Managed to get the new Deep Floyd image generation model to work in a Colab notebook https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/deepfloyd_if_free_tier_google_colab.ipynb
It's supposed to be able to handle words... it's not quite there yet though. Here's what I got for: a photograph of raccoon in the woods holding a sign that says "I will eat your trash"
@simon I guess that makes sense when you put it like that
@lurker Yeah, words are known to be really hard - I guess because a slightly off face still looks like a face, but a slightly off letter looks like not-a-letter