Supercharge Your Real Estate Ads with AI-Generated Images
In this episode of the Real Estate Growth Hackers, Zach, and Charlie have shared some incredible insights on how AI can be your secret weapon for creating compelling and eye-catching visuals for your real estate listings. They’ve unlocked the power of AI tools to craft images that not only grab attention but also convey emotions and ideas, all without breaking the bank on expensive professional photo shoots.
Now, you might be wondering how to get started with this game-changing approach. Well, the experts recommend beginning with concepts that resonate with your local market, such as waterfront properties, modern designs, or that cozy farmhouse aesthetic. Describe these dream homes in vivid detail, making sure they fit the unique vibe of your geographic area. Then, let AI image generators like Midjourney work their magic, producing various versions of these dreamy properties. Experiment with different lighting, impossible scenarios, and artistic styles to find what clicks with your audience.
The secret sauce here is having a treasure trove of creative image options to choose from, ensuring you can always keep your audience engaged. Zach emphasizes that AI makes this scalable, providing each real estate agent with unique visuals for their campaigns. No more ad fatigue and your costs stay in check.
In a nutshell, AI imagery is your ticket to capturing attention in the crowded real estate market, reducing expenses, and enabling you to fine-tune and test your ad campaigns. So, don’t miss out on this fantastic opportunity to supercharge your real estate ads with AI-generated images. Tune in now and get ready to unlock a world of possibilities!
Other subjects we covered on the show:
- How AI systems generate new images based on training data.
- Using ChatGPT to brainstorm ad image ideas.
- Differences between AI models like Midjourney and Stable Diffusion.
- Applications beyond ads like social media, websites, and direct mail.
- Importance of learning through testing what images will attract clients.
AND MORE TOPICS COVERED IN THE FULL INTERVIEW!!! You can check that out and subscribe to YouTube.
If you want to know more about Zach Hammer and Charlie Madison, you may reach out to them at:
[00:00:00] Zach Hammer: Welcome, welcome, welcome to another episode of Real Estate Growth Hackers. Today, we’re going to be talking about how you could supercharge your real estate ads with AI generated imagery. This is a game changer. It’s going to be really powerful for you guys. I got Charlie Madison here with me again, Charlie, welcome to the show.
[00:00:19] Charlie Madison: It is great to be back. I am really, really excited about this because I’ve seen some of the images you’ve created and I’m like, I want that. I want that for my real estate practice.
[00:00:33] Zach Hammer: Right? Right. Yeah. We’ve been, shoot, honestly now that I think about it, most of these images that I’ve generated that I’ve like, when I started down this path, you were one of the first people that I started showing like, man, look at what’s possible with this. Cause I knew you’d appreciate it.
[00:00:46] And I knew that you could see the the power of it. Like part of when we first got started, I was showing you how I could generate, very specifically, cool images of homes that look like they’re from your area and I could just [00:01:00] generate them with AI. So I was generating you know, Nashville looking homes or, you know, homes that look like not only homes that you’d find in Nashville, but look like they might be in the Nashville area and the suburbs and, you know, out in the more wooded areas, and and be able to create some really cool images.
[00:01:15] And so that’s what we’re going to be talking about today. We’re going to be talking about that concept and how it works and how do you leverage it and all of that. So, yeah, let’s give people a little bit of context.
[00:01:25] What I mean when I’m talking about some of these things. So, AI generated imagery. What we’re talking about there is we’re talking about images that are 100% completely new and unique that are created by software that creates these images out of thin air. There’s a lot of different technologies that do this.
[00:01:41] You don’t completely need to understand it, but I think it’s worthwhile to have a basic understanding to know how you end up with these images so that you understand a little bit about what’s possible and what might not be so possible. And the basic idea, there’s a few different tools that do this.
[00:01:57] But one of the concepts is that, that they [00:02:00] trained an AI to be able to recognize a whole lot of different images, right? So you give it a massive amount of images with a bit of documentation saying, this is what’s in this image. This is the style that it is. And all of that, that was done through these scraped databases of all this different stuff.
[00:02:15] So, you train them to recognize that and then what they did is they essentially generate noise, and try and modify that noise continually until it looks closer and closer and closer to something that represents some sort of text. So, if I’m generating say I want to generate an apple I would say, you know, Hey, I want you to generate an apple and it’s going to generate this noise field and try and change that noise continually until the result looks like an apple.
[00:02:44] And it gives each picture like a rating of how close to being an apple this image is, and it keeps modifying it further until it inches closer and closer and closer to being actually something that looks like an apple based on apples that it’s been trained on. And [00:03:00] so, that’s the basic concept.
[00:03:01] There’s a lot of different ways that, that works, but essentially, it’s starting from nothing, but able to kind of create from that by by having been trained on previous things. So, if something could be described and could be represented in pictures that you’d find, you know, freely available online or historically chances are those concepts could be recreated.
[00:03:24] Now it’s going to have a hard time creating anything that has never existed, or we don’t have any visuals to describe, right? If there’s not lots of representations of that, it’s going to be hard for it to do that. But a lot of these models are getting trained to be able to remix things well, right?
