Why stock photography will die fast … if it's not evolving. #ai #aigeneratedart

This is an excerpt from my upcoming ai art related presentation

Status quo and how it's going to change … fast!

For years creatives flocked to stock portals in the hope of cheap fast imagery. This let to a constant and steep growth of sites like shutterstock & istock and many more. During this time we all thought that the advent of subscription based services like envato.elements would bring doom to the latter.

We were wrong all along. #AI is here to push up this timeline and stocksites could go under if they do not evolve. Here's the deal:

Currently if I'm looking for a photo (or illustration) I can enter some vague keywords and will find several photos which fit my need. Take the header photo from this article. It can cost me between 17 to 21 € depending on the credit package I buy. You can check out the link for yourself https://www.istockphoto.com/de/foto/frau-und-mann-blick-auf-den-malerischen-blick-auf-gordes-dorf-in-der-provence-gm1040906722-278650506

Now on to the interesting part

It was easy for me to find this photo since keywords and a description were added to it (highlighted in pink). These are one of the most important assets because they help us to navigate the fast libraries. But since 21 € is too much for me. Let's take this description straight to DALL·E and enter:

Young Caucasian woman and man looking at scenic view of Gordes village in Provence

DALL·E ist setting the right base but the light and camera angle are off. So let's fix this with a few additional prompts.

Added camera angle and mood

Much better but still not ideal. So let's optimize again by giving it more depth and make sure that the couple ist shot from the back.

Adjusting positioning of characters

Great! No let's add some greenery to each side of our scene to further approximate the reference photo.

Add greenery

Awesome! Let's create more images to finde the right one.

Add sea in the background and retry a few times

Done ? Now you just created your own photo. You can even change details with DALL·Es edit function or adjust their clothes.

Let's compare. While the original photo has a lot going for it: Great location and composition, wonderful lightning and mood, we now can create something more versatile within minutes.

Our generations only cost me about 70 cents and 3-4 minutes compared to about 20 € for the stock photo

I fear that stock sites could see a sharp decline in revenue as soon as #stablediffusion spreads in the creative industry and I hope they will be able to adapt. Otherwise #ai will generate their doom.

What could save stocksites?

In my opinion, the only thing they can do is to hop on the ai train and try to provide the same solutions in a better more optimized UX. Let's take the our photo as an example:

• Provide the ability to change characters, lightning, mood, location etc. straight on their website
• This functionality has to be as easy as possible to encourage users to stay and create their ideal image directly without switching to their own a local or hosted #stablediffusion #midjourney #dalle3 system
• Share the revenue fairly with the artist, since the artists image is the base and the ai changes are provided by the stock company.

What I fear will happen

• Argument against this form of innovation (https://mixed-news.com/en/why-a-stock-photo-startup-isnt-afraid-of-ai-images/)
• Just raise praises as sales decline

This wave is coming fast and it will hit with full force, especially as soon as the first developers start to simplify the #stablediffusion installation process for local systems. And I'm sure that large and slow companies are not prepared at all.

I encourage you all to try out my process as well. Pick a photo from a stocksite and try to recreate it with your favorite ai #dalle #midjourney #stablediffusion #nightcafe I would love to see your results.

BTW: I did not use the DALL·E creation since I did not need that photo. It was just a quickly done test. So no revenue was lost for istock and the photographer.