AI car stunt video from a complete car photo
See a car climb a ramp, turn in the air and land in a fictional stunt. Roadster uses your car photo as the appearance reference for a prepared portrait scene.
Existing sample: 12.04 seconds · 496 × 864 · Reference audio · silent ending · Seedance 2.0.
A fictional AI stunt made from an AI-created car photo. Reference audio fades near 10 seconds; the ending is silent. Wheels, smoke and framing can vary.
Sample dimensions describe this clip; the 480p generation setting is not an exact pixel-dimension guarantee. This example is separate from your selected photos.
100 credits per generation. Sign in with Google and confirm photo permission before submitting. See one-time credit packs.
Photo requirements
Choose photos that fit this template
Show one complete car with visible wheels. A clear three-quarter view helps you check the body shape before submitting.
A suitable starting point
- One complete car with the body and wheels visible.
- A clear three-quarter view.
- An image you have permission to use.
Choose another photo if…
- A dashboard, headlight close-up or cropped bumper.
- Several cars or heavy foreground obstructions.
- A photo too blurred to distinguish the vehicle.
These are framing tips, not an AI quality check. Selecting photos does not change the example or submit a generation.
JPG, PNG or WebP, up to 10 MB per photo. Preview and replace photos locally; upload begins only when you start generation. Framing advice cannot guarantee identity, motion or quality.
Before you generate
- Watch the full example, including its ending and sound.
- Open this template and add a separate photo for each role.
- Check the photo requirements, sign in and review the credit quote.
- Start only when ready. Buying credits does not generate a video.
Find submitted tasks in your history. Confirmed technical failures release reserved credits; uncertain results stay under review and must not be submitted again blindly. Delivered videos are available for 7 days; save them before the displayed deadline. Credit releases and refunds are different processes.
Roadster Ramp questions
What this template supports today.
Will the car always be the red car in an example?
Your uploaded car is the appearance source. The template is not restricted to the example’s paint color. AI can still change wheels, body details, smoke and framing; an exact vehicle match is not guaranteed.
What format and sound does Roadster use?
The current generation setting is fixed at 480p and 9:16. The existing example is 496 × 864 and 12.04 seconds. Its reference audio fades near 10 seconds and the ending is silent; full-length audio is not promised.
Does this show a real stunt or let me upload a driving clip?
It creates a fictional AI scene from a car photo. The public sample uses an AI-created car image. Custom driving-video uploads are not available, and this is not a driving demonstration or stunt instruction.
Genjutsu Lite is an independent Seedance template tool, not affiliated with Higgsfield. Arbitrary reference-video uploads are coming soon.