Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I doubt it, since the Warhol work here has strong similarities to another one, while text-to-image models only directly reproduce existing art when something occurs as multiple (close) duplicates in the training data. Which can be quite easily avoided, like using a vector database to filter for similar items before training.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: