See that rectangle at the top? That’s my original photo. Well, almost. Check out the original, original photo below. The rest is generative fill. The original image was both extended and redrawn using Photoshop (Beta).
And, it's done in high-resolution. No puny 500 px x 500 px Ai images anymore.
To me, that’s exciting. And scary. And thought-provoking. Zoom in. Tell me I’m not imagining this, but I feel that the quality is astounding. Has Adobe (via Photoshop) just toasted the entire stock image industry?
I’ve been thinking about how easy it was to create this image. I simply two-clicked my way through the creation of the majority of this piece, only having to write in descriptive text prompts for the finer details, like "crow" and carefully selecting the area to generate the detail. The whole process took about four hours.
Is it art? Is it photoshopping? Is it something you can use to illustrate your products and services? I'd say yes to it all.
If you want to add finer detail, just select an area, select Generative Fill from the Edit menu, and type in a description. In this example, I added a flock of quail in diffused low light. It took three attempts (about 5 minutes) to get the birds on the dirt road looking decent.
So now, thanks to Generative Fill in Photoshop, it's possible to create entirely new scenes based on source photos, where the photos are the prompts. Text prompts, your time came and is going in a flash. Maybe this is the life preserver for stock companies...providing source images intentionally for generative extension and editing.
What about copyright protection/infringement?
What does this mean to creatives and businesses going forward? This was easy. Maybe too easy. I fully expect original photos and illustrations to be ‘borrowed’, extended, and presented as new work. In many circles, this would be considered creative license or fair use. In others, it might be seen as copyright infringement.
As a creative person and a business owner, I believe in copyright protection and am aware that substantial alterations may result in non-infringement rulings or fair use. Which test does an image re-created or extended with tools like Photoshop's Generative Fill meet? I’m not a lawyer or a judge, so I don't know the answer to that question.
Perhaps if a resulting image is an extension of the original (not substantially different) it may not pass the test. So, possibly infringement. What if it's seen as being only one component of a generative image? Extended vs incorporated.
This image I originally shot at Dry Island Provincial Park in Central Alberta has been extended dramatically, using only two typed words, "river" and "mountains". The rest of the image was created using the Generative Fill tool, making selections and clicking twice (many times over two hours). Please attribute me and link this page if sharing.
How will we know what's real, realistic, generative, composite, or unique?
For creators of original works, I ask…will behind-the-scenes footage and imagery (BTS) become the ultimate proof of origin? Should artists voluntary disclosure if a work is generative? Is that the right, ethical, and maybe legal thing to do?
And, will anyone really care?