay, I want to go do, and we'll see if this, uh, you know, if this actually works here and we'd be really impressed, but you know, typically this takes a number of runs, so we'll, we'll give it a whirl here, but let's put this prompt in here and then I will add a photo of myself.

Yeah, this is what I, what I had in mind actually, because you can replicate expensive photo shoots kind of, I mean, in a way like to, because that's what you showed before where you had, I mean, in a previous session we did, you showed actually how you replaced yourself, uh, you know, and put your headshot or something like that on it and it looked completely real basically.

Yeah. So there's a, there's a good process for it. And if you can get it dialed in, this is the hardest part of the process, right? What we're doing right now. Is we're going to use image prompting and image weight to get to a point where it looks somewhat like me and then we can use this tool called insight base, which basically swaps my face really the important part is having the facial structure somewhat match so that it looks the same like a fat person on a skinny face is not going to look right.

So, I mean, do you think that will actually be that you can just upload a photo of yourself eventually and it can just replicate it in Midjourney? Do you think that's something that might be coming like and getting better with this stuff? Yeah. I think so. There's some open source models on GitHub that you can do that with.

Um, you know, Midjourney, it'll probably get there at some point, but let's see if we can do this here real quick. So man, tuxedo and sunglasses are sorry, man, wearing a tuxedo and sunglasses. Um, actually let's change this sitting and let's see if this where I was sitting, man, wearing a tuxedo.

Okay. And maybe I'll just add a little bit on the back end here to see if we can get a little bit more pop and resolution. And then what we're going to do here, so I'm going to add some image weights and what I'm doing with these little brackets here, this is called permutation prompting.

So instead of having to go and prompt this each individual time, if you put these curly braces, On the end of it, essentially, it'll just generate all those prompts individually. So typically, I'll do like 0. 5 here, 6, 0. 7, 0. 8, 0. 9, and then 1. And then we'll see if this will actually work because sometimes, you know, this is, this is a pretty testy, uh, pretty testy process on here.

Now, once these generate, it might take a minute, essentially, again, what I'm looking for is, or if I should describe this a little bit better, the image weight, uh, parameter on the low end is 0. 25 on the high end is two. Those are the values right now on 2. 5 or 0. 25, that's going to be closer to the text prompt.

So what I wrote in the text bar. On the other side of it, if you go to image weight 2, that's going to be closer to the image. So it's going to be like, it's like a sliding scale of how much you want each part of that to influence. So, pretty much here, and I'm probably going to have to reload again just to get even somewhere close.

But let's see. Are we even generating? Alright, so we're starting to get there. Right now, I probably could have done this in a much easier format. Of course, this is how the whole process works, right? Like there's always, you know, there's always some sort of screw up along the way.

At least for me, I am not perfect. So even something like this, right? Like this, this could work for me, right? Like, even if, you know, we don't know what my hair looks like, maybe I just want to go here and I'll just copy this job. I'll just show you guys so you have an idea of what this looks like.

Job ID, I will go into Discord here, and I'll tell you show, there's a little trick. If anyone doesn't know this, you can copy your job ID and hit show and then open it in discord. And then from here, this will show that job in discord. And then I can utilize it within the, within the actual tool.

So let's use number three. And like I said, oops, that's not the one I want. Number four. So here now this insight face, it's a, it's a plugin. You can download it fairly easily off their website. All it does is run this app and it's called in swapper. And once you hit this, basically it'll send it, it'll put my face on the subject, and then I have a picture of me, right?

Might say to yourself, that's not my hairstyle. Again, you can generate this and let's go back to, let's go back in here and see if there's anything closer. I just wanted to show you that instead of just waiting for, you know, 30 minutes for this stuff to be done. Um, cause then let's see if there's anything closer here.

Can you get a lot of variations of this with the image weight specifically? Some of it will be close. Some of it won't be close at all. Right. But that one, that one worked the best. Cause then I really, where it gets. To the next level is when you take it into something like Magnific, right?

Like I just generated these before we got on here just to see if it will work. So, you know, again, you can look at something. Like this, where my face got on here. Now, Magnific will add these details. And I'll zoom in a little bit. I think it might be a little bit far out. So let's see if we can zoom.

Click right here. And then I'll try to zoom in. There we go. Okay. So let's. All right. So if you can see here, like this is what Magnific can do. That goes from like this grainy little photo, right? To this. Grady photo to adding all this depth. So it added depth in the beard. It added like a lot more, uh, contrast and different sort of texture to my hair.

My skin, even if you look at the skin right on the forehead, that starts to become real. I get some wrinkles up here, even down like on this sort of example here, which is a different one. What's interesting is like, look at the, when you look at the hands, right. Let's see if I can zoom in on these hands here.

Oops. So like the hands, right? Like this is the kind of stuff that you're going to have to play with. If you want to utilize this in real life, like in your business, because like those hands look so polished, but I add this and we add hair and we add all that we have the skin to the knuckles, we add, you know, the veins.

