I love this scientific curiosity towards DALL-E. Many people just say that it's bad at text generation (including me), but someone stopped to wonder if this is really gibberish or it has some logic to it. Classic "hmm, that's odd" case.
It will be fun to see people experimenting with extracting text prompts from generated images. I'd try something like "An open children book about animals" or "Random thought written on a paper". Maybe do a feedback loop of extracted prompts :)
Comments
I love this scientific curiosity towards DALL-E. Many people just say that it's bad at text generation (including me), but someone stopped to wonder if this is really gibberish or it has some logic to it. Classic "hmm, that's odd" case.
It will be fun to see people experimenting with extracting text prompts from generated images. I'd try something like "An open children book about animals" or "Random thought written on a paper". Maybe do a feedback loop of extracted prompts :)