Hey IH!
I just launched LumiYing and wanted to share the story behind it.
## The Problem
If you've tried using AI to generate videos or images recently, you know the pain:
- Sora has a waitlist, then a subscription
- Midjourney lives in Discord with its own billing
- Veo is buried inside Google's ecosystem
- Seedance, Kling, and others all require separate accounts
As a creator, I found myself juggling 5+ platforms, each with its own UI, its own credits, and its own learning curve. Every time a hot new model dropped, I'd start from zero — new account, new billing, new workflow.
I thought: what if there was one place to access all of them?
## The Solution
LumiYing aggregates the top AI video and image generation models into a single platform with one unified credit system.
You sign up once, buy one credit pack, and use it across any model:
- Seedance 2.0 / 2.0 Fast — video generation
- Sora 2 / Sora 2 Pro — video generation
- Veo 3.1 — video generation
- Midjourney — image generation
- Seedream — image generation
- ...and more coming
You can generate from text, images, video references, or even audio. Pick the best model for the job, compare outputs, and only pay for what you use.
## How It Works
1. Sign up (Google or magic link — takes 5 seconds)
2. Get 20 free credits to try it out
3. Enter a prompt, pick a model, hit generate
4. Download your video/image, or browse the showcase gallery for inspiration
No API keys. No subscriptions. No lock-in.
## Pricing
Simple credit-based pricing, no recurring fees unless you want them:
| Plan | Price | Credits |
|---|---|---|
| Free | $0 | 20 credits |
| Starter | $19.90 | 1,000 credits |
| Pro | $39.90 | 2,250 credits |
| Max | $69.90 | 4,500 credits |
Credit costs are transparent — you see exactly how many credits each generation will use before you hit the button.
## Tech Stack (for the nerds)
Built as a solo dev with:
- Next.js 14 (App Router) on Cloudflare Pages + D1
- Drizzle ORM + SQLite for the database
- NextAuth 5 for auth (Google OAuth + magic link)
- Polar + Paddle for payments
- AWS S3 for asset storage
- Tailwind + shadcn/ui for the frontend
- A custom background job processor that handles async generation, polling, and callbacks
The whole thing runs on the edge — fast globally, cheap to operate.
## What I'd Love Feedback On
- Pricing — Does the credit model make sense? Too expensive? Too cheap?
- Model selection — Which AI models would you want to see added next?
- UX — Anything confusing or friction-heavy in the flow?
Would really appreciate any thoughts. Happy to answer questions about the build, the tech, or the business model.
Thanks for reading!
Try it free: https://seedance-v2.app