What Are AI Photos? (AI-Generated Stock Image Potential in 2023)

It seems like AI images are everywhere, or at least they’re something everyone’s talking about. What are AI generated images, though and what counts as one? It’s not as straightforward as you might think. We’ll take a look at the types of AI image, their advantages and disadvantages, and how they might impact the world of stock photography.

What are AI Photos?

The word photograph is a bit misleading as there’s no camera involved in the process, and results could be artworks, rather than photographs. As a shorthand though, many people refer to AI images as photos.

The process of generating AI photos involves training algorithms on large datasets of images. These datasets could include things like landscapes, portraits, animals, or objects. Through this process, the algorithms learn to recognise patterns and relationships between different elements, and generate new visual content accordingly.

AI algorithms can also perform object removal or manipulation in photos. They can detect and remove unwanted objects or elements from an image, filling in the gaps with undetectable background content. Additionally, they can change or rearrange objects in an image, opening up flexibility and potential when it comes to composition, editing, or storytelling.

It’s important to note that AI photos are not meant to replace traditional photography or the skill and creativity of real photographers. Instead we should think of them as tools to assist us by enhancing their work, exploring new artistic possibilities, and streamlining certain aspects of the creative process.

What Does AI Generated Mean?

AI generated images are ones that are created entirely by artificial intelligence algorithms that mimic new content based on patterns and data that it’s trained on. AI images aren’t necessarily always a brand new photo or artwork though, they could be existing work, changed by AI in some way. Here some examples:

1. Enhanced image quality: One of the key advantages of AI photos is their ability to improve image quality. AI algorithms can reduce noise, adjust lighting, and enhance details, resulting in sharper, more vibrant, and visually appealing photographs. Adobe Camera Raw for example, has an AI algorithm tool called Super Resolution, which can greatly increase image size without reducing quality, by essentially rebuilding the photo.

2. Automated editing: AI-powered photo editing tools can automate some editing processes, such as colour correction, retouching, or object removal. This saves time and effort for photographers, meaning they can focus on the creative stuff!

3. Intelligent composition: AI can analyse the composition of a photo and suggest cropping, alignment, and framing adjustments.

4. Filters and effects: AI algorithms can replicate the styles of famous photographers or apply artistic effects to images. This enables photographers to experiment with different looks, though there are obvious ethical (and perhaps legal) considerations with asking AI to mimic someone’s work.

5. Image restoration: AI technology can restore old or damaged photos by removing scratches, dust, and other imperfections. This feature might be valuable for archivists and individuals looking to preserve memories, though the process of changing original sources should always be well documented so that any retouched picture isn’t mistaken for the original source.

What are Some ways to Generate AI Images?

There are lots of different ways that AI can be used to generate or alter images. Here are some of them, and an example generator that uses that method.

Text to Image

This is probably the most common method at the moment and involves you writing a text prompt into the image generator and the algorithm will create an image, or several images, based on that. You then often have the option to develop the generated image further, as a basis for new image selections, or to increase it in size for use.

While Dall-E and Midjourney often take the headlines for this kind of image generating AI, we cover those in another article, so let’s take a quick look at Stable Diffusion.

Stable Diffusion

My prompt was: a photo-realistic image of a person sitting at a desk, typing a prompt into an AI generator. The room is neat and tidy, with books on shelves, a plant on the windowsill, and nice evening light coming in through the window. There is a lamp on the desk casting a warm glow.

That produced the four images you can see above. I then have some further options to develop more images, or enlarge them. While they look fine from a distance, the quality of people and certain elements is still questionable – and they’re littered with anomalies.

The interesting thing about Stable Diffusion is that it has a Prompt Database that you can search to help you with seeing what has been generated by other people and how those results turned out.

Pricing. Free, or paid for access to 100+ APIs and models

Style Transfer

You upload an image and the generator creates a new artwork from that image. For example you could upload a photograph and have it turned into the style of a digital illustration.