[00:03:38] So, you know, like, one of the examples that you’ll often see in some of the demonstrations for this stuff is like the AI might be really familiar with what an astronaut looks like, right? And it might be very familiar with what a horse looks like you may not have many examples of an astronaut riding a horse that currently exists, [00:04:00] but AI can be used to combine those ideas so you could generate an image of an astronaut riding a horse, even though that’s not a common thing that you would find lots of imagery of it’s able to take kind of from its knowledge bank of those concepts, and merge them together.
[00:04:13] And so similarly, anything that you could think of that, that is highly representative, whether it’s specific concepts like animals people that sort of thing, or even specific styles, like, you know, paintings in the style of Van Gogh, or Vector based arts, or hand drawn illustrations, or photography done by professional photographers versus amateur photographers, so really the sky’s the limit in terms of what’s possible if it exists somewhere on the internet and in decent quantity then chances are that AI has been trained on it to be able to represent things in those ideas, and remix them a bit.
[00:04:47] So, that’s the basic idea of how that works. Is any of that unclear or confusing?
[00:04:51] Charlie Madison: That makes sense. Kind of the way that I like to think about it is, you know, what ChatGPT has done is it’s learned [00:05:00] all of this information. And now it takes all the information and can create something brand new. And so what images have done is they’ve learned all these different images and now they can create a brand new image based on your description.
[00:05:20] Zach Hammer: Exactly. Exactly. And there is some skill set to this in terms of knowing how you need to craft your language in order to get the result that you want. So as for instance You typically want to craft your image prompts. You know, the words that you say, I want you to create an image like this.
[00:05:35] You want to focus on very visual language as much as possible. Or if you’re going to go with emotional language you need it to be emotional language that is commonly represented visually. If that makes sense. So, happy and sad you know, dark like, those are, you know, more emotional language that do often have like a style of imagery that goes along with it.
[00:05:57] But like, I don’t know, I’m trying to think [00:06:00] of, the more that you get into it, you’ll probably find, like, if you’re trying to create something and it’s not coming out right, or it’s not representing it right, it’s, you’re probably using words that are more conceptual rather than visual and that can cause problems for you not getting at least what you’re hoping for. And so, there’s that other problems that you can run into with this. Trying to combine too many different ideas into one thing.
[00:06:22] These AI’s could struggle with that. So, if you’re trying to get it to generate a left-handed cowboy wielding a lightsaber while doing the Macarena you know, on a spaceship that’s floating in the ocean. Like it trying to generate all those different elements, it’s going to have a hard time making them all come together in a way that makes sense.
[00:06:45] And you often get results where you look at it and you’re like, Oh, I kind of see how it’s thinking all these elements are here, but it doesn’t actually really make sense and it’s not what you were picturing. So there are strategies for how to do stuff like that. And so typically I find You [00:07:00] want to keep the elements fairly simple where it’s like you describe one, maybe two things that are in the image and then you can incorporate a style, maybe some colors, that sort of idea, but you don’t necessarily want to try and cram too much into one of these there are strategies for how you can incorporate getting more things into them, but that’s more like sequential.
[00:07:17] Where you, you know, merge images and stuff like that, and that’s more advanced, but yeah, so those are some of the things, you know, that kind of go into this concept, but let’s talk about like, why, why does this even matter, right? Like, why do we want to generate images? Why why does this help?
[00:07:29] Why does this help us achieve a better end results? What, you know, like the technology is cool, but why does this even matter? So, there’s a couple of key areas, especially in creating ads, and creating online advertising where we’re being able to generate these images is really powerful.
[00:07:43] So, first off foundationally for having success with your online ads, you actually need to have a ton of different ads just to find which ones work. Right. You need to have a lot of different options. You want varied options. More often than not I mean, you run a lot of ads.
[00:07:59] Charlie, I know that [00:08:00] you’re familiar with this. I have found that like, when I spend a bunch of time putting together the perfect ad and I’m like, man, all of my wisdom and intelligence is in this thing. I’ve picked the perfect image and I’ve got the perfect copy and I’ve got that. When I compare the results of me trying to do that versus.
[00:08:19] Let me just put up like five to 20 different ads that I haven’t really thought about a bunch, but they’re all different and unique five to 20 different ads always outperforms my one that I thought was perfect every single time.
[00:08:36] Charlie Madison: Never know. We never know which one it is.
[00:08:41] Zach Hammer: Exactly, exactly. That’s the biggest thing. Like, it’s very humbling to realize the only thing that like, like the smarts that you need to have is you need to have the smarts to know you need a lot of variation.
[00:08:51] And then the market sort of decides what it actually is going to engage with. So in order to have success with these, really, that’s the key is trying to get a lot of [00:09:00] variation so you can actually see what hits and what doesn’t, and more often than not, it’s about getting out of your own way to be able to have that varied creation happen, which AI is great at and we tend to not be, we tend to have a hard time coming up with 20 unique ideas.
