Unforgettable Car Prompt for AI Video (Copy & Paste)

The Unforgettable car prompt below describes the whole 15-second reveal for an image-to-video AI tool: a vertical handheld shot, red 2 and 4 cards at 3.3 seconds, a hand covering the lens at 4.5 seconds and a car already parked behind you at 5 seconds. Attach your photo, fill in the place and the car, and generate.

It follows the timing of the original clip, so the cards come up early and the car appears rather than drives in. Below: how to adapt it per tool, the beat sheet, variations and common failures.

The Unforgettable car prompt to copy

Vertical 9:16 phone video, one continuous handheld take, 15 seconds. The person from my photo stands in [place], framed chest-up, with an empty spot behind them. 0–3 s: they glance away, turn to the camera and talk. At 3.3 s one hand raises two red number cards, a 2 on the left and a 4 on the right, reading 24, digits not mirrored. At 4.5 s the hand sweeps across the lens and covers the whole frame for a moment. At 5 s the hand clears and a [colour, car] is parked behind them, completely still. They point back at it with a thumb, put both hands on their head and walk out of frame; the car stays alone for the last 2 seconds. Keep the person's face, hair and clothes exactly as in the photo. No text, logos or music.

How do you fill in the brackets?

• [place]: a short, concrete setting with room for a car. For example "a quiet parking lot at night under sodium lights", "a coastal road at sunset" or "a clean concrete garage".
• [colour, car]: colour first, then a body type or model. For example "grey coupe", "red convertible" or "black SUV". A specific model name can help, but simpler descriptions tend to keep the car's shape more stable.

Keep the rest as it is; the timings, "completely still" and "digits not mirrored" each prevent a common failure.

How do you use it in an image-to-video tool?

The prompt is written for tools that take a reference photo of a person plus a text prompt. Names and menus differ between tools, but the steps are much the same:

1. Attach your photo as the reference or start image. One person, face to the camera, shoulders in the frame, even light, no heavy filter.
2. Set the aspect ratio to 9:16 for TikTok, Reels and Shorts, or 16:9 to match the original.
3. Set the length to 15 seconds if the tool allows it.
4. Turn off any automatic music or sound. You will add the trend sound when posting.
5. Paste the prompt with the brackets filled in, and generate.

If your tool limits clip length

Many tools generate shorter clips than 15 seconds. The reveal is what matters, so keep the first 6 seconds intact: the cards at 3.3 s, the swipe at 4.5 s and the car at 5 s. Shorten or drop the ending ("hands on head, walk out of frame") instead of compressing the start.

If your tool takes a start frame and an end frame

Use a photo of you in the setting as the start frame, and an image of the same setting with the car parked behind you as the end frame. Then describe only the swipe and the reveal in the prompt. This tends to keep the car still and in the right place.

If your tool has a negative prompt field

Add things like "car driving in, moving car, mirrored numbers, extra fingers, text overlay, music". If there is no such field, the last line of the prompt already covers the main ones.

What beat sheet does the prompt follow?

These are the beats remakes follow, measured from @y0kun's original TikTok:

• 0–1.5 s: glance away, then turn to the camera.
• 1.5–3.3 s: talk to the camera, one hand near the face.
• 3.3–4.3 s: the 2 and 4 come up to the lens, reading 24.
• 4.3–4.8 s: the hand sweeps across and covers the lens (about 4.5 s).
• 4.8–5.5 s: the hand clears and the car is parked behind you.
• 5.5–11 s: point back at the car with a thumb and look back at it.
• 11–12.5 s: hands on the head, walk out of the frame.
• 12.5–15 s: the car alone in the frame.

If you shorten the prompt, keep the three timestamps and "completely still".

Variations: car type, time of day and location

Change only the bracketed parts and, where useful, one line about light:

• Car type: a sports coupe or supercar matches the original's feel; an SUV, pickup or classic car works the same way. For a car you actually own, describe its real colour and body type.
• Night: "a parking lot at night, lit by street lamps, the car's paint reflecting the lights". Start from a well-lit photo, since night scenes darken faces.
• Day: "an open car park in soft afternoon light". Daylight shows the car clearly, and every artefact.
• Location: a driveway, a quiet street, a garage, a coastal road, a pit lane. Keep it uncluttered, with an empty spot where the car will appear.
• Frame: for 16:9, change the first line to "Horizontal 16:9 phone video" and keep everything else.

What are the common failures, and how do you fix them?

• The cards come up late. Text prompts on their own often push the cards past the swipe. Keep "At 3.3 s" in the prompt, and if your tool ignores timestamps, generate a shorter clip that starts closer to the cards.
• The car drives in. Keep "completely still" and "is parked behind them", and add "car driving in" to a negative prompt if your tool has one.
• The numbers read 42 or are mirrored. Keep "a 2 on the left and a 4 on the right" and "digits not mirrored". Don't flip the video afterwards in an editing app.
• The hand doesn't cover the lens. Say "covers the whole frame for a moment". Without it, many models just wave a hand in front of the face.
• The face drifts. Use a front-facing photo with the face larger in the frame, no sunglasses, cap or filter.
• Strange hands or shadows. Start from a photo with hands down and a plain background.
• The car's colour or shape changes. Use a simpler car description, or a start and end frame as described above.
• Music or text appears. Keep "No text, logos or music" and switch off automatic sound.

How do you add the sound?

Add "Unforgettable" by French Montana featuring Swae Lee when you post. On TikTok, adding it from the library also puts the post on the sound's page. Use the trend's own sound rather than the full song, and line the swipe up at about 4.5 seconds so it meets the drop. Label the clip as AI-generated when you post; TikTok, Instagram and YouTube ask for it. Only use someone else's photo with their permission.

Why doesn't this site rely on the prompt alone?

Text prompts alone often bring the cards up late or let the car drive in. So the Unforgettable Car AI generator first makes two still frames with your face, then follows a motion map of the original clip. For background, see what the Unforgettable car trend is; to film it yourself, see how to do the Unforgettable car trend.

Skip the prompt and use one selfie

If you'd rather not test prompts, open the generator: upload one photo of you, pick a preset car or a photo of any car, and get a 15-second MP4 with the trend sound and no watermark. Your first 480p video is $5.99 with the Single Reveal pack; other packs are on the pricing page.