Bots Can Create Almost Any Image. What Real Estate Agents Should Know

There’s a lot noise on the market on the best way to navigate a difficult market. This April, let Inman assist you to lower via the muddle to make sensible enterprise choices in actual time. All month lengthy, we’re taking it Again to Fundamentals and discovering out how actual property execs are evolving their methods and investing personally and professionally to drive development.

In late March, the world gasped after images emerged exhibiting police arresting former President Donald Trump.

In one of many images, officers grip Trump’s arms and lead him away. Others present Trump working from police, and nonetheless extra photos depict Trump’s members of the family wanting on in misery. It was a scene not like something that had occurred beforehand in American historical past.

And because it seems, it nonetheless hasn’t occurred.

That’s as a result of the images weren’t images in any respect. As a substitute, they had been the handiwork of a synthetic intelligence picture generator referred to as Midjourney. The photographs first appeared on Twitter, earlier than going viral on different platforms comparable to TikTok. Initially, many customers assumed the pictures — which at a look seemed plausible — had been actual and subsequently shared them as such. (Trump’s precise arrest a pair weeks later was a far much less dramatic and photographed affair.)

However earlier than lengthy, eagle-eyed viewers observed errors within the photos and shortly dozens of reports shops wrote posts debunking them.

However the episode however captured the superior, terrifying and sudden energy of synthetic intelligence picture mills. Seemingly in a single day, peculiar folks gained the flexibility to conjure realistic-but-fake imagery that previously might need required intensive coaching or a Hollywood price range. And whereas the Trump images had been extensively seen as a cautionary story about misinformation, others have used AI instruments to create high-quality artwork, illustrate youngsters’s books, and assemble skilled branding supplies.

All of which is to say that AI picture mills at the moment are attaining unprecedented ranges of sophistication, and are thus poised to disrupt industries comparable to actual property. That actuality raises a number of challenges, however it additionally means actual property professionals are about to have new alternatives to streamline their work, prolong their attain and combine know-how in ways in which even months in the past had been unimaginable.

In that mild, right here’s what you should learn about AI imagery.

Desk of Contents

Why is there all of the sudden a lot buzz about AI-generated photos?

As Inman beforehand reported, the present frenzy over AI started in December when know-how firm OpenAI debuted the free, public model of its chatbot, ChatGPT.

OpenAI has equally pushed the explosion of curiosity in AI picture mills. The corporate first introduced its DALL-E software — named for artist Salvador Dalí and Pixar robotic WALL-E — again in 2021. It debuted an upgraded model referred to as DALL-E 2 final spring, then made that software extensively accessible to everybody in September.

A handful of competing picture mills have popped up alongside the best way, however DALL-E’s debut and development alongside ChatGPT is the largest driver of consideration to the sector.

The large three picture mills

A lot as is the case with chatbots, there are quite a few AI-image mills on the market proper now. However in latest months, three have emerged because the dominant gamers. Inman examined these three by asking them to create a picture primarily based on the next immediate:

a home on a hill surrounded by flowers, with the milky method galaxy seen within the background

The aim of the experiment was to check the AI, however on this case Inman was additionally making an attempt to provide header photos for a latest collection of tales on the spring market. So it wasn’t a wholly tutorial train.

Right here’s how the experiment went:

Secure Diffusion

Secure diffusion is by far probably the most user-friendly of the big-name picture mills. Customers don’t have to sign up or join. You merely go to the web site, enter a immediate and wait seconds or minutes (relying on how busy the location is). Higher nonetheless, it’s free and there aren’t any limits on the variety of prompts customers can enter. Because of this, Secure Diffusion is an effective entry level for anybody wanting to check the AI waters.

Listed here are the photographs Secure Diffusion produced from that immediate:

The photographs above are usually not horrible per se, however for those who look intently they’ve an uncanny valley high quality to them. And so they spotlight one in every of Secure Diffusion’s massive tradeoffs; the platform is free and simple to make use of, however its photos typically aren’t as spectacular as these of some rival platforms.

Fortunately, Secure Diffusion has a bit of its web site dedicated to efficient prompts, so it’s simple to get used to the platform and determine the best way to generate higher photos.


