Hey HN,
I had an idea a while ago to build an app that uses AI generate a kids picture books, but I didn't know how to train the models. After ChatGPT and StableDiffusionXL came out, it made it entirely possible for me. I don't know if im solving a problem here, it was more of a passion project.
Some parents currently allow their child to use a phone or tablet. I wanted to build something that could show the child, these are more than just consumption devices.
The idea is the child can imagine a story, and with a single prompt, bring a multipage picture book to life.
Tech Stack:
- Typescript
- React (vite)
- Node.js
- Firebase (Auth, Functions, Hosting, Firestore)
- S3/Cloudfront/Lambda (images)
- StableDiffusionXL
- ChatGPT (3.5 turbo)
Haven't implemented change username yet, so sign up and see what funny kids tale author alias you now go by!
Roadmap:
- Add change username button
- Find cofounders
- Improve story quality
- Expose page length, target age, art style variables through the UI
- Improve character continuity between images
- Git like clone/fork/merge features to remix tales
- ?
- Profit
Sometimes ChatGPT returns invalid JSON which causes the generation to fail, this is probably the biggest issue in the system right now.
We are currently limiting sign ups to 100 users, as we gather feedback and iterate.
Look forward to hearing your thoughts and answering questions.
