Two runs, shot from directly above a bed. No prompt was kept for this one, so there is nothing to paste.
These are the two closest runs in the whole collection. Same overhead camera, same crop from the top of the head to the ribs, same cream lace camisole with the scalloped edge, same dark hair fanned out across pale rumpled bedding, same warm underexposed light with heavy grain, same damp sheen on the collarbones. Everywhere else here, two runs of one prompt differ enough to fill a paragraph. These take a minute to tell apart.
What does differ: the second adds a fine chain with a round pendant and a pair of stud earrings, neither of which is in the first. The head is tipped a little further back and the mouth is open slightly wider. That is the whole list.
Which is the useful observation, even without the prompt. Two runs converging this hard almost always means the prompt was long and concrete — the more of the frame that is nailed down in words, the less room a second roll has to move. Compare it with the set where only the costume repeated: there the wardrobe was specified and the room was not, and the room changed every time. Convergence is a rough readout of how much of the picture the prompt actually described.
Both frames are small — around 612 pixels wide — so they are working files rather than finals.
