Two flows: drag and drop an image from a previous prompt's output into the prompt area, OR drag and drop an image from the canvas/media into the prompt area — so you can iterate on a variant without leaving the gen UI.
Captured during Telethon 2026.