Creating Brand Stories Through Video With AI

Build the world once, then every clip is a moment inside it, and the product only ever appears in a tight shot.

Night outSix girls, one door, one compliment at the bar.

After partyThe steps outside, and the one who came to get her.

Getting readySame world, one girl, the jewelry going on.

How do I decide the world before I generate anything?

Write the world in one message. The place, who goes there, and the one rule: the brand is never said out loud.

Prompt 1 · the world

I am building a fictional world for [brand], a [what you sell] brand. The world is [a place: a private members' club, a summer house, a hotel, a tennis club], called [the name of the place]. It is chic, quiet, if you know you know. The product is never the focus. It shows up in tight shots on the people who live in this world, and nobody ever says the brand name. Here is the brand: [paste your website link] Here are the pieces I want to appear: [paste your product page links, one per line] Read all of it, then wait. Do not generate anything yet.

Returns a confirmation, not an image. Everything below reads from this message.

How do I cast the scene before spending on video?

Generate one still first. It is the cheapest thing you will make and it decides everything: the faces, the outfits, the place, the light. Approve it before a single second of video exists.

Prompt 2 · the hero still

Make one still image, vertical for Instagram, of the opening moment of this world: [describe the moment in one line, where we are and what is happening]. [Number] people, [who they are in one line: age, energy, how they carry themselves], caught mid-movement like a candid frame, not posed. Dress them in these exact outfits, one each, in this order: [paste one outfit link per person]. Jewelry or product, real scale, worn naturally: [name the pieces from my product links and who wears what]. The place carries its name in one small detail: [a plaque, a monogram, a matchbook, a towel] and nothing else is written anywhere. Shot on 35mm film, shallow depth of field, fine grain, [your light: warm evening lamps, hard midday sun, soft morning window]. Diverse cast, natural skin texture, real hair. No text, no captions, no logos other than that one detail.

Returns one still in about a minute, 2 credits. Ask for two and pick. When one detail is wrong, ask for that one change and nothing else: "Keep everything the same. Change only [the one thing]."

How do I write eight seconds so it does not feel like one long shot?

Eight seconds holds six moments, about 1.3 seconds each. You define them. Not the tool, not me. Watch how a film, a TV show or a fragrance commercial opens a scene: arrive, notice, touch, react, leave. Save the reference on Pinterest or Instagram, then hand it over.

Prompt 3 · six moments

Here are my references: [paste Pinterest or Instagram links, or attach screenshots, one per moment or one for the whole feel]. Write six moments for an 8-second clip in this world, about 1.3 seconds each, in a table, in this order: [name the beats you want, for example: arrive, meet, go in, cross paths, the compliment, the close-up]. Take the mood and the framing from my references, not the content. Rules: 1. One line of dialogue in the whole clip, and it is never the brand name. 2. The product appears in exactly one tight shot. 3. The last moment ends on a look, not on the product, so there is somewhere for the next clip to go. Show me the list first. I will approve it before you write the film prompt.

Returns the shot list. Costs nothing. Change the order here, not after the video exists.

How do I turn the still into the film?

The approved still drives the video, so the faces and the outfits hold across every cut. Fill the six moments in from your approved list.

Prompt 4 · the film

Use the approved still as the starting image. Make it 8 seconds, vertical for Instagram, with sound. Cinematic fashion film, six quick moments in eight seconds, hard cuts, 35mm anamorphic, shallow depth of field, fine film grain, slow deliberate camera inside each moment. Same [number] people, same faces, hair and outfits throughout. [Time of day], [place in three words], [the light in one line]. Moment 1 (0 to 1.3s): [your first moment, one or two sentences, name who is in it by their outfit and say where the camera is]. Moment 2 (1.3 to 2.6s): [your second moment]. Moment 3 (2.6 to 4s): [your third moment]. Moment 4 (4 to 5.3s): [your fourth moment]. Moment 5 (5.3 to 6.7s): [the moment with the one line of dialogue, in quotes]. Moment 6 (6.7 to 8s): tight insert on [the piece, described as it really is], catching the light, then tilt up as [who] lifts her eyes toward [someone or something off camera]. Sound: [three or four sounds that belong to the place], low music, one spoken line only. No on-screen text, no captions, no logos other than the small detail that names the place. Palette: [three or four colors]. Mood: [three words].

Returns the 8-second clip in about five minutes, 72 credits, or 36 at the smaller size. If a named preset is offered instead, decline it and run the prompt as written.

What breaks this build: putting the brand name in the dialogue. The moment a character says it, the world becomes an ad and every clip after it reads as one. The place carries the name in one detail. The product carries it in a tight shot. Nobody says it.

How do I make the next story with the same cast?

Pick one person from the approved still and put them somewhere else in the world. Pass the still back in as the reference and say who they were in it. Define six new moments the same way as before.

Prompt 5 · the second story

Use the approved still as the reference for the face. This is [which person, by their outfit and where they stood in the still]. Make it 8 seconds, vertical for Instagram, with sound, no dialogue. [Pick the style for this one: cinematic like the first clip, or an iPhone-style handheld montage]. [Where we are now, in one line, still inside the world]. Same face, same hair, [what they wear in this scene]. Six moments in eight seconds, about 1.3 seconds each, hard cuts. Moment 1 (0 to 1.3s): [your first moment]. Moment 2 (1.3 to 2.6s): [your second moment]. Moment 3 (2.6 to 4s): [your third moment]. Moment 4 (4 to 5.3s): [the moment the product goes on or comes into frame, close-up]. Moment 5 (5.3 to 6.7s): [your fifth moment]. Moment 6 (6.7 to 8s): [the exit, ending on a look back at the camera]. [Sound in one line], no dialogue, no lip sync, no on-screen text, no captions, no logos. [Light and color in one line], natural shadows, no airbrushed skin.

Returns the second clip, same cost as the first. Keep them dressed in every moment. A robe coming off or a dress going on gets the whole clip rejected before it renders, so any change of outfit happens between cuts, never on camera.

The three clips on this page are unpaid concept work, made as a sample for a jewelry brand I was pitching so the prompts read against something real rather than something invented. None of it was commissioned, I am not affiliated with the brand, and every prompt above is written to be filled in with your own product and your own world.