Deep Art Effects

Deep Art Effects works on desktop or mobile and lets you create works of art or process your existing images in new ways through the Style Transfer method. You can scale photos up to 4x their original size, colour greyscale and black and white images, and choose from a number of Deep Art Effects filters.

Pricing

Deep Dream

This technique uses deep learning algorithms to make images that are surreal. It repeatedly modifies the image, focussing on specific elements and enhancing patterns and features.

Deep Dream Generator

Pricing

NB: ‘Energy’ is like credits, and ‘Recharging’ is how long those credits will take to replenish, so on the Advanced plan, every 12 hours you’d get 12 points (energy).

What are the Advantages of AI Photography?

Efficiency

It can be quick and easy to create stock photos using AI rather than by traditional photography. You can very quickly create a whole bank of images to use or sell just by writing some prompts. The flip side is it might actually take you a little while to start getting the results you want from your prompts.

Diverse or Unusual Content

If you can describe it, you can generate it, which means you can create stock photography in particular niches, or unusual scenarios. This is useful to target gaps in the market.

Customisation

You might have a stock image that’s great, but not quite right for a client’s needs. With AI, you can customise that photo to tailor it to the client and make something bespoke without needing a photo shoot.

What are the disadvantages of AI Photography?

Copyright, Legal, and Ethical Concerns

Let’s address one of the most important issues with AI photos right off the bat, the algorithms are trained on the work of other people and as much as that’s definitely an ethical consideration, it’s also a potential legal one if something is ‘copied’ from another artist.

There’s also the licensing issues of each generator – they can all be different, so it’s important to read the terms carefully to see if something you produce is really yours to use as you wish.

We might also see an increase of ‘fake news’ stock photographs and this is something else that will need close monitoring as the technology gets better, as these could mislead, misinform, and cause real world problems once social media takes hold. We tend to believe what we see, and we need to train ourselves out of that when it comes to AI.

GIGO

Early on in information technology settings we learn the phrase garbage in, garbage out. Inherently problematic issues can and are already, being replicated in AI offerings. Racism in AI is often perpetuated by drawing from already biased systems. AI photo generators offering to change existing photographs of people into professional head shots are lightening the skin of non-white people, making brown eyes blue. Features on women are being sexualised, with breast size increases, for example. This is a major issue with training algorithms on data from already problematic systems.

Quality

I’m sure it’ll get there eventually, but right now there are a lot of quality control issues across most AI image generators. We’ve all seen the jokes about ‘AI can’t do hands’ and there are numerous other problems, particularly when it comes to people. Even ‘perfect’ pictures have an unnatural glow about them that immediately identifies them as AI.

Overreliance

We’ve already started to see an over-reliance on AI technology. Recently, CNET were forced to take down a number of stories they’d published as it turned out they’d been AI written and were full of mistakes. When we contact a company we’re often forced to battle through an AI ‘bot’ before we can talk to a human, even when time is of the essence. As great at time and labour saving as AI tech is, it still needs to be balanced with doing the job properly, and if we can do that with AI, great, but it shouldn’t be a cheap replacement.

Envato Elements

You can find plenty of traditional stock photography at Envato Elements, and if you’re a photographer, you can look to contribute your own stock photos to sell by becoming an author.

Each month you’ll find a bunch of freebies too, including free stock photos, so have a look and see if there’s something right for you.

Conclusion – What Are AI Photos?

AI photos offer a range of advantages, like better image quality, efficiency, and realistic filters and effects. In this sense, they’re perfect for creating AI stock photography, but we should keep in mind some of the downfalls of this technology too: the inherent biases, licencing implications, and quality concerns. Legislation is always slow to catch up to technology.

Right now, if you’re looking to create stock photography using AI, you might want to compromise – the technology currently is perfect for improving on and customising your own photographs, rather than starting from scratch with a prompt. As the algorithms improve though, I think soon we’ll see more and more stock images created artificially.

Still Curious About AI? Try More of Our Top AI Articles

About This Page

This page was written by Marie Gardiner. Marie is a writer, author, and photographer. It was edited by Gonzalo Angulo. Gonzalo is an editor, writer and illustrator.