Over the past few years, AI image generation (yes deep fakes fall into this category) has been making some amazing progress. But over the past few months some sites have introduced new iterations and are getting much closer to fully AI generated stock photos complete with subjects interacting with their surroundings.

In 2019 Fast Company reported that AI was coming to kill stock photos, but at the time it was just AI generated faces. In itself, that was a big leap as :

• There was no model or photographer that needed royalties

• You could easily create a face that was unique to your needs or brand

• You could generate the gender, race and potentially emotional state of the subject

These tools are now widely available on sites like Generated.Photos which just introduced a real time face generator.

While this is a big step forward, it is missing a key element of stock photos, context. What we need next is to be able to place those subjects into backgrounds. In the past few months we’ve gotten a step closer with the introduction of two new AI tools.

NVIDIA’s GauGAN2 allows you to use text to image and other tools to create realistic stock photos without a subject (person). Once an image is generated there are lots of tools to manipulate the image to make it truly unique. You can select areas to add more sky, clouds, and trees. You can draw your composition using different colors to represent objects like water, forest, sky, and clouds.

DALL-E 2, which was just introduced, takes the text to image concept one step further by allowing you to designate a subject (Koala, teddy bear, and astronaut for example).

The ability to write a text description with a subject is a huge leap forward. And as you can see in the video description below, DALL-E’s AI is able to understand the relationships between the subject and the other objects so that a koala can actually drive a motorcycle and the astronaut can actually ride a horse in the correct position.

The implications of DALL-E 2 are huge in terms of blowing the stock photo market wide open with AI images. We are now only one step away from being able to specify a human subject and have them interact with their AI surroundings.

The other amazing thing about DALL-E is the ability to select a style for the image. So we are not just talking about upturning the the stock photo industry. AI is about to take a huge swipe at graphic illustration as well.

As we can see from the DALL-E 2 example, AI can understand the relationships between a subject and its surroundings. The next step is to be able to change the level of detail of the subject in the text description from “astronaut” to “Smiling African American female with a baseball cap”. Once AI reaches that plateau, the stock photo industry will forever be changed.

After that, it is just a matter of processing power to be able to generate the larger file sizes of traditional stock photos.