Improving face generation quality and prompt following with synthetic captions

Open in new window