Comment by paulcn

No this isn’t implemented, it’s image-based only, meaning it transforms the image directly based on a prompt.