Cofi Trend Video Prompt for Two Photos

The Cofi Trend video prompt

Vertical 9:16, one continuous slightly handheld shot from a low angle at the edge of a kitchen island. A round red mug starts in the middle of the counter. Person A (photo 1) stands close to the camera on the right, holding a jar of instant coffee and a teaspoon; person B (photo 2) stands a step behind on the left. A acts confident and smug; B acts grumpy and unimpressed. Neither laughs.

0-1.5s: B slides the mug to the left and picks up a kettle; A scoops coffee powder. 1.5-4s: without looking, A tips the powder onto the middle where the mug was; B pours hot water into the mug on the left. 4-6s: B frowns at A and slides the mug back to the middle. 6-8.5s: A turns to look at B and tips the next spoonful onto the left spot where the mug had been. 8.5-10.5s: B turns away and leans on the counter; A breaks into a smug grin. 10.5-11.5s: A picks up the mug and holds it at his side. 11.5-13.5s: B pours a plain white milk jug straight onto the middle of the counter; the milk splashes over the powder and A’s grin fades. 13.5-15s: A turns to look at B, who shrugs with both palms up.

Realistic powder and liquid physics, natural hands, no text or labels, no music.

Why each part is there

The joke is not that someone misses the mug. It is that each of them keeps moving the mug away from where the other is aiming. Prompts that only say “spills the coffee” get a clumsy clip where the mug never moves; spelling out every mug move, second by second, is what makes the scene read.

The camera line fixes the classic angle, and naming who stands where keeps the two faces from swapping roles. The character lines matter as much as the actions: A is smug and ends up grinning, B is grumpy and finishes with a palms-up shrug. Without them most models make both people blank or laughing, and the scene loses its punchline.

Photo rules for two people

Use one photo per person, each showing a single face looking at the camera with even light. Group photos, sunglasses, masks and heavy filters are the most common reasons a face drifts or changes.

Only use photos of people who agreed to it. Do not use photos of minors, celebrities or public figures.

When the result goes wrong

If the mug never moves, give each move its own time range and say who moves it and in which direction. If the milk goes into the cup, make sure an earlier beat has A pick the mug up and hold it to the side. If the carton shows garbled text, ask for a plain white carton with no text. If the faces swap, put the role description before the timeline.

If you do not want to troubleshoot prompts at all, Cofi Trend AI has this scene built in: upload two photos, pay once and get a 15-second vertical video.