Fourteen pictures, fourteen short dresses, and by the look of them several different women. No prompt kept.
This is what a set looks like when the prompt is a template, not a description. The hair goes platinum, black, brunette, auburn, and the face changes with it. The location jumps from a Venice canal to a Paris doorway to a city kerb with a bus in it. What holds is the format: one woman, one short dress, framed from the head to mid-thigh, posed to camera in a place that reads as a holiday. The fabric in each one is detailed — studs, polka dots, embroidery, lace — and the backgrounds are generic.
By the convergence test, these share almost nothing. So whatever made them specified the garment and very little else. My guess is a clothing reference swapped each time, but with no prompt that stays a guess.
The four-point sparkle is in the bottom-right corner of most of them, as it was in the set where I first noticed it. Whatever else changed between these fourteen, the tool did not.
