It’s easy to think that AI can generate any type of picture, but these tools have their limits. To put this to the test, I took a series of smartphone photos to prove that AI can only go so far—some with a focus on aesthetics, and others on practicality.
1Burst Photo Series
On my morning walk one day, I captured a burst photo series on my iPhone of people cycling along a bridge. I wanted these to look a bit like a magazine or comic strip and show a specific sequence of events. As you can see, I managed to capture every couple of seconds between the bikes crossing along the bridge.
While AI is good at generating solo pictures, it’s not really capable of creating image series. For example, ChatGPT will often change the picture completely when I ask it to make simple adjustments. Moreover, using the same prompt will often yield completely different results.
2Photographing Signs
I’ve spoken in depth about generative AI being bad at creating text in images. While I sometimes get decent results when using software like ChatGPT, these tools are often way off the mark. I’ve had numerous experiences in which the text was illegible, so I won’t rely on generative AI for these kinds of pictures anytime soon.
Trying to create text in AI images has made me realize how I often take the simplicity and usefulness of smartphone cameras for granted. Whenever I want to take a picture of a sign, I simply have to point my camera at it and capture the shot.
However, I also wanted to make the image a bit more interesting. So, I made the sign part of the foreground. Generative AI tools can be good at creating bokeh and background effects, but since they fall short with text generation, these kinds of pictures are better when snapped on a phone.
3Leading Lines
Generative AI is pretty good at creating landscape-style pictures, but I’ve sometimes found it to be lacking when I try and create pictures with more complex leading lines. It’s fine for straight roads, but beyond that, the AI tends to struggle. On the other hand, it’s much easier to achieve leading lines with your smartphone.
In this example, I took a picture on a beach while standing on the walkway. As you can see, I simultaneously captured a lens flare shot with my smartphone. I liked how the light shone on the walkways, and it’s quite cool how the boardwalk starts bending to the left as it reaches the background.
I could potentially create a similar image with generative AI, but it would fall short with the core leading line element. You may, however, want to try some key AI prompting tips and tricks to improve your chances of getting the result you want.
4Nighttime Smartphone Photography
Generative AI can make some interesting night shots, but it tends to falls short in getting the colors right. In my opinion, generative AI is better for daytime pictures. In this nighttime smartphone photo, I wanted to specifically capture a building with an interesting shape. This kind of architecture is typically harder for a generative AI tool to replicate.
The light reflections in this picture are also unique. While AI can create reflections in the water, it can make them look more like watercolor paintings than the real reflections you’d see in photos. To take even better nighttime pictures on your smartphone, consider trying these night photography apps.
5Silhouettes
Taking photos of silhouettes can be very difficult, and to be honest, it often happens to me by luck rather than actually trying. Silhouette photography is not just about making something look darker, you also need the light to hit your image just right. For example, in the picture below, the light hit the modern building on the left at just the right angle.
The different shades in the sky of silhouette photos are also difficult for AI to emulate, even with specific prompts. Moreover, I feel like these tools often overstate the shadows; if I want my pictures to look like that, I can just use my phone’s photo editing app.
6Macro Photography
AI can be good at generating close-up shots, but with some limitations. While these tools are great at generating pictures of people, I find that they struggle with images of plants. To put this to the test, I took various macro pictures of flowers and leaves.
My smartphone camera was able to capture these plants’ various textures, which was precisely what I wanted. It was also good at capturing the real colors without over-saturating them; generative AI tools tend to do this with bright colors. Nonetheless, as I found out when trying macro photography with my smartphone, this style requires a lot of practice.
7Crouching Down
I always try to find new angles with my smartphone photography, and crouching down is one way to do this. I can ask an AI platform to create an image that looks like I’ve done this, but it hardly gets it right. Since I like having more control over composition, doing it with my smartphone is just easier.
The signs in this picture were my main focus area, and I first went beneath a bridge before crouching down further. As a result, I also managed to get enough of the sky in my photo. Asking an AI tool to do this would probably not give me what I wanted.
8Frames Within Frames
AI is good at generating images that only have one or two layers, but things become trickier when I try adding more elements to the picture. For example, I like capturing frames within frames; ChatGPT holds its own in this area, but the results are far from perfect. To try and do something that AI couldn’t, I took the picture below.
I wanted to include the car in the distance, which is something an AI would have struggled with. Since it often struggles with different patterns, I don’t think it would’ve succeeded in creating the lines on either side of the bridge either—AI tends to be better at curves.
AI does have its merits, but it isn’t reliable for every type of image. Smartphone cameras can often outperform AI, and while this might change in the future, I still find smartphone photos superior.