Dall-E 2 continues to be most likely the best-known AI picture generator, and Inman has used photos from the platform a number of occasions as headers for tales. Right here’s the way it responded to the immediate:

The software is a strong one. That mentioned, the outcomes above aren’t terribly inspiring. They’re not fairly as uncanny as Secure Diffusion’s photos, however they’re additionally drab and darkish. That’s to not say all of DALL-E 2’s photos are drab — this Inman story includes a vivid drawing-style picture of rockets that the platform created — however the photos don’t simply prove spectacular if a person doesn’t determine the best way to write efficient prompts.

Customers have to enroll to make use of DALL-E 2. The positioning operates utilizing “credit,” with every new immediate costing one credit score. Most customers obtain 15 free credit every month, which isn’t sufficient to do a lot experimenting, although relying on when an individual indicators up they could additionally obtain an preliminary bucket of free credit. The positioning presently prices $15 for 115 further credit.


Midjourney might be the buzziest AI picture platform, and for good motive: It persistently generates unimaginable pictures. The photographs it got here up with for the immediate had been far and away higher than what DALL-E 2 or Secure Diffusion created:

After some tweaking through follow-up prompts, Inman in the end used Midjourney’s photos for the headers on the spring market collection.

Midjourney photos are additionally adequate that they persistently go viral; along with the Trump episode, there was additionally the so-called “Balenciaga Pope” photos that made the rounds on social media final month.

However Midjourney’s draw back is that the training curve to get began may be very steep. The platform just isn’t really an internet site by itself, however quite works through Discord, a social and messaging platform. So, would-be customers first have to enroll in Discord, select a “server” by which to take part, determine the essential code to get the bot to begin working (prompts should be preceded by “/think about”), after which determine the place these photos are going to indicate up.

When you get used to the system, it’s pretty simple. However Inman’s first a number of makes an attempt to make use of Midjourney concerned a number of Googling of directions.

Midjourney has traditionally supplied a free trial model of its platform, however in latest days has lower off the free model, citing extra demand. So a further a part of the training curve now entails signing up and paying for a subscription as properly. As of this writing, Midjourney’s least expensive tier begins at $8 monthly. The costliest tier prices $48 monthly.

How can brokers use AI-generated photos in actual property? 

Whereas chatbots have apparent purposes in actual property, AI photos are concurrently extra thrilling however more durable to deploy. Brokers, in any case, don’t essentially want authentic imagery of solely synthetic environments. That mentioned, listed here are a couple of of the methods trade execs are utilizing this know-how:


Headshot era — for which there are a variety of suppliers — is likely one of the extra standard methods brokers are utilizing AI.

Lauren King — an agent with PureWest Christie’s Worldwide Actual Property in Whitefish, Montana — is amongst those that have given the idea a attempt, telling Inman she used the corporate Attempt it On. The service asks customers for 10 to twenty authentic images, then after a day or two sends again about 100 AI generated headshots. The service value King $17.

These are two of the unique photos King despatched to Attempt it On:

Credit score: Lauren King

Of the greater than 100 photos Attempt it On created, King mentioned there have been a handful that she favored and may really use.

“I might say I might really feel comfy utilizing 5 or eight,” she mentioned. “They’re those I feel nobody would query.”

Listed here are two of these photos:

Credit score: Lauren King

The photographs Lauren shared with Inman present that Attempt it On added heat, golden hour lighting and a shallow depth of area (which means the background is out of focus), amongst different issues. They appear skilled.

However the outcomes additionally assorted, with King saying that most of the images both seemed nothing like her, had unusual lighting, or had been in any other case unusable.

Listed here are a few the much less profitable examples, which did not precisely seize quite a lot of King’s facial options. The second picture particularly additionally has an oddly unreal high quality.

Credit score: Lauren King

Regardless of a few of the weirder renderings, King spoke positively of her expertise.

“It was a enjoyable experiment,” she mentioned, including that it was a better course of than capturing precise headshots. “I discover photograph shoots and headshots to be sort of aggravating.”