[00:09:13] We might find 20 very similar ideas that are minorly changed, but it’s hard to come up with 20 unique ideas. The other thing that we want to do is, we want to make sure that we’re creating images that actually stand out, that we’re leveraging imagery, whether it’s something that we’ve created or whether it’s something that is just using stock imagery, regardless.
[00:09:30] You want something, that gets noticed, that stands out in a sea of everybody else using the same images, the same stuff, the stuff that blends into the background, and just doesn’t stand out. And and that’s one of the keys here is, being able to find something that is actually interesting enough to get engagement.
[00:09:49] And that’s, You know, that could be hard when you’re just trying to operate from your own creativity, but when you could generate whatever comes to mind and the sky’s the limit, and this isn’t expensive, like, man, it’s a lot [00:10:00] easier to stand out and and create unique images. Right.
[00:10:03] Charlie Madison: Yeah.
[00:10:04] Zach Hammer: And so with Good.
[00:10:07] Charlie Madison: you know, it used to be it was popular in real estate. We had the ugly yellow signs and it was. Yeah, you know, and the whole purpose was that you get past banner blindness, you know, you get past what everyone sees. So to have something unique, something that’s different, I mean, that’s the whole idea cause our brains are made to filter out what’s familiar and look at what’s new.
[00:10:33] Zach Hammer: Right. And there’s lots of ways to do that. I mean, you could throw up an Ad that’s just like a big red image with the word “Stop” written on it. Right. And that would probably get people to like, stop and look at it and shoot. Like I kind of want to test that now actually. But it may not be relevant to whatever you’re trying to advertise.
[00:10:51] And it may not stop the people who are interested in the thing that you’re trying to engage with. And so, there’s always this blend and balance of, it’s not just about being noticed. It’s about [00:11:00] being noticed in a relevant way that sort of fits with what you’re trying to do.
[00:11:04] And it can be hard to find that nuance, right? Where it’s like, well, I need to find people who are interested in houses, but I need to stand out against the other stuff. So I can’t just like, throw up. You know, images of scantily clad women and think that that’s going to be, you know, enough to attract people who are interested by a house, right?
[00:11:20] So you want to find something relevant and not just that stands out and and that blend is part of what we’re looking for here .
[00:11:27] The other thing that we’re looking for is we’re looking for images that meet these qualifications, right? We want to be able to have a lot of them. We want to be able to have them stand out and we want to be able to do that without breaking the bank.
[00:11:38] Right? So I know you were talking about earlier. It’s like, there might be some great images in real estate, pictures of homes do really well pictures of homes that look beautiful do incredibly well as kind of the foundation for some of your creative imagery on, you know, on Facebook and on these advertising platforms.
[00:11:54] But if it’s not your listing and you don’t have the right to use that image, you’re not allowed [00:12:00] to, right? You can’t just decide to use somebody else’s image and get away with that legally. And so like, has that been something that you’ve seen where you’re like, man, this would be a great image. I’d love to use that, but it’s not my listing.
[00:12:11] Charlie Madison: Exactly. Exactly. Yep. So I’ve spent a lot of time image searching for royalty free, and you know, most of the royalty free images are not good . Right. And you know, I don’t want to pay Getty images for, you know, a million dollars for a picture of a front door. And so yeah, it’s.
[00:12:34] How do you get images finally? This is better than searching royalty free for an hour.
[00:12:42] Zach Hammer: Exactly. Exactly. Yeah. Because you could say very clearly, like, I know I want this image. I know what I want it to look like. I know what emotions I need it to convey, or how I need it to stand out. And yeah. You know, finding an image that already exists that perfectly matches that and doesn’t break the bank can be hard. Again, especially considering we want to get a lot of [00:13:00] variety in order to make these ads work. Right. So yeah, that’s definitely a key problem that we could solve with this AI imagery.
[00:13:06] And then, the other thing that, you know, especially for this audience people who have have, you know, agents that they work with, maybe you lead a team and you want to be able to supply your agents with tools and assets they can use in their own marketing to be able to do this at scale where you could say, how can I help my people be able to do the same things that work for me, but not just have everybody literally rubber stamp, copy paste, everything that I’m doing.
[00:13:30] Cause that degrades the effectiveness, right? That increases the level of sameness. And especially if you’re in one market, you’re going to have a bunch of people using the same images over and over and over, it could be useful to be able to say, Hey, I know this style of image works well, but let’s create a unique one.
[00:13:48] Or a unique variations or teach even my team how to create unique variations so that not everybody is using the same thing over and over and over. So you actually get the benefit of knowing what works, with the [00:14:00] ability to create a unique version of what works every single time and that level of power for being able to scale up uniqueness while keeping the foundations, you know, the same that’s really powerful.
[00:14:11] I know you’ve led a team at various times how much of your struggle and working with the team is literally just giving them the tools to be able to do some of the things that you’ve found success with. Right?
