Long-time followers of my photography may recall that I like cityscapes. I am finding AI to be a perfect tool for creating some cool-looking cityscapes. (I want to acknowledge Tracy Hirai of Adobe for the suggestion leading to these images.)

The prompts for these images emphasize geometric abstract design and bold color blocking. These three images are produced by roughly similar prompts and progressively more powerful generative models. (Click each image for larger views with full-screen option.)

Changing to a horizontal view and adding more details, including asking for the inclusion of a city park, we get the following.

With a more powerful generative model and some refinements to the prompt, we get this.

I wanted to do the downtown cityscape with a main street included. Thus:

Relaxing the prompted requirements slightly (less geometry), this is the result. We get a more realistic image all around.

I find it interesting that several of these images include buildings unmistakably like the Empire State Building and, even, the Freedom Tower (that replaced the Twin Towers after 9/11). The prompts made no mention of New York City. This is a built-in bias of the generative models. (AI models are likely to have numerous built-in biases, some obvious, others much less so.)

That’s it for this week. I’ll try getting back to photographs next week. Watch for a new installment midweek, on Wednesday or Thursday.