Is AI a Better Landscape Photographer Than Me

Perhaps you have heard recently about AI image and text generators and how good they are getting. With an image generator, you enter in a description of an image that you want to see and the AI produces a realistic image. There has been a lot of hubbub within the photography press about how this technology will eventually replace photographers with its superior abilities as people will not be able to distinguish machine-generated images from actual photographs.

DALL-E 2 Depiction of a “Robot photographer walking in the desert at sunset”

One famous AI image generator is DALL-E by an organization called OpenAI. Their description of the product is “DALL·E 2 can create original, realistic images and art from a text description. It can combine concepts, attributes, and styles.” Sounds cool! So, I signed up for an account to have a little fun with another blog I was reading and after seeing some results, it occurred to me to see if DALL-E 2 could best me at a photo of a sunrise in Big Bend. This should not be a particularly tough task for this AI as I don’t think I got great sunrise photos at Big Bend on my last outing.

First, my sunrise: This was taken on another clear morning last week from a desert overlook. I framed some desert plants in the foreground with the sun coming up over the mountains in the distance. This was the scene as it occurred Monday, January 23 2023 in Big Bend National Park. I adjusted the exposure and saturation in Lightroom, but there isn’t anything faked in it unless you consider the sensor dust spots that were removed from the sky portion of the image.

Image generated by natural intelligence using a fancy camera and Lightroom

Now the robot’s interpretations: I entered the description “Beautiful sunrise at Big Bend National Park” into DALL-E 2 and waited for about 10 seconds for my results. The tool then presented me with four images that it dreamed up of beautiful sunrises seen in Big Bend National Park. Below are my results from DALL-E 2:

I have to admit that at first glance, those are some nice looking images. There are even a couple without clouds in the sky to play fair against mine. There is a lot of color saturation in the images, and in mine as well to be honest. Compositionally, they are pleasing scenes. If you look close you will see that the plants are kind of a mess and don’t look very real. It did include a dusty horizon in some of the images which is a realistic touch. The images do look like an amalgamation of Big Bend landscapes and not actual places, though. I think if you put the DALL-E 2 images on Instagram (low resolution and small screen), I don’t think that most people wouldn’t notice that they are fakes. As soon as I looked at them full screen on my computer monitor, I began to see the imperfections though.

So, what do you think? Did my mere mortal attempts at capturing a sunrise a Big Bend get thoroughly outclassed by software sifting through a massive database of pixels and synthesizing a sunrise a Big Bend? Or does average human meatware still prevail in capturing and presenting the natural beauty of the world in a way that no robot could hope to comprehend?

Whatever your answer, I’m sure that the technology will improve and my days are numbered. I will eventually have to resort to going out hiking and shooting landscape photos merely for the personal enjoyment as everyone else will be dazzled and amazed by the intelligent computers. I could still enjoy writing blogs about hiking and photography, but it turns out that the AI can do that to:

“Hiking in the desert is a unique and rewarding experience that provides a sense of adventure and escape from the daily grind. The vast and open landscape, combined with the captivating beauty of the arid environment, creates a serene and peaceful environment for those who appreciate nature. For photographers, the desert provides an endless supply of opportunities to capture stunning images, from the intricate patterns in the sand dunes to the striking colors of the sunsets and sunrises. The combination of physical activity and artistic expression makes hiking and shooting photographs in the desert a joyous and fulfilling experience. Whether it’s a solo journey or a group outing, a day spent in the desert is sure to leave a lasting impression and create memories that will be treasured for a lifetime.” – ChatGPT

Thanks for reading my blog, for now.

Santa Elena Canyon in first morning light (image taken by a human)

51 thoughts on “Is AI a Better Landscape Photographer Than Me”

        1. My post was mostly just for fun based on the weekly freak-out post on the photography news sites about AI replacing photographers for whatever task. Society may be entering into a another period of techno-phobia like we had in the late 70s.

          Liked by 1 person

    1. I guess it’ll be gradually accepted into our lives without a ton of thought or resistance as the internet was, or microwave ovens, or whatever. I wonder if one day people will watch/listen to entertainment entirely composed and created by AI.

      Like

  3. The AI photos look more dramatic and spectacular to me, than does your photo. I like the first AI photo the best, as it has the most realistic looking foreground. The remaining photos have strange, unnatural looking foregrounds, which I noticed right away, and without having to study them. And the second photo looks more like a painting, to me. I like your photo better than any of the AI photos, because it contains a lot more fine detail, even though it isn’t especially spectacular.

    On the other hand, I approached this analysis knowing ahead of time which photo was natural, and which were AI. I’d like to see some natural photos mixed with AI photos, and have the opportunity to try to tell them apart from each other. I wonder how well I’d do.

    Liked by 3 people

    1. You risk angering the robotic overlords, so you better do well. Maybe I’ll try something like that if I get the right photos.

      I think that the AI generated photos are based on some large number of real photos that were tagged a certain way and it used all of that to generate photos. So, probably a lot of the photos that it had access to were the better photos as people wouldn’t publish their mediocre photos as often. I guess I could have contributed to the AI’s knowledge previously without knowing.

      Liked by 2 people

        1. My interpretation is that it is like an averaging of photos that it has seen. Those generated landscapes don’t look like any particular place I have seen in Big Bend but look like a landscape that you might expect to see in a place like Big Bend.
          I don’t know how you would trace-back all of the copyrights from that. It’d probably make some lawyers happy. Or maybe some AI lawyers.

          Liked by 1 person

    1. Yes, that explains it well, lacks the human touch. I agree that’s very important! Thats what makes it unique, for noone has the same touch as you do, whether its with photographs or writing, etc. And now back to writing my post. 🙂

      Liked by 1 person

  8. Ah, AI. Why are folk who are using it so lazy? That they need and want assistance to create art, an image or whatever that so many before us pursued and strived to create. It’s the personal ability to overcome imperfection (or because of imperfection) and enjoy the art created. Sometimes making stuff takes a little talent and ambition and work to make it happen.

    Now if only we could get AI politicians to make sensible laws and work for no pay but still get stuff done. Just saying.

    Like

  9. I’m going to confess to having watched and listened to probably twenty-hours of ChatGPT, about half as it interacted with a software engineer. At first, I was stunned. But after awhile, I started to recognize that the conversations, no matter how seemingly well-informed, didn’t actually understand what they were saying. A “Chinese room” effect becomes evident. AI may be aware in some sense, perhaps of “thingness” and interaction; but it isn’t aware of meanings. So while it can generate an algorithmic version of reality, including what it determines will be most appealing, it still makes sometimes glaring errors because it doesn’t understand the *why*.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

Gravatar
WordPress.com Logo

You are commenting using your WordPress.com account. ( Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. ( Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. ( Log Out /  Change )

Cancel

Connecting to %s

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