Roadster ai Guide
Subject images, stunt generation and result checks.
Make your video
- Open the homepage workspace and watch the full 14-second reference.
- Choose one sharp image of your car, vehicle or other subject, fully in frame. For people, keep the full body, face and outfit clear and confirm consent. JPG, PNG and WebP are supported, up to 20 MB.
- After sign-in, your selected photo is uploaded for generation. The template uses about 14 seconds, a 4:3 landscape frame and no audio. Choose Standard 480p or HD 720p.
- Confirm photo permission and review the selected quality: 50 credits for 480p or 100 credits for 720p, then generate.
- Continue in the result panel or My creations.
Check the result
Check identity in the closing shot, continuity through the climb, flip and landing, and the subject’s hands, feet, belongings and any vehicle. Download the result once reviewed.
Waiting and failures
A waiting timeout is not a confirmed task failure. Continue in My creations instead of submitting again. Credits are returned for confirmed failed tasks. Task records let you resume after a network interruption.
Scope
The fixed template replaces the subject and vehicle using the uploaded photo. With no vehicle in the photo, generation is instructed to remove the original vehicle and use movement suited to the subject. Custom reference videos and prompts are not supported. The reference demonstrates the target effect; complex motion and identity can vary. Use authorized photos and identify generated content appropriately.
For screenshots and a longer walkthrough, read How to make a Roadster ai stunt video.