Some image prompts describe everything.
They tell the model who is standing where, what they are wearing, what the weather is like, what the mood should be, and what historical event is being reconstructed. That can work well. But sometimes the more powerful prompt is the one that hides its own subject.
A few numbers can be enough:
Coordinates 31.7785° N, 35.2296° E, Time: April 3, 33, 3:00 p.m.
Feed this into an image model with the right framing, and the result is not just a landscape. It becomes Golgotha. Jerusalem. The Crucifixion. A moment so visually overdetermined in art history that the model recognizes the coordinate-time combination as a historical trigger.
(more…)