This is a submission for the Cloudflare AI Challenge.
What I Built
A cool site where you can upload an image, select a language, let the AI do it's magic and receive a fairytale about that image, in your selected language, with 2 images!
Demo
My Code
💻 If you want to see my code, here it is: https://github.com/programORdie2/fairytaleAI
Journey
🚀 During this journey, I explored cloudflare workers and AI models. I only used pages before, so it was a whole new experience for me!
📚 Everything was very well documented, what made it very easy to use multiple models.
🧠 I really learned a lot of this challenge, and as soon as a text-to-speech model is available, I'm going to add this feature too. 🎙️
Multiple Models and/or Triple Task Types
I used triple task types (see the tabel) and multiple models (text generation), here is the list:
Type | Name |
---|---|
Image-to-text | @cf/unum/uform-gen2-qwen-500m |
text generation (SFE check) | @hf/nousresearch/hermes-2-pro-mistral-7b |
Text generation | @cf/meta/llama-2-7b-chat-int8 |
Translation | @cf/meta/m2m100-1.2b |
text generation (prompt generation) | @hf/google/gemma-7b-it |
Image generation | @cf/lykon/dreamshaper-8-lcm |
That's it, if you have any feedback or questions, please let me know in the comments!
Top comments (11)
A few months back I used this awesome post to generate a heroic version of myself:
And now I got a story to go along with it! 🙌
Haha! This is seriously fun. Thanks for sharing with us. 😀
Love this! Just generated my fairytale about a robot named Timekeeper, "who was tasked with keeping the time for all the creatures in the land." 🤖🕰️
Very nice! Good luck in the contest :)
Really good use of various models, quite creative. Love it!
Thank you!
Very creative 👍
Nicely Done. Best wishes
Amazing
do you find that the dreamshaper model is producing consistent images for your story, or does the character often look slightly different?
They are almost never the same
Oh wow, amazing work.