Deena Serna agreed that capturing headshots will be “a ache” and informed Inman that she too consequently used Attempt it On for headshots. Serna, a Compass agent in Vero Seaside, Florida, mentioned that of the 100 or so photos she acquired, many “got here out simply tousled,” with unusual wanting eyes and tooth. However like King, she acquired sufficient usable photos to make the experiment price it.

Deena Serna

“There was a handful that turned out fairly good,” she mentioned. “I’ll positively use them. The resemblance to my actual pure self is satisfactory.”

Such sentiments are circulating quickly via actual property social media teams. And whereas the weirder outcomes spotlight AI’s present shortcomings, that is additionally one of many main frontiers by which synthetic intelligence-generated actual property content material is definitely making it out into the true world.

Renderings and visualizations

One other buzzy method that persons are deploying AI visualizations in actual property is through renderings of buildings. The photographs under, for instance, are AI-generated pictures of townhouses.

The Twitter person who generated these photos describes them as having “immaculate design,” and it’s exhausting to argue with that evaluation. The person additionally means that it’s solely a matter of time earlier than such photos will be enter into the software program that architects use to design precise buildings.

From there, it’s additionally not an enormous stretch to think about a full AI-based constructing pipeline that begins with fairly footage and ends with 3D-printed buildings.

Dave Jones

That’s a speculative end result, however already actual property professionals are imagining methods to deploy this sort of visualization know-how to clean out the house search course of. For example Dave Jones, co-owner of Windermere Abode in Tacoma, Washington, informed Inman he might think about utilizing AI to assist shoppers higher visualize and talk what sort of property they need.

“What if I used to be serving to somebody who wasn’t there they usually might clarify to me what the home they wished seemed like,” he mentioned. “So now I’ve a visible of what you’re searching for.”

Kent Czechowski, chief information scientist at OJO, informed Inman that these instruments may additionally enhance current actual property visualization and digital staging instruments.

“For instance, AI instruments can let a homebuyer think about what a presently empty lounge would appear to be if their furnishings and residential decor had been current, or assist a potential vendor see what an enhancement to their property would appear to be and even value,” Czechowski mentioned.

Various actual property corporations are presently experimenting with enhanced AI know-how. Inman will dive into such instruments in a later publish.

Advertising content material

Actual property professionals may use AI of their advertising, in response to Nick Niehaus, co-founder at actual property coaching agency Enterprise Video College.

Nick Niehaus

“It must be utilized in advertising, and I feel it must be used even at the moment,” Niehaus mentioned. “Advertising is all about saying one thing completely different than your competitors and that’s one thing the instruments enable us to do extra effectively.”

Niehaus has spent latest months experimenting with completely different AI platforms, and mentioned that instruments comparable to Midjourney and DALL-E supply each pace and originality benefits. They will additionally assist brokers create a cohesive visible advertising marketing campaign that interprets throughout mediums.

“You can also make much more photos lots quicker,” he added. “You’ll be able to have a postcard marketing campaign that drives them to a touchdown web page and results in a collection of emails. And you may have all of that basically in minutes.”

Czechowski additional famous that AI might let actual property professionals higher goal their advertising to their shoppers.

“The proliferation of content material will result in an explosion of hyper-localized and highly-targeted content material,” he mentioned. “Customers will anticipate an expertise tailor-made to their private preferences, communication types, and timelines.”

Whereas such a content material is a innovative AI enterprise utility, it’s obvious that it’s additionally gaining popularity: As a result of Midjourney operates on Discord, customers can see every others’ prompts and outcomes, and combined in with the limitless fantasy landscapes is a lot of firm logos and different branding photos.

Tackling the training curve

Whereas the potential of AI picture mills is important, it is usually not essentially simple to make use of them. Or not less than, they don’t robotically produce nice photos.

Niehaus, for instance, mentioned that with a view to generate the sort of complete AI-based advertising marketing campaign talked about above, a person may have to enter wherever from 20 to 50 completely different prompts. And people prompts want to make use of language that the AI platform understands.

