12 Copy-Ready Prompts for the Unforgettable Car Reveal
Unforgettable AI Car Video TeamPaste-ready prompts for the Unforgettable car reveal — four locations, four cars, and four fixes for the failures that ruin most renders.
A prompt for this trend is not a description of a car. It is a set of instructions about what must stay the same and what is allowed to change. Get that structure right and every prompt below works; get it wrong and no amount of adjectives will save the render.
The structure, in four lines
Every working prompt has these four parts:
- The action sequence, in order: cards up → sweep across the lens → clear → car revealed.
- The identity anchors: "same face, hair, build and outfit throughout".
- The composition: one camera move at most, plus the aspect ratio.
- The look: one lighting decision, one colour note.
The presets in the studio already contain parts 1 and 2. What you change is parts 3 and 4, plus the car itself.
This clip: 30 credits — about 6 a month on Pro for $39.90, or 80 a year for $319.
Four locations
1. Quiet car park, late afternoon
Two plain cards showing 2 and 4 are held to the lens, sweep across it, and reveal a parked coupe behind the same person. Continuous shot, 9:16, empty concrete car park in warm afternoon light, long shadows, same face and outfit throughout.
2. Neon street, night
The cards sweep across the lens on a rain-slick city street at night and clear to reveal a matte-black coupe under a streetlight. Neon signage reflecting in the wet asphalt, cool blue highlights, continuous shot, 9:16, same face and outfit throughout.
3. Golden-hour forecourt
The cards clear the lens and reveal a silver coupe parked on an empty forecourt at golden hour. Low sun glinting off the paint, warm haze, shallow depth of field, steady camera, 9:16, same face and outfit throughout.
4. Dark showroom
In a dark showroom, the cards drop away to reveal a glossy performance car on a turntable under overhead spots. Controlled reflections, premium commercial lighting, one slow push, 9:16, same face and outfit throughout.
Four cars
5. The one you own
…reveal a [colour] [body style] with [one distinctive detail] parked behind the person. Keep the body shape, paint, wheels and visible details consistent throughout.
Naming a real model by brand tends to produce a slightly-off replica, which reads worse than an unbranded car that looks right. Describe the shape instead.
6. The one you want
…reveal an unbranded low-slung sports coupe in deep green with a wide rear haunch, parked behind the person. Continuous shot, 9:16.
7. Two cars instead of one
…reveal two cars parked side by side behind the person — a silver coupe on the left and a dark SUV on the right. Both stay stationary, wheels grounded, continuous shot, 9:16.
8. A car that is not a car yet
…reveal a concept vehicle with a smooth unbroken surface and no visible doors, parked behind the person. Muted silver paint, studio light, continuous shot, 9:16.
Four prompt fixes
These are the repairs for the four failures you will actually hit.
9. The face drifts across the sweep.
Add: The person remains the same individual before and after the lens is covered — identical face, hair, skin tone, build and clothing. Do not change their identity when the cards pass.
10. The card digits come out mirrored or unreadable.
Add: Exactly two cards, showing 2 and 4, reading 24 from the viewer's left to right. Do not mirror, rotate or change the numbers.
11. The car melts or changes colour.
Add: The car is stationary and unchanged throughout: one body shape, one paint colour, all four wheels on the ground, no deformation.
12. The clip cuts instead of wiping.
Add: One continuous shot with no cuts. The transition is hidden by the cards covering the lens, not by an edit.
How to iterate without wasting credits
Change one line per attempt. If you change the location, the car and the camera move together and the render improves, you have learned nothing about which line did the work — and the next render will regress.
Two attempts is usually enough for a usable clip. If a third is still wrong, the photo is usually the problem: sunglasses, a group shot, or heavy filtering all make identity anchoring harder than any prompt can compensate for.
FAQ
Can I use these prompts on other models? The action sequence transfers; the identity-anchor phrasing is tuned for Seedance 2.0 and other models respond differently to it.
Should I put the credit-cost spec in the prompt? No. Resolution and duration are studio settings, not prompt text.
Do longer prompts work better? Not reliably. Past a certain length the model starts dropping clauses, and the two that get dropped first are the identity anchors — the ones you needed.
What if my photo has two people in it? Crop it to one before uploading. The model picks a face from a group shot, and it does not always pick the subject you meant.