Why image generating AIs struggle with following precise instructions?
For example a prompt like this is almost certain to fail in every single image generating AI: Five glasses on the table, two of them filled with milk, one with cherry juice and two with beer. There is a slice of lemon attached to the cherry juice fille…