Turning multimodal image generation from "prompting" into "directing" by letting users draw, write, and mark intent directly onto the image itself.
More to come...