[00:14:21] Charlie Madison: Yeah. And make it super simple, you know, cause you know, I’m super techie and like each person on the team has their own unique strength and weakness. And so to be able to make, give it, you know, have something that’s useful for the individual people, you know, without being just a low common denominator.
[00:14:41] Like it’s so simple to use that it’s not even effective or or whatnot. So like with this, you know, like how cool would it be if you could have a list of your IDX website for homes with pools and you’re able to generate a picture of a pool or a horse farm, [00:15:00] or, you know, like you could just think of the different groups.
[00:15:03] And I love the thought of, because it’s pretty simple. You got a template. You could give your agents the ability to do this. And that way they create each image would be completely different. And it would lead to their IDX website, which would be awesome.
[00:15:20] Zach Hammer: Absolutely. Absolutely. So yeah, let’s dive into kind of how you leverage this concept, how you flow through this, and be able to leverage this technology to get this result. So, Charlie was just talking about kind of one of the foundational styles of campaigns that this works for and that is when you’re talking about a concept, an archetype of the kinds of homes that people might look for in an area.
[00:15:41] That is a fantastic style of ad. You know, typically, each area has things that are more often desirable in the area. You know, you might have in parts of the country, it’s homes that back up to forested land. It might be You know, homes with a creek running through it, or it might be waterfront homes on a lake, [00:16:00] on the ocean, on, on something, right?
[00:16:03] You might ha have homes that are up in the mountains with mountain views. You might have homes that are very specifically desert theme, desert landscapes, right? You might like, like homes that, homes in Vegas obviously look very different than homes in Nashville. And being able to represent that can be useful.
[00:16:18] The key thing that I’ve always looked for in running those style of ads is that you want to think through what are the desirable attributes. What are people naturally looking for in this area? And being able to offer those up a single family home or a single story homes, homes with pools, all of those ideas, you know, your market and know the likely things that people are looking for in the area, go ahead.
[00:16:39] Charlie Madison: Homes with amazing kitchens amazing bathrooms, you know, people love looking at kitchens, bathrooms. Will it create one of like a nice master bedroom?
[00:16:49] Zach Hammer: Yeah, yeah, absolutely. So, anything that you could describe visually you could definitely create images of, and all of these things, what’s interesting is some of these elements may make sense for an ad and some of them won’t, [00:17:00] you’ll find that some things are going to generate leads more effectively than others.
[00:17:03] Oh, and then on the other thing or on the other element for some of these concepts, really what you want to be asking yourself is you want to be asking when the person looks at this. Do they say to themselves, yeah, that’s what I want, right? Like, that’s me. That’s what I’m looking for.
[00:17:17] That’s what I’m hoping to find information on. Other things that you do are our price points homes under this price point with these features, right? So, that’s a good one. And so, to generate these kinds of, you know, to be able to have the variety to generate these what you do.
[00:17:32] Is you start with those concepts and then the other key element that you want to leverage when you’re like leveraging a prompt like this into into an AI image generator is that you want to specify that it looks like it’s from a specific area because the key thing here is when somebody looks at it, they think, yes, this is what I’m looking for.
[00:17:53] And I’ll tell you what, if you are in Las Vegas and you’re running ads of amazing looking homes with [00:18:00] pools, and all the things that people are looking for, but it looks like it’s from Nashville. they’re going to look at it and they’re going to think that’s not what I’m looking for because I know I’m looking for a home in Vegas.
[00:18:10] Similarly, if you run the desert looking homes from Phoenix. And you run those in Nashville, people are going to look at that and think, that’s not what I’m looking for. Right? Like they’re just going to pass on by. Cause it’s not the concept that they’re looking for. Or similarly, if I’m somebody that’s looking for a high price point home and you’re showing me an image of a home that it may look nice, but it doesn’t look like a luxury home that I’m looking for.
[00:18:35] I’m going to say that that’s not the fit for me. Or the other way around, if I know that I’m looking for a home in kind of this starter price point, and you’re showing me something that is obviously like a multi million dollar mansion. I might find that cool, but I may not click on it because it doesn’t feel like the kind of thing that I feel like is what I’m looking for.
[00:18:55] And so all of this is about being able to really zero in and say, put [00:19:00] yourself in the shoes of the person looking at your ads and be able to say, this shows you not necessarily exactly what you’d get when you’re buying, but like an idealized version of whatever it is you’re looking for.
[00:19:15] You don’t want to show, you know, somebody looking for a starter home, the rundown house that doesn’t look good. You want to show them the best looking version of what they might imagine would fit as a starter home. Does that make sense? It doesn’t need to be real or actually exist, but it needs to feel like that concept.
[00:19:32] Otherwise otherwise they won’t connect with it. And all of this is stuff that you test, right? You test these ideas to see what actually hits and what actually works. So, again, first off you’re going to start, you’re just going to write down and list out. What are the different things that people are looking for in my area?
[00:19:47] What are kind of these major buckets? You know, there’s lots of good starting points, you know, you could think through the price points that people are looking in. You could think for the features, a single story with pools those kinds of concepts. And you can take those ideas, you could remix [00:20:00] them, put them together.
