There's a thread on the user forums with an example of doing that, with some images from various of these AI image generators. I think the most obvious flaw wit

There's a thread on the user forums with an example of doing that, with some images from various of these AI image generators. I think the most obvious flaw with the idea is that you can feed in a prompt like "a prawn and a horse are waiting in the railway station", and the image generator will happily produce an image missing at least one of the keywords, or weird things like a horse coloured like a prawn... So the images look cool but fail as cues to remember all the keywords that are supposed to be prompting you to remember the pieces of the character.