Set the duration to 15 seconds and the aspect ratio to 9:16. Turn audio generation ON (some sites call it Create audio). Tip: if the site has a prompt enhancer, turn it off so it doesn't rewrite your prompt. Then paste the full prompt below.
Using your own characters? Edit the prompt before you paste it:
1. INPUTS: write your character names next to @image1, @image2 and @image3.
2. CHARACTERS: rewrite each block to match your sheet (skin, hair, outfit). Keep one Voice line for each. Delete the paragraph that starts "All three have the same surreal, uncanny look" unless your characters share that look.
3. THE SCENE: keep one time block per line. In each block, change who speaks, what they do and the line in quotes. Use ChatGPT's lines word for word.
4. DIALOGUE RULES: change who says which line.
5. SOUND and CAMERA AND LOOK: swap the diner details (rain, fridge hum, neon, mustaches) for your own.
Only 2 characters? Delete the third character block, upload your location as @image3 and change every @image4 to @image3.
SETTINGS:
Duration 15s · Aspect ratio 9:16 · Create audio ON · Highest quality
INPUTS:
@image1 = THE TALL HEAD (1-tallhead-sheet.jpg)
@image2 = THE GINGER (2-ginger-sheet.jpg)
@image3 = THE BOB (3-bob-sheet.jpg)
@image4 = LOCATION: retro diner at night (4-location-diner.jpg)
GLOBAL RULE:
Treat every reference as a fixed identity token.
Keep each character's face, hair, mustache and outfit identical in every frame, from the first frame to the last.
The location is @image4: same booth, same window, same table.
CHARACTERS:
THE TALL HEAD (@image1):
Deep dark brown skin, very tall cone-shaped black afro rising high above the head,
long narrow face, thin black handlebar mustache, small calm eyes,
olive-green houndstooth three-piece suit, white shirt, black-and-lime-green striped tie.
The tall hair never shrinks, flattens or changes shape.
Voice: very deep, very slow, completely calm.
THE GINGER (@image2):
Pale freckled skin, bright ginger helmet bowl-cut with blunt bangs,
big curled ginger handlebar mustache, long narrow face,
mustard-yellow tweed suit, white shirt, blue polka-dot bow tie.
Voice: thin and slightly nasal, quietly heartbroken.
THE BOB (@image3):
Pale freckled skin, glossy jet-black helmet bob with blunt bangs, thin drawn-on eyebrows,
long neck, burgundy plaid suit with a big white pointed collar, pearl earrings.
Voice: flat, quiet, monotone.
All three have the same surreal, uncanny look as their sheets: unnaturally tall elongated heads,
long narrow faces, completely deadpan expressions, yet fully photorealistic.
Nobody smiles. Nobody laughs. They play it like the most serious drama in the world.
THE SCENE (follow this order and timing exactly):
Seating: THE GINGER sits alone on the window side of the booth.
THE TALL HEAD and THE BOB sit side by side on the bench across from him.
One full plate of spaghetti sits in the middle of the table between them.
0–4s
Close-up: THE GINGER, perfectly still, staring at the plate of spaghetti.
Slow dramatic zoom-in on his face. He looks up into the lens and says, slowly:
"I'm leaving you both."
4–7s
Quick cut: THE TALL HEAD does not react at all. Dramatic snap zoom onto his face.
He looks down at the spaghetti and says, very slowly:
"Who gets the spaghetti?"
7–10s
Quick cut: THE BOB slowly pulls the plate toward herself with both hands.
Dramatic snap zoom. She says, flat:
"It needs a stable home."
10–15s
Wide shot of all three. THE GINGER slowly takes the plate back,
holds eye contact, and says, very slowly:
"Then none of us."
He slides the plate off the edge of the table. It falls out of frame with a loud crash.
Nobody blinks. All three stare straight into the lens in silence.
Hold the final frame.
DIALOGUE RULES:
Only these four lines, spoken in English, exactly as written.
THE GINGER says lines 1 and 4. THE TALL HEAD says line 2. THE BOB says line 3.
Each line is spoken ONLY by the character named, with perfect lip-sync.
Long dramatic pauses between lines. No other speech, no laughing, no narration.
SOUND:
Quiet diner ambience, rain on the window, a soft fridge hum,
and a cheesy dramatic soap-opera music sting on each snap zoom, and the plate crash at the end.
CAMERA AND LOOK:
Real iPhone footage, photorealistic, not CGI, not animation.
Telenovela-style dramatic snap zooms on each line.
Warm tungsten light with pink and teal neon from the window, natural skin texture with pores and freckles.
Do NOT add text, subtitles, logos or watermarks.
Do NOT swap faces between characters. Do NOT change hair, mustaches or outfits.
PRIORITY:
1. Each character matches their sheet exactly.
2. The right character says the right line, with lip-sync.
3. Deadpan acting and dramatic zooms.
4. The diner from @image4.
ChatGPT (writes the script)