[00:20:01] Charlie Madison: It knows even stuff like luxury, modern farmhouse or, you know, with land in a barn or like, it’s and it’s really, really good at the luxury homes. Like, it makes really nice luxury homes. And then you can also do, you know, another little tip, which you’ll probably get to.
[00:20:19] You know, you can say at dusk or at dawn, which I love that because, you know, you got that kind of that golden hour so to speak, the images are amazing.
[00:20:31] Zach Hammer: Yeah. And that’s actually, that’s one of the areas where you can really make the images stand out is that you can get the results of hiring a professional photographer without having to actually hire a professional photographer to get those images, and be able to do it at, you know, these tools are crazy cost effective.
[00:20:46] And even better than that, again, going to this end of, we want these to stand out and be unique. One of the things that I like to do is I like to actually generate images that are impossible to actually capture because they may not actually [00:21:00] exist. So, as of for instance, you can tell it that you want to have the Milky Way galaxy visible in the sky in the background.
[00:21:08] And I don’t know how much you know about astrophotography and doing that, but literally You know, the Milky Way is in specific places at specific times and visible sometimes of the year, and not others depending on or at least the positioning is going to be in some places.
[00:21:25] In the sky at sometimes a year and not others and so in order to get the perfect shot where like the galaxy is positioned just right over this home with these things in the background, it’s literally like there are a handful of days of the year where that might be possible and it may not be possible to capture it because the lighting in the area may not allow you to capture it or whatever, right?
[00:21:46] So like, you could create these scenarios that couldn’t actually be captured by a photographer and it would be impossible. But you could make them come into existence just by imagining them coming together and yeah, and generating these [00:22:00] beautiful nighttime photos that people haven’t seen before because it’s impossible to get and so you could do things like that.
[00:22:06] You could experiment, maybe it’s not even a photo of an actual home that works best. Maybe it’s actually a home that’s done in the style of Picasso that performs best. Right. And so there’s all sorts of different ways that you could test out these concepts to figure out what works.
[00:22:20] And so that’s actually one of the other things that I like to do in this process. You know, we’re talking about being able to generate lots of different variations. I don’t know about you, but me personally, I find it hard to come up with all those different ideas. And so I will actually leverage a different AI tool.
[00:22:35] I’ll leverage ChatGPT to give me ideas for these different image concepts. And I’ll say, I’m putting together an ad that is about this. That is about, you know, it’s going to offer you know, a list of single story homes in Southern California below this price point, please give me 20 unique concepts without repetition [00:23:00] for images that I could use to go along with this style of ad that are designed to stand out while still conveying the idea.
[00:23:08] Right? Like, so I’ll use a prompt like that and literally just have it give me… 20 different concepts, right? And then I could take those concepts and I often need to modify them.
[00:23:18] I actually have a prompt that I’ve put together that is very specifically, you give it a concept and it outputs a perfectly structured prompt for an image generator. So, I have a prompt that does that, but even without that, you might have to, you’ll look at the, you know, the ideas that it gives you.
[00:23:33] And some of those concepts are going to be. You know, you’ll see like, this isn’t really visual language, right? Like I get why they’re saying this but like, how do you like, as for instance, it’ll say something like an image of an orange that you could just tell looks successful. What does a successful orange look like?
[00:23:55] Right? Like, what does that mean? Right? And so, like some things they’ll be like that where [00:24:00] it’s like, I don’t know what a successful orange would ever look like.
[00:24:02] Charlie Madison: That sounds like the name of a punk rock album the successful orange by rancid.
[00:24:10] Zach Hammer: Right, right. Exactly. And so, you might have to modify them a little bit in order to actually generate the image where it’s like okay, what does a successful orange mean? So maybe instead of a successful orange, you would say something more like, an image of a perfect orange.
[00:24:25] That’s shining and glistening because it so good to eat or something like that. Right? And like, that would be the way that you turn that concept into, you know, visual language. And then you take that and you throw it into an AI image generating tool. Now there’s a lot of these, we haven’t talked about this yet.
[00:24:39] One of my favorites is one called Midjourney. Midjourney, as of at least the recording of this show, you actually work through, with Midjourney through Discord. So Discord is a chat app. It’s kind of like Slack if you’re familiar with Slack. If you’re not familiar with Discord where typically it’s used for communities to like communicate with each other it’s like a Old AOL [00:25:00] chat rooms, but in the modern age is kind of what what discord is?
[00:25:03] But, you actually, you use Discord in order to communicate with Midjourney to tell it to generate stuff. And that if you’re looking for information on how to use that, you could literally look this up on YouTube, how to get started with Midjourney. And it’ll teach you everything that you need to know about like how to work with it in that way.
[00:25:17] If you’re looking for something that has an easy to use web interface, there aren’t many that are as good as Midjourney right now, but that’s changing. There’s some that are getting closer. Stable Diffusion is one of the companies that’s making big headway in this. And they have some tools on their site that allow you to do a lot of this sort of stuff as well.