Like that looks so much different. Just that little element that if I can even zoom in a little bit over here, let's see, go in a little bit further. So we can see this thing. Sorry, it's a little bit, this one's a little bit hard to control. So yeah, so this, this hand, I mean, that looks, Totally different.

Right. And that's the, that's the nice thing about tools like these. So I mean, yeah, that's, that's pretty simple in terms of how fast Rory in terms of the speed of improvement, like the hands in Midjourney, just, just a few months ago were kind of horrendous. The word didn't look anything like this and now it's actually pretty good.

Would you say it's nailing hands and stuff like that more these days, like with V6? Definitely. And I think you have to be, you have to be very consistent about what you want. Oftentimes, if you're, there's a little trick that I've used a lot of times when I can't, you know, regenerate the hands.

Often now with the vary region function, you can just paint over the hand. Uh, you know, you can highlight the hand and just have it regenerate until it gets it right. There's a few other things you can do is utilizing image prompt within, uh, you know, the vary region to get that hand if it's bad.

But, you know, luckily enough, I think I have one in That is, let's see. I'm going to scroll down where I was just, I was testing the hand theory. I wanted to see how, how many hands I could get in an image, right? So we did pretty good, right? Like you can do it. Like everyone's always did Midjourney, Midjourney, figure out hands yet.

I mean, look, I mean, we have 1, hands in here. Like this was a, this was a concept I was doing for a skincare brand, right? Like how could we actually figure this out? So, you know, it's possible, right? And this tool, like tools like this, again, it'll make it look even more real.

Like that was, I had to generate a super low res to get the hands, right. And then I can bring it up more high res with a tool like this. So there's a lot of things, right? Like even your brand photography that we have somewhere in here, right? Like this is a, you look at this face again, this goes from polished to almost looking real.

Right. Polish. So, you know, tools like this, it's just insane. The level of development, um, and let's get back out of here. So there we go. That was. No, this is awesome. Do you have any, do you have any beginner tips for people who are just getting started with the AI image generation tools?

Of course they can use, Dalai three, but if they want to take something more advanced, uh, what do you have to say there to get started with Midjourney? Yeah. Yeah. I think Midjourney really like. If there's one thing I can say with it is to go into it having fun first, like this is a really dumb thing to say, but if you go in there, like wanting to create things that aren't for your brand and you just want to have fun, you'll end up learning more than if you're just like, I need this image for my brands right now.

Um, but typically learning the tool, like learning the tool itself, like the functionality of the tool will get you so much better. Then like just trying to prompt or like copy prompts from people. So if you go in there and you understand, like have a little bit of prompt structure, right?

Like what I talked about before, then essentially from there, like learning what the parameters do. So what does it, you know, what does the aspect ratio parameter do? What does the stylized parameter do? What does chaos do those things? Because oftentimes you can switch up your image with just a parameter, like a lot, like a big variation.

So you can get to these different things. Cause I think people just get lost in tinkering with the prompts where it's just like, Oh, maybe it's this word, it's this word, sometimes it's just a parameter or sometimes it's the structure. So knowing the tool really well and what it can do, cause then you can get creative with it.

So that's kind of my, my advice. And what mistakes do you see people are making? Are there any mistakes you can make? Or it's just a creative outlet? Like, what, what do you see? Like if some don't like some, I mean, you covered it a little bit in your presentation, but is there anything that you see that you absolutely don't do this when it comes to Midjourney and then you will kind of fail with this.

I think it's knowing that there's, there's some things that just aren't like never going to generate consistently. Like if you just want pictures of hands, I mean, like just go somewhere else, you know, that's never going to be, uh, that's never going to be the best spot for it. Bye.

There's things that it has problems with, like stares. I always tend to find it has a problem with stares, like getting stares right. Um, you know, different, like, reflections. Reflections are really hard to get it perfect, like a nailed reflection of, like, someone's face and it being reflected perfectly.

It has a problem with that. It's a lot of things that typical artists have problems with, right? Artists have a really hard time drawing hands and have a really hard time drawing eyes. Like, you know, eyes in Midjourney tend to be really good, but I mean, you know, it's hard. So those are things, those are things I wouldn't say stay away from, just be cognizant of, because if you're just trying to generate that, that and also having too many, too many subjects in your photo.

So meeting. I want Michael Jordan and Gandhi and Barack Obama and Donald Trump, like, playing in a basketball game with the sun, you know, exploding, it's like, when you have too much of that, it's never going to generate the right way, you know, so it's, uh. Keep it, keep it simple.

Keep your focus simple. Um, and then you can expand out on it the same way that we talked about iterating and going further. So that's kind of why. Yeah. I also like the community aspect with Midjourney with discord. I mean, hopefully that still stays, even though they have their native app.

Now, I don't know if you see anything with that, because actually if you're new to Midjourney, you can go into discord and actually see in real time what others are creating. I'm, I'm think they're going to replicate something like that with their native app as well. I mean, you, you know more about this story than I do.

