Reverse engineered prompt

Build me a polished AI video web app like the demo in this repo. I want people to sign in with Google, get a starting credit balance, and make videos from a text prompt, from one or two images, or by uploading up to three reference images for an edit style workflow. Let them choose aspect ratio, resolution, and model quality, then handle the generation in the background and update the result when it is done.

Please include a nice modern responsive interface that feels premium on mobile and desktop, a personal creations page with saved history, preview modals, and easy downloads. I also want simple pricing and credit purchase flow with Stripe, automatic credit deduction per generation, and webhook handling so billing and generation status stay in sync.

Set up the database, auth, billing, and environment config so I can run it locally and deploy it easily, ideally on Vercel. Use the existing project structure and make it feel production ready. Look up current docs online if you need to.

Want more depth? Deep Reverse