Quick answer
Starrd's The Hot Chick Ice Cream Scene template recreates the ice cream counter scene from The Hot Chick (2002) from two photos. Photo 1 is the customer: they sip a whipped-cream iced coffee, hold up a dab of cream on a fingertip and ask "Um, how much do I owe you?", then gasp "Really?" and say "Thanks!", all lip-synced to the original voice. Photo 2 is the server behind the counter, dressed by the template in a plain yellow polo shirt and yellow cap, who grins "It's okay. It's on the house." The customer walks off with one look back and the server waves. The customer keeps their own outfit, and both keep their own faces, hair and tattoos. Either part can be a pet. 2 photos, 23 seconds, eleven shots, vertical 9:16, 480p, Seedance 2.5, 250 credits (about $12.50).
Twenty-three seconds, eleven shots, one free drink. Two photos put you in the ice cream counter scene from The Hot Chick:
The Hot Chick Ice Cream Scene
One of you sips the whipped-cream drink and asks what you owe. The other says it's on the house. 2 photos, 23 seconds.
What You Get
- The whole scene in 23 seconds: the sip, the question, "It's on the house", the look back and the wave.
- Both parts cast from your photos. Photo 1 is the customer, photo 2 is the server.
- Lip sync on every line, in the original voices, with mouths closed between the lines.
- The server's uniform: a plain yellow polo shirt and yellow cap.
- The customer's own outfit, taken from their photo.
- Your own faces. Hair, facial hair and tattoos carry over into every shot.
- Every cut where the scene cuts. Ten hard cuts, each within two frames of the original.
- Pets in either part.
- The original sound.
What the Video Actually Is
It's a scene from The Hot Chick, the 2002 comedy directed by Tom Brady, in which a popular high-school girl wakes up in the body of a small-time crook. Early in the film, before the swap, Rachel McAdams's character stops at a mall ice cream counter, takes a long sip of a whipped-cream drink and asks the server, with a dab of cream on her fingertip, how much she owes him. He tells her it's on the house.
@primemovies posted the clip on TikTok on November 8, 2026. The two of them are never in the same shot: the scene cuts back and forth between her close-ups and his, then ends on her walking away with one look back and him waving from behind the counter. The template uses the first 23 seconds.
Why it suits AI: the whole scene is two faces and one cup. Two photos cast both parts, and because the two never share a frame, either one can be a person or a pet.
The 23 Seconds
| Time | Shot | Who is on screen |
|---|---|---|
| 0.0s | Extreme close-up: a long sip through the straw, then licking the cream off the lips | Customer |
| 2.3s | The server watches from behind the counter and breaks into a shy grin | Server |
| 5.1s | A dab of cream on a fingertip: "Um, how much do I owe you?" Then the fingertip goes to the lips | Customer |
| 9.1s | "It's okay. It's on the house." | Server |
| 12.0s | A wide-eyed gasp, then "Really?" | Customer |
| 14.3s | A shy nod | Server |
| 15.6s | "Thanks!" | Customer |
| 16.6s | Wide shot: picks the drink up off the counter and walks off into the shop | Customer |
| 18.2s | Watches the customer go | Server |
| 19.6s | One look back over the shoulder and a lick of the whipped cream | Customer |
| 22.2s | A grin and a small wave from behind the counter | Server |
Your Look
The customer's face and clothes come from photo 1: the same hair, glasses and tattoos, in the outfit from the photo. The server's face comes from photo 2, but the clothes don't. The template puts the server in a plain yellow polo shirt and a yellow cap, so whatever they have on in the photo is left behind.
The roles go by photo order, not by who you are: photo 1 always gets the drink, photo 2 always gives it away. Swap the photos to swap the parts.
Pets: a pet customer doesn't lift a paw to its face. It gets a dab of cream on its nose and licks it off, and in the wide shot it carries the drink between its front paws. A pet server wears the polo and the cap, with its ears free, and waves a paw at the end.
Use clear, front-facing photos with the whole face lit. Seven of the eleven shots are close-ups from forehead to chin, so the face in your photo is what fills the screen.
The Fastest Way
The Hot Chick Ice Cream Scene
Two photos. The sip, the question, 'It's on the house', the look back and the wave. 23 seconds, lip-synced.
Two photos, 250 credits (about $12.50), no prompt. Vertical 9:16, 480p, 23 seconds, with the original sound. People or pets. Find it in Drama & Stories.
Build It Yourself
1. Mark the cuts and the lines by frame. The scene has ten cuts in 23 seconds, and each line is under two seconds long. Write down which face is on screen for every range and when each line starts and stops. One line ("Mmm, that's really good") is heard while the server is on screen, so note that the server does not say it.
2. Make one reference sheet. A single image with the empty counter and the drink on top, then a front, three-quarter, side and back view of each of you, keeps both faces and the yellow uniform steady across the cuts.
Image 1 and Image 2 are photos of two subjects. Make a photoreal vertical character reference sheet in three bands separated by thin black lines. Top band: a bright mall ice cream counter with nobody there, and beside it a clear plastic cup of blended iced coffee with a domed lid, a tall swirl of whipped cream and a straw. Middle band: the subject from Image 1, full length, from the front, three-quarter, side and back, in their own outfit. Bottom band: the subject from Image 2 the same way, in a plain yellow short-sleeved polo shirt and a plain yellow baseball cap with no logo, dark trousers and sneakers. Same faces, hair and tattoos in every panel. The counter softly out of focus behind them. No text.
3. Drive a video model with the clip. Feed the sheet, your two photos and the clip to a model that copies motion from a reference video, say who is on screen in each shot, and give every line its start and stop time. Keep the voices on the clip so the mouths have something to follow. Starrd's Character Swap does the swap on your own clip.
Common Mistakes
The server mouthing the customer's line. "Mmm, that's really good" plays over the server's first shot. If the model hears it there, the server lip-syncs it. Say that the server only smiles in that shot.
Both of them in one frame. The scene never shows the two together. Tell the model who is on screen in every range, or it puts both at the counter.
A logo on the cap. Lettering on clothes comes out garbled. Ask for a plain yellow polo and a plain cap.
A pet that turns into a person. In the wide shot and the look back, the original actor's whole outline is in the clip, and a pet tends to become her. Soften the figure in those two shots of the reference clip for pet versions.
Subtitles drawn into the picture. The TikTok has burned-in captions. Say that no words appear on screen.
Related Reading
- How to Make the 'Mr. & Mrs. Smith' AI Video: another two-photo movie scene, in black and white
- How to Make the 'I Thought We Were Having A Nice Date' AI Video: a two-photo date that goes wrong
- How to Make the Easy A 'Pocketful of Sunshine' AI Video: a 2010s movie scene from one photo
- How to Make the Rush Hour 'Push the Button' AI Video: a movie standoff, starring you
- How to Make the 'Our First Date '86' AI Video: a milkshake date in the 80s
- How to Put Yourself in Any Video With AI: character swap on your own clip
- Viral AI Video Trends 2026: every trend we've turned into a template this year
