woah real ophi bsky account verified confirmed

👀

www.youtube.com/watch?v=YHb7...

but what if im into biting

oh hey :3

Hello silly eepy

morningdm

Webfishing!

Personally, the biggest advantage you have is quality control of the actual jpeg quality when uploaded, the resolution, and the ability to watermark or crop certain elements can be a big factor. I wish I could offer more ideas, but unfortunately it's becoming harder to combat this stuff :\\\\

Partially, but a lot of the time they'll either hand-describe the actual photo themselves or use other AI tools like ChatGPT to fully describe it. The high wall for AI is that good data in results in good data out, but badly captioned data, or data that's not high resolution is basically bad

A majority of the stuff they scrape are re-captioned with other tools and danbooru/gelbooru tags to train their models, generally the alt-text has no use unless you're spending every letter to fully describe it as if someone is completely visually impaired, and even so its usually not enough

🥺 💙💙💙