[00:25:34] I forget what they’re called, but if you look up Stable Diffusion and look up the tools that they offer they have some hosted solutions. There’s another one that’s called Leonardo. I think it’s Leonardo.ai is the website, but I Leonardo AI art generator. If you Google that, you’ll be able to find it.
[00:25:48] That’s another one that does some good work on this. That has a web interface and actually has a decent amount of like free credits that you could get in to try it out. Playground AI is another one that has a lot of cool tools. So, there’s lots of these tools out there[00:26:00] that I encourage you to play with it, experiment with to see what you could generate that fits in this vein.
[00:26:05] Over time, the best rules are constantly changing. We’re in a space right now where these things are developing very quickly, very rapidly, who’s doing it best and what the best options are changing incredibly quickly.
[00:26:16] As of right now, though, for me, Midjourney is one of my favorites. I tend to get the most consistent images that look good while also make cohesive sense. So, they don’t just have a feel and emotion, but like, you know, early AI images you’d have, like, if they were trying to generate hands, like people would have, like.
[00:26:32] 30 fingers or like, they’d have three really big fingers or something, right? Like, where they just looked off, it was like, you look closely and it’s like, something’s off about this image. Anyway, so, Midjourney at this point is most consistently able to generate solid images.
[00:26:48] Go ahead.
[00:26:49] Charlie Madison: I would say from what I’ve seen Midjourney is the best at getting past the uncanny valley, like a lot of the AI generated, [00:27:00] like you said, it’s off enough that it’s just like little wacky, just a little like disconcerting, but like the Midjourney images, a lot of times that like, they just look like a really great photo.
[00:27:13] So, it just looks legit.
[00:27:16] Zach Hammer: Exactly. And, luckily you know, some things our human eyes are very tuned to like recognize even the minorest thing being off anything that has a person in it. We see that even if something is just really simple and off, we see that really quickly without even like consciously detecting it.
[00:27:36] Other things don’t hit us quite that same way. A lot of objects don’t hit us that same way. So if there’s something a little bit off, we don’t pick up on that quite as quickly as we do somebody having 30 fingers or like, you know, their nose being the wrong shape or their eyes being just slightly off or whatever, right?
[00:27:54] Like we don’t notice that as quickly. Or we don’t know those objects as quickly as we do humans. So, depending on the [00:28:00] tool that you’re using, you might be able to strategize what you’re asking it to generate, to be able to more consistently get images that work well for you, but overall Midjourney handles all of these things pretty well, pretty consistently.
[00:28:10] And so then, yeah. So, then what you’re going to do so, we generate a list of concepts through something like ChatGPT we take those concepts, we feed them into something like Midjourney. Midjourney is going to actually output a number of variations for you. Each time you generate something so you’ll get like four different images.
[00:28:26] And then from that, you might have to remix them. You might have to regenerate them a couple of times, but then ultimately you’ll find like an image where you’re like, yeah, this makes sense. This is one that I like, or what I often find is that I’ll start generating something. And then that sparks a new idea for me where I’m like, oh, wouldn’t it be cool if this was there or if this was a little bit different?
[00:28:45] And then that’s a great way to you know, start down a path and be able to like mold it into something that is more your own and and really be able to direct it a little bit better. Just getting into the process of developing it gets you thinking about it more, thinking about it more deeply, [00:29:00] and coming up with some good ideas as well.
[00:29:02] And, and then, yeah, so, you know, you could take this concept and be able to roll with it from there. You could take the same image and generate it in all sorts of different styles, right? So, let’s go to that orange example. You could have a photo of an orange. You could have a drawing of an orange.
[00:29:16] You could have a painting of an orange. You could have an orange done up in neon lights. You could have an orange that looks like it’s generated by a computer. You could have an orange that looks like it was drawn by a five-year-old. You could have an orange that looks like it’s actually the earth.
[00:29:30] But it happens to be an orange. You can have an orange drawn by Picasso, Van Gogh Rembrandt. You can have an orange that looks like it’s a hieroglyphic. You can have an orange that looks like it’s a cave painting. Right? So, all of these different things. Any style that you could think of, you could potentially apply to your concept.
[00:29:49] So, in the real estate world some things that are specifically useful more directly, more obviously when you’re looking at different ways to generate houses golden hour, blue hour, dusk, [00:30:00] twilight, nighttime photography you could use words like with plenty of natural light, you can you know, talk about throwing the moon into the background, the you know, the galaxies into the background.
[00:30:11] You can you know, you could talk about like you said, you could say, you know, this is modern country farmhouse. This is, you know, strictly modern. This is this is mid century modern. All of these different styles of architecture that people know, like, and desire, you can take the same house, put it in the same location, but remix how that house is built.
[00:30:31] And all of these things can be used to get you plenty of variations to be able to run these styles of ads, run these style of concepts, be able to test them effectively. And even more importantly all of these ads end up fatiguing at some point, right?
