photobyjohnbo

Lens-Artists Challenge #302 – Artificial Intelligence and Photography

President Reagan and Josh share a joke.

For this week’s challenge, we ask you to share your thoughts and images on a controversial subject – the use of Artificial Intelligence (AI) in your photographs. The amount of editing photographers use has been discussed for years, with controversy growing since the proliferation of AI.  From before the release of Photoshop in 1987, the extent to which photos could be manipulated has steadily improved – to the point where, today, an image can be completely computer-generated. This week, we’d like to explore that topic specifically.

In 2002, while working on a graphic arts degree, my son, Josh, created a photograph demonstrating image manipulation to emulate a newspaper photograph. He inserted himself into a newspaper photograph of President Ronald Reagan and Queen Elizabeth, using only Photoshop and two pictures. Today, applications like Photoshop, MidJourney, Firefly, and others can create photorealistic images using only text descriptions.

Reporters have been fired over minor modifications of newsworthy photographs made with only basic editing tools, never mind the use of AI. Generally, those who subscribe to Lens-Artists do not claim to be journalists and often edit an image to better conform to our artistic vision. The question for today is to what extent AI makes the work something other than photography and at what point should it be identified as such.

Pinnacles National Park
Pinnacles National Park Altered.

The image above of Pinnacles Park was captured on a cloudy day, but some peaks and valleys of the pinnacles were getting extra light. I found a more dynamic sky in my library and used Luminar Neo’s Sky Replacement tool. Does sky replacement constitute AI, or is it a simple edit? To what degree have I manipulated the image, and should I include a note to that effect in my text?

Generative Expand - Convert Vertical to Square Format
Converting vertical to square format.

In September 2023, we visited the Texas Ranger Museum in Waco, Texas. I captured a vertical image of this statue near the entrance. Have you ever realized you cropped too closely in-camera when you got home and reviewed your photos? Photoshop’s Generative Expand came to my rescue, but not without much editing work.

The original format is on the left. It took a couple of iterations to see a creation I liked. One tip for using Generative Expand, or Luminar Neo’s GenExpand AI tool, is to make the additional area beyond the crop larger than you would like. I wanted a square format, so I elongated the crop wider than necessary. That allowed me to determine how much of the newly added image to crop and make it square. If you look closely, you’ll notice that my final edit omitted some of the sky in the original (as well as the light posts.}.

Ancient ruins AI Final-12
Another World.

For this article, I created an image background using only AI.  I started with a royalty-free image of a woman and generated a background using Luminar Neo’s GenSwap tool. My prompt was “Ancient ruins on sand and desert background in haze backlit by a large moon.” I tried to get Neo to generate a round moon, but it always generated an oval moon. In one attempt, it generated two moons. That became the background I chose. I added a light source outside the right edge of the image and appropriate shadows to match the light on the model’s face.

Photoshop River-1-3
Desert scene altered by Generative fill.

The image above, which I haven’t published before, features an example of Photoshop’s Generative Fill tool. It is a desert scene with a road bisecting the extreme right of the picture. I masked the road on the right and carried the mask to the bottom left. In the prompt, I entered the text “raging river,” and Photoshop gave me three versions. This was my favorite option.

We recognize that not everyone is interested in AI editing tools. Either way, we invite you to share your images with or without AI and ask you to let us know your opinion on its use. What are your thoughts on what constitutes AI vs. enhanced editing, i.e., tools like sky replacement, removal of distractions, replacement of backgrounds, etc, vs. total creation of an image? At what point (if any) should we identify the use of AI? Except for experimental purposes, I don’t expect to create fully AI images for my blog, but if I do, you will see my admission to that fact in the caption or text.

I will close by asking you to comment on the following two images. One is a photo of a sunset in the Gulf of Mexico. The other is an AI-generated image created in Photoshop with the text prompt, “Sunset over the ocean with an island peninsula.” It took maybe six attempts to end up with similar images. I’d love to see your guess as to which is which. So as not to spoil the surprise, I will update the post with the solution on Thursday.

Sunset on the Gulf of Mexico
Image A.
Sunset from Tropical Island-10
Image B.

Edit: I promised a reveal and found it quite interesting that the majority picked the AI image as real. Image A is the original captured by my Samsung S23U off the coast of Naples, Florida. I did very little editing on this image, but it is not “straight out of the camera.”

In addition to the HD versions of the photos here, I’ve included some other images that use Artificial Intelligence on my Flickr site here. I look forward to reading your comments and posts regarding artificial intelligence and its application to those of us who consider ourselves lens artists.

Last week, we saw many signs of spring in posts featuring Sofia’s theme, Floral. Donna will host the challenge next week. If you’d like to participate in the challenge responses each week but aren’t sure how to get started, check here.

John Steiner

127 comments

  3. This is an interesting challenge for sure! I don’t have Photoshop (only Elements) and that’s maybe why I haven’t experimented much with AI, beyond occasionally dabbling with sky replacements in Luminar Neo. But your examples prompted me to try something more ‘out there’, and here’s the result: https://www.toonsarah-travels.blog/gallery-ai-versus-human/

    On your last pair of shots, my guess would be that A is the AI one although I’m not sure why I think so!

    Reply

    • I commented on your page, but it didn’t appear to post. Here’s what I wrote:
      Your own images are far superior to the AI-generated examples. Truly, that black-and-white could have easily persuaded me that it started with a real image and was edited for a mystical effect. Only the face appears to be “illustrated” rather than photography.

      Now, I think you ought to submit that AI bridge to the powers that be. I’d like to see that “for real.” >grin<

      All in all, it's been interesting to see people's attempts and read their thoughts on the new world of photographic AI.

      Reply

      • I think you must not have been logged into WP when you commented, as it went into my ‘pending’ folder as an anonymous comment. I’ve since approved it and responded, although I didn’t make the link to this comment at the time so couldn’t thank you by name!

      • I’m not sure how that process works, as usually we are allowed to login while making the comment. It’s been not working like that lately. I am logged into WordPress whenever I comment, but unless I do the Reader, it doesn’t always work as it used to. 🙂

  13. […] John from Journeys with Johnbo has come up with an interesting topic. Most of my exposure to the topic of AI (or machine learning) has been restricted reading a few articles related to health care and how machine learning can benefit the industry. I have also played a little with Chat GPT in helping me relearn Spanish. But I digress……Regarding AI and photography, it seems that the editing programs have been evolving in that direction and whether we chose to use these features little, a lot or not at all is a personal choice. […]

    Reply

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.