0
$0.00
0 items
I am 55 years old, which makes me part of Generation X. I've always felt that we were the luckiest generation because we were born and raised and educated and went out into the world of work at the start of this modern post-boomer silicon era when absolutely everything changed.
We were at school and college when the Internet was born and when the world-wide web was invented. We witnessed the arrival of the home computer and the games console. We were young when first music and then photography went from analog to digital and we were young adults when smartphones were invented.
We were there when suddenly you could pause live TV, when air travel was democratised, when social media was invented and just as we could legally drive, we got sophisticated digital navigation devices for our cars. We were there when Google search first came online and when they cloned the first animal and genetically modified the first food crop.
And now we're all middle-aged as the new A.I. era starts building momentum and everything changes forever again. As a generation I reckon we're pretty well placed to deal with what was once science-fiction and is now commonplace reality.
We need to talk about stock photography specifically and how A.I. is going to kill a substantial portion of the revenue of companies such as Getty Images.
There are, it seems to me, two kinds of stock photography - the specific and the non-specific. Specific stock photography would be something like a photograph of an actual location - Big Ben, the Grand Canyon, Machu Picchu for instance. Or an actual person - Barack Obama, Kanye West, Vladimir Putin for instance.
Meanwhile, non-specific stock photography is the generalised topic or themed based photographs used to illustrate blog articles or brochures. You might search for photographs of an elderly man pushing a walker down a suburban street in Liverpool. Or perhaps a group of young children enjoying a birthday party in the back garden of a house in Canada. Or a colourful sunset over a forested mountain range.
Non-specific stock photography will be rendered irrelevant within the next couple of years and I strongly suspect that specific photography will go the same way over a slightly longer timeframe.
The three best known generative art models currently available to the public are Dall-E, Midjourney and Stable Diffusion. Most of the public facing services such as Canva's new AI tool are based upon the APIs of one of these three models, but some companies such as Adobe have built their own. Between them, these models, each with their own strengths and weaknesses clearly point the way to a future when we can summon up literally any image we want with a short text query.
In many ways, we're already there. The advertising industry is already embracing generative art systems like Midjourney. Instead of hiring an expensive product photographer and some studio space and possibly a model or two - they can get what they want near-instantly with a text prompt in Midjourney. At the very least, something like Midjourney can create a near-flawless backdrop for a product placement. But it won't be long until the AI can do complete the whole job.
Any stock photograph is always a compromise because it's never precisely what the client had in mind. If you've ever purchased a stock image you know the drill - you click through page after page of thumbnails looking for the photo you've got in mind, you shortlist a bunch of them and eventually you settle for 'close enough'.
But now, someone creating a brochure, for instance, need no longer pay a royalty to a stock photo library by using a real photograph. They can get exactly what they want and they do not have to trawl through hundreds of images to do it. They do not have to compromise in any way and they get exactly what they want - near-instantly. Honestly to me it feels as futuristic as Captain Picard summoning up Tea Earl Grey Hot in the Enterprise's replicator.
If you're the sort of person that has to trawl through shots on Alamy, Getty or Shutterstock then you've every right to feel euphoric right about now. No more boring searches, no royalty payments and no copyright issues, ever.
And let's not forget that these generative art models are improving dramatically with every release. My first introduction to them was a video by MKBHD which, incredibly, was just under a year ago. Replaying that video now already feels like I'm watching some retro news clip from years ago.
The images Marques created in that video, of an astronaut on a horse and a wise elephant staring at the moon at night, seem ludicrously primitive compared to the kind images you can now produce in something like Midjourney. We've gone from toddler's crayon scribbles to art school graduates in under a year.
Non-specific stock photography is going the way of the dinosaur and any revenue you currently earn from stock images will probably fizzle out to nothing within a couple of years.
I'm sorry that photographers will lose a little bit of income but I'm also belly laughing at the stock agencies because they're a bunch of thieving soul-sucking scam artists who've been ripping photographers off for decades. So fuck those guys.