[00:30:43] Like you can’t run an ad forever and have it keep working. People end up getting used to the ad. It stops working. You need something new, something different. But what’s cool about this is that, you could generate a new image that looks mostly the same and be able to keep an ad live, but yeah, so there you go.
[00:30:58] That’s you know, that’s how you would do [00:31:00] this how you would leverage it what I have found in my testing. These style of ads when we use these AI generated images, the fact that they are unique, literally for me at this point, nine times out of 10, they’re outperforming the standard images that I would use otherwise that these slightly off, slightly different, slightly unique images are drastically better performing because you can make them do that.
[00:31:23] And so, yeah, you can take this concept, you can run it with your ads, but don’t stop there. All of this learning that you’re going to be able to do through generating AI images like this we just hit the tip of the iceberg and applying it to ads.
[00:31:34] You can also take this and leverage it everywhere else that you need images. As long as you are not trying to create an image of a specific house or a specific person, even that at some point might become easier. But right now if you’re not trying to say, Hey, this is a specific home that I’m trying to sell, it’s not going to work for that.
[00:31:51] But what it will work is for people who are looking for types of homes or anytime that you’re looking to represent a concept or the feel of an [00:32:00] area where it doesn’t necessarily matter if somebody looks at it and says, yes, this is definitely Nashville. I could see the skyline that I’m used to like if you’re not looking to do that, but you’re looking for it to just feel like it could be Nashville, or it could feel like it’s a neighborhood in Vegas, or it could feel like it’s a, you know, homes in Florida or Washington or wherever.
[00:32:21] Right. When you’re looking for those types of concepts, you’re looking to create an emotion or a feeling AI is going to be great. Anywhere that you need that kind of imagery, whether it’s on your websites, your print materials, your direct mail, your social media graphics, anywhere that you can imagine, you can leverage AI for those concepts where you’re looking to implement a concept.
[00:32:39] And unlike before, you don’t have to pay somebody for that individual image, you don’t have to have exactly the right image you know, to get it going. You could very quickly test and iterate on images and ideas, and see what works. You’re only limited by your ability to be creative enough to either ask [00:33:00] ChatGPT to be creative for you or to have enough varied options of what those different things look like.
[00:33:06] And from there be able to generate effective leads, reduce your costs and be able to work with a team in the process. So, you know, there you go.
[00:33:13] Any final thoughts you think Charlie on on this concept and leveraging AI to increase your, your ad performance,
[00:33:20] Charlie Madison: I’ve heard, I think some places where people said, like, you want to put certain like ratios, or you want to mention a certain professional camera is that still needed to make these photos? Do you still put that type of stuff in there?
[00:33:37] Zach Hammer: So it’s just like anything a lot of these tools, AI tools, especially, they right now are more so a mirror reflecting how clear you are in terms of what you get back. And so if, you know, I want to achieve a very specific style of image. And can reference it’s [00:34:00] like, somebody who knows cameras will be able to tell you, yeah, this picture was taken by this kind of camera using this kind of lens with this sort of F stop, this exposure, this time of day, like they’d be able to be very specific.
[00:34:14] And if you want AI to give you back something that feels like that, that references that specificity, or looks like that kind of professional image. Yes, sometimes. You will get better results by being very clear and very explicit with, I want it to have these characteristics. But really it depends on the model.
[00:34:34] Midjourney at this point actually has a back end AI that’s running with it where it will try and improve your prompt for you so that you get a better more interesting image and it doesn’t show you that it’s doing that. But that’s part of what’s going on. You can actually turn that off if you want to so that you could be more specific about just using the prompt that you’re using.
[00:34:55] Stable diffusion is one of the other models, that one typically seems to perform [00:35:00] better if you tell it a lot of what you want and what you don’t want. Whereas sometimes Midjourney can run into problems if you get too descriptive one way or the other.
[00:35:08] So really like part of this is, there’s a little bit of a learning curve to seeing what gets you the result that you want. What I recommend for most people is either take the shortcut of looking up. You know, YouTube videos or training that other people are doing to get a similar result to what you’re looking for, or be ready for a little bit of a learning curve to say, I’m not quite getting what I want.
[00:35:29] Let me massage it a bit. Let me change my words up. Why is it giving me this when I’m asking for that? Could I be more clear? Could it be more specific? Am I getting better results that way? And there was a little bit of a learning curve to it. It’s not something that you should expect. You’re immediately going to get into it and output perfect images that are exactly what you’re looking for.
[00:35:44] But with a little bit of effort into it. You could very consistently get massive scalable output of exactly what you’re looking for. So, just like anything there’s a learning curve and yeah, sometimes specificity does matter. Other times it’s not as [00:36:00] important, but really you only see that by getting in and playing with it to really know for sure.
[00:36:04] Charlie Madison: Perfect.
[00:36:05] Zach Hammer: Does that make sense?
[00:36:07] Charlie Madison: Yep. That made a lot of sense.
[00:36:09] Zach Hammer: Awesome. So what do you think? What do you think for you? The difference would be implementing these kinds of AI images into a real estate practice. Where are the key areas where like, if you’re going to go out and use this today, where would you start?