So, okay. Speak to that. Yeah. I hope the, I really hope discord sticks around. Like I'm a fan of discord because I love the community. And the great thing about Midjourney and what they do on their discord is they, they take the feedback verbatim. Like they'll ask questions, they'll, you'll get to have a voice in like the development of stuff.

So they'll put out polls, you get to vote, then they'll share them on their office hours, you know, so I really think they're going to stick around there. Um, But the, the one thing that is a little bit of a pain to me on the new website is there isn't like a channel feature where I can, like, I can create projects in channels like I can in Discord, and there's no way for me to share that yet unless I'm missing something where like I couldn't have like myself and my team member in my, in the same channel generating together, which I think is really good for interesting.

I kind of see that problem with a lot of AI tools. Like I think team collaboration needs to be a bigger focus for a lot of these tools to like, bring this in even, I mean. Canva and stuff like that got it with, I mean, that's, that's bulked in, but I think like specific AI tools, they're making it really difficult.

Even like when I tried some tools like Opus Clip and stuff, they make it more difficult. Although you can invite people, it's still not, it's still not perfect. You know, that's, that's kind of what they need to get a little bit better on so we can use this together with our AI. technology can do for us to make us more efficient and stuff like that.

So, well, do you see anything for the future, by the way, if it Midjourney, you, I mean, B6 was released, but anything else you've seen in terms of Midjourney or AI in general that you would like to share? Yeah. So I think, you know, Midjourney is. Really going to push, you know, to, like I said, to fix some of the things that people want, right?

I think a lot of people want consistent characters, so that's really hard to do. Just a caveat again, from, you know, anyone that's tried this before on Midjourney, it's hard. You can go do that on stable diffusion. If you know what you're doing, right. Then you can do that in Dolly.

If you know what you're doing, the quality is not going to be the same. Really, I think that's one of the things they're pushing forward towards on their roadmap. So, you know, getting a picture of me and then putting me in different situations, right? Like that's going to be one of their focuses at some point moving forward.

So I think that's a really good piece, especially for brands that want to do this or creators or You know, entrepreneurs that want to like do, you know, very specific things and keep like an aesthetic throughout the entirety. So on that, the style, style reference that they're, that they just released to where you can keep a certain style a little bit more consistently than just having to prompt it.

Yeah, that stuff's, that stuff's really, really interesting. Something I thought of, like you imagine in their native app, being able to add your color scheme or something like that into that, that would be really cool. So you don't even need, it's kind of like custom instructions a little bit then.

Catchivity, but it's kind of saved. I think that would be extremely cool. I don't know if that's something that you've seen or that's something you have seen others other wish for. I mean, I w I would love that at the same time. I would like us to be able to upload like a photo of myself to use that photo in the image and not being able to go through the process you just.

Went through that. That's a little bit cumbersome. I mean, it still works. I think it's cool that you can do it, but I think it's going to be a lot easier at some point. I don't know when, but hopefully in the near future. It's a little hacky, you know, I guess that's kind of like, you're just like, you're kind of playing with the system to get to what you want.

Maybe someone knows how to do it better than me. Probably. I'm sure. But if you know better than the Midjourney master, let us know in the comments, but, you know, I, I think this has been really valuable. I mean, I really enjoyed the session. Do you have any final words of wisdom regarding AI Midjourney?

Anything else you'd like to share here? And also where people can connect more with you or. Yeah, man, I think there's the one thing I will say is like, I don't want to be all doom and gloom about this stuff, right? I think a lot of people there's a lot of like doom and gloom Okay, I was gonna take everyone's job and whatnot But really what it is is just staying up with these tools as they develop I think this is the easiest way to get into a better position You know, my analogy is always if you gave an iPhone To your grandparents and said, we'll learn it.

They're probably like, this is hieroglyphics, right? Like, I don't even know what this is. We've known it because we've had every iteration of it. We can pick it up and use it today. Same theory applies with these tools, right? It's going to be easier to learn them when they suck.

And when they get better, you know, you're going to be a professional with them. So learn it as it develops, instead of trying to get it all when it's like the iPhone now, like, and then trying to figure that out. So that's my little word of wisdom is just like, stay up to date, use them, test them.

Try to put it, you know, try to implement them, but yeah, other than that, like I'm on a, you know, I'm going to be on LinkedIn and if anyone wants to chat, I'm always on LinkedIn. I'm on Twitter. Uh, you know, roar underscore fly there, but yeah, that's where I hang out. And if anyone has questions, you know, always, always down to chat and see what, you know, see what we can come up with.

Yeah, I'll link up the resource Rory mentioned as well in the QR code, but I drop a link in the description and also to where you can connect more with Rory on LinkedIn. Definitely comment below if you have any takeaways from this session, anything you'd like us to cover more in the future and subscribe to the channel and like the video if you want more like this, and I'll see you in the next one.

Ciao for now. Bye.