It should guide the AI to generate an image closer to the original All versions prompts

no results