Also, the focal point (red character) is meant to stay as the same spot as the words progress

I made a similar thing at wordflashreader.vercel.app, cool to see a different implementation. https://github.com/SpyC0der77/wordflash

Agreed. I have also built many LLM apps, and they aren't one line changes, but it is relatively simple to switch the model.

Fingers crossed this is right!

Can I have the link?

Hello there

A better way would be a VLA as opposed to a VLM. VLAs are meant to take action, where as vlms are for geneeral use. https://cognitivedrone.github.io/

Did some research, found a model that is exactly that. https://cognitivedrone.github.io/

The Black Mirror speedrun continues

Slaughterbots: https://www.youtube.com/watch?v=O-2tpwW0kmU

Thanks will check this out!

For fun

Buy an arduino kit and just try to solve some things around you