Then there's the specific kind of stock photography. Some celebrity exiting a nightclub, a politician standing behind a podium, an athlete winning a race. I'd say there was a strong argument that even these kinds of photographs are threatened by A.I. Models such as Stable Diffusion and Midjourney can already work from reference images and their ability to regenerate alternative images based on that reference is getting more powerful by the week.
Imagine you're a pop singer and GQ want to do a photo-shoot with you. At the moment you've got to interrupt your schedule, get to a studio and spend an entire day being photographed by an expensive photographer who may or may not nail the brief the magazine gave them.
In the extremely near future an alternative 'shoot' will take place. The magazine will suggest a scenario for a shoot, the pop singer will supply a couple of simple and up-to-date reference photographs and a prompt engineer will magic the whole thing up in a generative art system. The singer can simply and easily collaborate with the magazine to get the exact look they're all happy with and everyone's a winner. Everyone that is, except for the photographer and whoever owns the studio and the retoucher and the graphic designer.
But what about photographs of specific places? And to that I say, have you ever playing Microsoft Flight Simulator? Ok - that's a render engine based on a massive topographic and texture database, but I can't imagine it would be difficult to combine the two. Give Midjourney access to topographic data, add some real-world texture reference images to its dataset and set it loose. You'll be able to create an entirely convincing fake image of a real location that looks precisely the way you want.
Tourist agencies, hotels and property owners, real estate agents - they're all going to absolutely love it. Instead of searching online for images of a location that may or may not be popular and therefore may or may not have been heavily photographed, they can pick a location, an angle, a time of year, a time of day and specific weather conditions and the A.I. will produce four stunning shots in under 30 seconds.
And while I realise that many photographers are purists who are steadfast in their belief in the superiority of actual photos taken by actual people with actual cameras - nobody will care that the generative images in travel brochures, adverts or blog posts are not real. Nobody. If I was one of the overpaid leeches that works for a company like Getty, I'd be starting to get a big panicy round about now.
And while we're at it - how hilarious is it that stock agencies now have sections of A.I. generated imagery? Who is going to pay Getty for an A.I. image when they can type a prompt into a text box and get their own one in seconds flat. In the short-term I wonder if the real money is going to be made selling prompts - not photos.
Now I mentioned at the start that of this video that any market that relies on imagery of any kind is in deep-shit and that, of course, includes porn. A.I. created porn is already a reality, readily available and easily generated. I'm not talking Deepfakes here - those are just faces mapped onto someone else's body - I'm talking about 100% fake imagery.
The weapon of choice at the moment for generative porn producers is Stable Diffusion. The reason for this is that Midjourney currently blocks 'adult' imagery, but Stable Diffusion does not. They did initially, but they re-enabled the nudity option at the end of last year.
With A.I. generative porn you can create exactly the sort of images you want, the stuff that 'works' for you, without having to trawl through Pornhub's archives. And yes, that will of course mean that at some point in the future - possibly now - I don't know - you'll be able to render your own otherwise banned pornography. You'll be able to pick an age-range and summon up images that would land you in jail if you downloaded them.
Don't tell me that restrictions will be put in place to stop this kind of thing. The genie is well and truly out of the bottle and I have no doubt whatsoever that there will be a buoyant black market for generative art systems that can churn out porn catering to any and all niches. It probably exists right now.
What are the ethical arguments around people generating AI porn of banned subjects? It's preferable to the current real-world situation of course, but once Pandora's box is opened there's a whole bunch of other stuff to consider around giving people access to otherwise prohibited imagery.
Anyone that earns money from porn, whether it's soft glamour style or hardcore porn will probably see their income drop over the next few years. All the folks involved in the cottage industry that is Onlyfans will also undoubtedly see a downturn in profits too. Will people care that the images that turn them on aren't real when they can tune precisely into whatever kinks are theirs? Of course not. And we're not just talking still images here either. It's early days but AI generated adult video is already being produced and it will only improve in quality and sophistication as time goes by.
I have absolutely no doubt that we are on the beach, watching the water recede as a tsunami appears on the horizon.
Everything is about to change in ways that we could never have dreamed would come to pass in our lifetimes and you'd better be ready for it.
As Malcom X said, "The future belongs to those who prepare for it today."
