Every character here is fictional, created by the studio. Each one has their own trained AI model that has learned their face and look, so the same person can appear again and again, in any outfit and any place, and still be recognisably them.

They also have a personality and a backstory. Your story partner knows these, so the story you build will feel like something that could really happen to this person.

• One character keeps the story intimate: every picture is about them.
• Two or three lets the story cut between them, like a film moving between storylines. Each picture still shows one person (step 3 explains why).

A World is a complete look: an Alpine ski resort, an English village, a Cold War spy thriller. Everything in it, from the furniture to the light and the clothes, belongs to that one style.

A Setting is a specific place inside a World: a firelit lounge, a cottage kitchen, a hotel bar. The studio has built many of these as 3D rooms, so they look the same from one picture to the next.

A story happens in one Setting. That's what makes the pictures feel like one continuous scene rather than a scrapbook. Not sure? Choose "Surprise me" and the story partner will suggest somewhere that fits.

A story is a short run of beats, three to six moments, and each beat becomes a handful of photographs. A good beat is something you can see: "she reads the letter by the window and laughs" works, "she remembers her childhood" doesn't, unless we show it.

One person in each picture. The AI can keep one person looking the same from picture to picture, but a second person changes face and clothes every time. So other people are implied: the camera is their point of view, they're just out of shot, or we see evidence of them (a second coffee cup, a jacket on a chair, a hand at the edge of the frame). A story is allowed one picture with two people in it, for the moment that matters most.

Every great story builds to a peak moment: the one picture everything leads to. A look, a gesture, something we haven't seen this character do before. The story partner will name it on each card.

Each conversation has a limited number of back-and-forths, so make them count.

• Who's in frame. One character per beat. With more than one character, this is where you choose whose moment it is.
• What we see. Where they are, what they're doing, what they're wearing if it matters, and how they feel. Concrete beats make better pictures.
• How it's made. Described beats are composed by the AI from the words alone: perfect for mood, faces and simple poses. Staged beats get a 3D stand-in posed by hand first, for poses words get wrong: kneeling, lying down, curling up, leaning on furniture.

Three to six beats is the sweet spot. Open with a Described beat to set the scene, and save the Staged ones for when the pose really matters.