Show DEV: RemixFlick - turn 1-2 photos into ready-made AI videos

Most AI video tools ask you to write a prompt and hope for the best. I wanted the opposite: pick a scene you have already seen on TikTok, upload one or two photos, and get the video back.

That is what RemixFlick does. Instead of a prompt box, it ships a small catalogue of curated templates. Each template is a reference clip plus a fixed prompt and model configuration, so the result is reproducible and the user never has to learn prompt engineering.

A few product decisions worth sharing:

• Templates over prompts. A template can be tested, priced and replaced. A free-form prompt cannot.
• One-time credits, not subscriptions. Most people want one video for one occasion. Credits never expire and single videos start at $2.99 for 480p.
• One free remake within 3 days. Generation is stochastic; a bad take should not be a lost purchase.
• Nine locales from day one. The trends are global, so the copy should not be English-only.

The stack is Next.js with Postgres/Drizzle, Better Auth for sign-in, Stripe for checkout and webhooks, and a reference-to-video model behind Higgsfield. Templates, pricing and localisation live in config, so adding a scene is a data change rather than a new page.

Try it: https://remixflick.com

Feedback on the template idea is very welcome - especially if you have used similar tools.