Fflatlay studio

Method

Four images are not a photoshoot

Generate four frames separately and you get four different women in four different rooms. What makes a set a set is chaining.

4 min read

The drift

Ask for a front view, a back view, a detail and a three quarter view as four separate generations and you will get four plausible photographs of four different women, in four different rooms, at four times of day.

Each one is fine. Together they are not a listing. A shopper scrolling a product page reads inconsistency as an error before they read it as anything else.

Chaining

The first frame of a set is generated from your reference. Every frame after it is generated against the frame before, so the model, the room, the light and the time of day carry forward rather than being re-rolled.

That single structural change does most of the work. It is also why a set takes about thirty seconds rather than a few: the frames cannot be produced in parallel, because each one needs the last.

The face lock

Chaining alone still drifts across a long set, so the identity is named explicitly and carried as its own instruction. The same person, in every shot of this set, stated as a rule rather than implied by the reference.

For a saved model this extends across sets and across seasons, which is the difference between a lucky generation and a model you can actually cast twice.

Where it also matters

Colourways are chained to the original set, so eight colours of one dress are eight photographs of the same woman in the same room rather than eight separate shoots.

Reshoots are the interesting exception. A reshoot is deliberately not seeded with the frame it replaces, because the old photograph would outweigh the instruction to change it. It is anchored to the set instead, which keeps the model and the room while letting the change actually land.