[00:36:22] Charlie Madison: I mean, the easiest is types of homes for an IDX search, you know?
[00:36:27] Zach Hammer: Right.
[00:36:28] Charlie Madison: You know, if I can you know, there’s been times where I would say, here’s the best homes with pools that are on the market now, or where I could get a list of that.
[00:36:43] So, I mean, what I would try to do right now is have five of those a week, you know, I’ve got pool homes on Monday, you know, farmhouses on Tuesday, Wednesday is Lake homes Thursday would be homes with great kitchens you [00:37:00] know, just something to stay on top. And then the other thing is I would play with you know, there’s lots of free reports that I can reuse that I can find online, but I would play and see if I could come up with like a cool cover for it or something to make it look mine, that those are probably the two things.
[00:37:19] And then I really liked the idea so that it doesn’t get boring to, like try the different concepts hieroglyphics or Andy Warhol or, you know, different things like that.
[00:37:33] Zach Hammer: Yeah. Yeah. And I’ve been amazed what works and what doesn’t, and none of it could I have directly predicted. It’s more often you look at and you’re like, all right, interesting. The market seems to be responding to that. And and yeah, so I definitely, I’d encourage, I’d encourage people.
[00:37:47] I think this is a skillset in and of itself. Try not to prejudge what you think is going to work and give yourself enough of an opportunity to test a variety of things. As for instance, you [00:38:00] know how we were talking about the whole uncanny valley and all of that. You’d have to think through whether it makes sense for your market or not.
[00:38:06] But I’m running tests on an ad right now where one of our best performing images is one that definitely has that uncanny valley, like the faces look off, the hands look off, you look at it. And like, I jokingly in the comments of this thing, like I tell people like, hopefully the therapy as a result of this image, isn’t too expensive because I’m pretty sure we’re giving people nightmares when they’re looking at this, but it’s working well because I think it does. It sort of causes people to stop and look at it and it’s not offensive, but there’s something that like causes somebody to stop and look. And what’s funny is we actually get a lot of comments of people who are who are saying like, man, this image is really off.
[00:38:44] This image looks really weird. Like you need to improve your prompts. And cause like they could tell that it’s AI generated. I hop into the comments and I’m like well, we keep running it because it keeps generating leads at the best cost of any of our ads that are doing this. So like, Hey, the proof of the pudding’s in the eating.
[00:38:58] Right. So it’s you know, you [00:39:00] may think, Hey, I’ve got to fix this image in order to run it. And maybe not, maybe that’s exactly what you need to get an ad that’s working right, but you don’t know until you test it. So, definitely what you want is you likely want to generate leads.
[00:39:13] Without coming across dishonest, like all of those things, right? You want to protect your reputation while while doing what you need to, but trust the data of the results that you’re getting more than what you would assume is going to work.
[00:39:25] And that’s likely a way to get better results with this. So, there you go. There’s some thoughts. Hopefully. There’s some practical implementable ideas in there for you. If you guys have questions on this and you want more information, this is a an area that I’m looking at potentially putting together more training direct step by step stuff.
[00:39:42] That’s something that people want to reach out to us, RealEstateGrowthHackers.com. And we might be looking at putting together, some more information on this. If this is useful for you and you do put it into practice, definitely share that with us, let us know that you’re putting this into practice, what kind of results you’re getting.
[00:39:58] I find that this is really, [00:40:00] it’s the forefront of what people are doing with this. And if you’re getting out there and putting it into action, I want to hear about it. But yeah, otherwise that’s that’s what we got for you today. Hopefully you get out there get some results and leverage AI.
[00:40:10] There’s a phrase in AI right now. That is the concept that this is the worst it’s ever going to be. And when it’s this good. At it’s the worst it’s ever going to be or it only gets better from here. Right? Like, we’re on the horizon of some really interesting things that I think are coming with with this technology.
[00:40:26] And I think a lot of the skillset is just being ready to play with it, experiment with it and see what’s on the cutting edge, cause it’s going quick. And so I encourage you to do that, get some results from it. And there’s some practical ways that you can implement it. Oh, there you go. Any any parting words from you, Charlie?
[00:40:43] Charlie Madison: I think we covered it. I really like it.
[00:40:46] Zach Hammer: Perfect. Well, there you go. That is this episode talking about AI art for your real estate ads and how to get better results with them as results. This is the Real Estate Growth Hacker Show. I’m Zach Hammer. This was Charlie Madison with me. [00:41:00] Until the next one, we’ll catch you later.
[00:41:02] Charlie Madison: Bye.
Real Estate Growth Hackers Founder
Zach Hammer is the co-founder of Real Estate Growth Hackers. Over the last 36 months Zach and his team have managed ad budgets well over $100,000, generated over 25,000 real estate leads, and helped create over $50,000,0000 in business revenue for their clients. Zach is also a highly sought after speaker and consultant whose work has impacted some of the top Real Estate teams and brokerages across the country.