Today’s AI experiment – I decided to see how successfully Midjourney would recreate an image I had taken. One way to do that would be to upload my photo as an image prompt, but that’s not what I did – I wrote a text prompt describing the image and then ran a number of variations with slight tweaks to the settings. The results below don’t look very much like my photo, but they are pretty decent examples of the kind of locations and people I might have photographed on that day in Florence’s Uffizi gallery.

Here (below) is my original photo

And finally, the image below is what I got when I used both the text prompt and the original photo as an image prompt. ok, I lied at the top – I was curious to see what difference it would make to also give a visual reference, and it has clearly helped to compose the image a lot more like the one I shot.