I built Unforgettable Car AI, a small web app that turns a portrait into a car reveal video.
You may have seen the format: someone holds up a 2 and a 4 card, covers the camera with a hand, and reveals a car when the lens clears. The app follows that sequence, with your photo as the starting point.
The idea is easy to explain in a sentence. Making the creation flow equally easy to understand is the part I care about.
A small set of choices
The homepage has examples you can watch before signing in. There are short previews and full 15-second clips, so you can see the movement and the reveal before deciding whether to make one.
To create a video, sign in with Google, upload a clear portrait, and choose a car and a setting. A regular selfie works; you do not need to be holding cards or standing beside a vehicle.
There are three cars: a black SUV, a green supercar, and a red coupe. You can pair any of them with an underground garage, a rooftop parking area, or a city street. That gives you nine combinations without a long configuration form.
There is no prompt to write. The reference sequence guides the card gesture, the hand over the lens, and the reveal. When the job finishes, the result appears in My videos, where you can watch and download it.
What sits around the video model
The web app uses TanStack Start, React, and Cloudflare Workers. It connects to an existing video generation service.
The application work is in the flow around that service: collecting a portrait, keeping the car and scene choices separate, tracking the generation job, and making the finished file available to the person who requested it.
For this kind of app, those steps are a big part of the experience. Uploading a photo is quick. Generation takes longer. The interface needs to make it clear that the request is still being processed and where the result will appear.
The examples matter for a similar reason. A screenshot cannot show whether a hand movement or a transition feels right. Watching a full sample gives someone a more useful basis for deciding whether to try it.
The output and its limits
The result is a 15-second, 16:9 MP4, with a 720p output setting. It is silent: the Unforgettable song is not bundled with the download. Adding music or making a vertical edit is a separate step.
The model creates new footage, so facial details, hands, clothing, and timing can vary. I would watch the entire clip before sharing it, especially the moment when the hand crosses the face.
Viewing the public examples is free. Personal video generation is paid through one-time packs, with a single-video pack currently at $12.99 USD before applicable taxes. There is no subscription, and a generation confirmed as failed returns its credit.
Uploads and generated videos are private to your account. Photos are retained for up to seven days and videos for 30 days, so you should download anything you want to keep.
You can see the samples at unforgettablecar.org.
If you take a look, I would be interested in feedback on the creation flow. Do the examples make the output clear enough, and is there anything you would want to know before uploading a photo?
Top comments (0)