harrypotterme

Allow users to quickly access AI through a third party application which can transform any image into the world of wizards and make believe.

The API key has been hidden, but once replaced with a real key the app should work to produce a new image based on the previous one but with a wizard theme added.

This app previously used austica for image descriptions and has since been changed to only use openAi. There we plans to develop this further, but it has since been abandoned.

Below were the outstanding tasks and notes from previous development.

Next Steps

fed from task list

1. SPIKE: Get simple end-to-end example for "harry potter me" with openai only
   • use vision model: https://platform.openai.com/docs/guides/vision/vision
   • set up account?
2. "I'm sorry but i cant do that?" - how to get that indication back as refusal?
   • Edit and adjust description for now, with new page, accept, refuse, clear.
3. With spike finished then, design final approach
   • UML, whiteboard, testing first

Task list

consumed by next steps

Permanently:

1. Create a chatbot in openAi for this purpose
2. Show image on description page, non-editable until refused.
3. What if the description is crap?
   • allow users to cancel it, and retry.
4. Add chatgpt login form
5. small image first request, when happy, option for larger image?
6. Use openai's tokens/payments/pro access to image editing
7. Provide a link to open ai's sign up, if you dont already have an account
8. rename the project, avoid harry potter references entirely, instead, describe harry potter characters in detail
9. tip jar, provide users a way to pay the developer
10. performance?
11. UX?

Done pile

keep a track of what you did

Spike work

• remove the need for two different ai references
  • remove austica
  • replace with single chatgpt call