“A number of of us they’ll put in a single, or possibly a pair prompts, and be like ‘I’m not getting what I would like,’” he mentioned. “The way in which you speak to Google, to seek for one thing, I feel that’s a superb analogy for what we’re studying right here.”

Amongst different issues, Niehaus mentioned that picture mills usually need customers to enter as few phrases as potential. Moreover, the order of phrases issues, with these firstly of the immediate exerting larger affect on the top end result.

“You don’t wish to converse in full sentences,” he added, drawing a distinction between picture mills and chatbots, the latter of which are inclined to excel at human-like speech.

Niehaus additionally suggested customers to suppose critically in regards to the product they wish to produce. He recommended, for instance, that folks with a pictures background inform the picture mills what sorts of lenses and lighting temperatures they’re searching for in photos. Individuals who have a specific artist’s type in thoughts ought to embrace that artist’s title within the immediate.

“You may say, ‘brownstone home, feminine actual property agent in entrance, cloudy day, 5600k lighting,’ and embrace a sure sort of lens,” Niehaus defined, including that studying to jot down efficient prompts takes time.

Niehaus’ ideas, nevertheless, increase a delicate however essential distinction between the best way AI photos work in comparison with extra standard searches for pictures. Up to now, somebody searching for a picture to placed on an internet site or piece of selling might need visited a photograph gallery, seen a wide range of disparate pictures, after which determined what to make use of from amongst these choices. In different phrases, a person may begin with solely a obscure thought of what picture they need, after which make a decision in the course of the search course of.

Nevertheless, producing a picture through AI is solely completely different. As a substitute of utilizing obscure search phrases, customers need to have a particular thought already in thoughts of what they’re making an attempt to generate. So, the second at which a person decides on a picture occurs a lot sooner within the course of, earlier than the search. It entails, in different phrases, a inventive course of extra akin to portray than to scrolling via a gallery on Getty Photos.

Huge dangers and massive rewards

As is the case with chatbots, the rise of AI picture mills has prompted quite a lot of questions on the way forward for work and mental property.

For instance, one contentious difficulty surrounding AI picture mills revolves round how a lot their merchandise pull from current work. In some case, AI photos have even included what appears to be like just like the signatures of actual life artists, prompting issues that the bots are plagiarizing — and consuming up the enterprise of — actual folks. Such issues have since prompted a number of lawsuits, and it stays to be seen how mental property legal guidelines may in the end apply to the gamut of AI artwork.

These points might impression how a lot actual property professionals lean on AI imagery in the long run. Within the shorter time period, AI’s reliance on current paintings may make it more durable and more durable to generate actually authentic content material — or not less than for learners to take action with out critical prompt-entering chops.

Past mental property, there’s additionally dangers in relation to misinformation. The Trump arrest photos spotlight this difficulty, however Niehaus additionally famous that “deep fakes,” which might superimpose one particular person’s face onto one other’s physique, increase quite a lot of moral issues. Niehaus moreover famous that there’s the potential for AI to interchange jobs, comparable to those who have historically revolved round content material creation.

“It appears to be like like a number of these instruments might be outfitted to interchange white-collar jobs greater than blue-collar jobs,” he mentioned.

Niehaus even imagines a future by which customers work together with AI-based actual property brokers.

“Should you can mix a chatbot with a dwell video that appears like an actual particular person, even for those who inform folks it’s AI, I do suppose there’s a section of the inhabitants that’ll be okay with that,” he mentioned.

That will sound ominous, and certainly for some brokers it might be. As extra AI instruments are capable of work collectively, their collective attain and potential disruptive qualities will get larger and larger.

However Niehaus additionally framed the current shift as a possibility. Brokers who learn to use AI may be capable to prolong their attain to far off areas by providing customers in new markets a chatbot model of themselves. And whether or not that particularly occurs or not, Niehaus’ level was that brokers who fail to discover AI are on the best danger, each from the know-how itself and from their human rivals who be taught the ropes.

“That’s why I’m actually emphasizing the concept of studying the instruments,” Niehaus mentioned. “We will’t predict precisely the place it’s going to go, however there’s a number of worth in brokers actually experimenting with it proper now.”

E mail Jim Dalrymple II

Back To Top