I can already see it. Many of you are clenching your fists and saying, “Never!” Certainly, there are many cases where using AI-generated images for photos is unacceptable. First, let's talk about some instances when it might be okay to use AI-generated images for photos. Next, look into such a great application.
Acceptable use of AI-generated images
I can consider three uses that I think it's okay to use some AI-generated images in your photos.
1. Canvas Wrap Printing
I wanted to make some photoshoot prints for my client who requested a photo printed as canvas wrap. She wanted it on a certain dimension. When I uploaded the image to a printing service online, I found that I was cropping some of the photos. I was able to use black to stretch the edges. This will make the sides of the frame black. However, the client did not want this. What shall we do?
I used Adobe Photoshop Generative Expand Feature to resolve this issue. This generated boundary was thought to be the authentic part of the photograph. However, it only appears on the side of the frame, not as a main photo. After that, the client decided she wanted them to frame too, so they became even more invisible!
2. Eliminate distractions from photography art
Sometimes we make art with photography. Of course, not all photos are photojournalistic or historical. If you have tree branches or street lights that distract you in your photos and you're creating art, I'm going to remove them.
Historically, I trimmed it. I also used clone tools, content recognition, or deletion tools. In most cases, this works. But sometimes, using something like a production fill leaves fewer artifacts behind. After all, I currently have a photo with a small amount of AI to analyze images and remove distractions. Is this unethical? Is it more unethical than using content? Is it more unethical than a portrait photographer who removes zit on someone's nose?
In an ideal world, the camera would have been positioned in a way that didn't include tree branches or streetlights in the photograph. But we all know that this is not always physically possible.
I think we can all agree that if we are making photojournalistic or historical essential photos, then we should not delete anything.
3. Viewpoint correction
The reason for creating AI-generated boundaries here is conceptually similar to the original reason. We are not trying to expand our borders to achieve anything and create a deceit notification.
Sometimes, when you're taking photos, you can't avoid distortions in the building's keystone (parallel lines appear to converge or branch out. Also, you can't back up physically enough to shoot big enough to fix keystones. After all, when you try to fix keystones with post processing, you lose too much of the edges.
A simple fix for this is to expand the edge. That's right – generative extension. You can use Content Aware Fill, but it rarely looks as good as a generative extension, and takes quite a while to achieve great results.
Expand the edge and then deal with the keystone. It's okay to lose the edge. In fact, what usually happens is that most or all of the edges generated by AI are supposed to disappear. Certainly there may be a few in the corner. However, in many cases it is much easier and looks better than the filling and other methods that content recognizes.
As mentioned before, the best solution is to shoot as broadly as possible. But we all know that in real life this is not always possible. Give yourself grace and use other technologies at your own whim (again, not for photojournalism or historical photography, but for photography art).
No, it's not okay yet!
If you think the above should not be done, then considering other attempts of “deep fake” or blatant deception, you're really going to brush your teeth. Here are some. In my opinion, it's never okay.
Deception
I think everyone would agree that they are trying to deceive using images generated by AI. This erodes trust. For example, some people no longer believe that my night photos are real. But, much worse, misinformation and deceptions erode public trust. The disinformation campaign was amazing. If people no longer believe what they see and read, where does it leave us as society?
theft
Photographers and other artists often argue that AI is imitating or stealing other people's work without permission, as it is trained in existing works. Adobe tries to avoid the plagiarism issue by learning models from Adobe stock images and public domain content without copyright restrictions. And some people argue that we all learn and integrate different art to some extent and that isn't too different to what AI does. But that discussion makes many of us feel uneasy at best. Being inspired by the art of another person may be completely different to ingesting it and refluxing it without any purpose or emotion.
Loss of revenue and employment security
There is a big debate about how much loss the images generated in AI cause. I don't think the microstock industry has benefited.
Energy consumption
Also, AI uses incredible electricity and water, so you can create successful cases without using AI. International Energy Agency (IEA) By 2030, electricity demand from data centers around the world is set to double from current consumption to 945 terawatt hours. This amount just exceeds Japan's total electricity consumption today.
This significantly reduced the use of ChATGPT, AI-generated images, and other similar tools. However, AI is now implemented everywhere, so even if you call a dentist or do a search online, it is quite difficult to avoid using it.
Where should the division line be?
A strong guideline is that if a reasonable viewer of a photo tells them how you created your photo, if you feel you felt you crossed the line, you will be deceived.
Apart from that, it's a matter of personal comfort. As mentioned above, it's comfortable to use images that are sometimes generated by AI in my photos. But I admire a lot of people not. I recently discovered how much energy AI is consuming, so I'm very concerned about how I use it.
