Show HN: Turn photos into Wordle puzzles with AI that runs 100% in your browser

[ud0]

I built Moments to get this game idea out of my head, finally. My original goal was to run on-device models specifically in mobile browsers, but running local vision models directly in phone browsers is still very much too early, so I focused on desktop.

How it works:

- You upload a photo.

- A local vision model running entirely in your browser captions it and picks a prominent object from the image.

- You guess the word just like Wordle.

It uses a very tiny model so it is not very smart https://huggingface.co/onnx-community/Florence-2-base-ft

[codingdave]

So I upload a map, and it gives me a 3-lertter wordle whose answer is: map.

This is probably a really clever coding exercise, but an enjoyable game it is not. Maybe share more about the code?

[ud0]

It uses a very tiny model, small enough to run in a browser, so it is not very smart https://huggingface.co/onnx-community/Florence-2-base-ft