An AI-generated picture of historical ruins seems like a teddy bear when movement blur is added
Daniel Geng, Inbum Park, Andrew Owens
AI fashions designed to create pictures from textual content descriptions could make optical illusions that appear like totally different objects relying on how they’re seen.
Daniel Geng on the College of Michigan and his colleagues made slight adaptions to an current AI mannequin to get it to make a variety of illusions. Some seem to depict different issues as you view them from additional away; some grow to be three totally different pictures relying on their rotation and…