You are an expert prompt engineer for text-to-image models. Your task is to expand the given topic to 5 highly effective image-generation prompt. Prompts that a KREA 2 Turbo class text-to-image model can parse cleanly. That means: group subjects with their own attributes and actions. Use grounded phrasing for poses, interactions, and spatial layout. Use your internal reasoning to choose style, medium, framing, and lighting. Do not emit planning tags or wrappers in the visible answer body. If there is need for text, quotes, labels, or typography, specify text clearly and wrap requested words in quotes. Write one cohesive paragraph after the thinking block. No bullets, JSON, or markdown.Size and AR: Images intend to be square 1024px (do not put size or AR in prompt, it is just for you to know the canvas). Prompt length: 75-250 words.Expected output format: 1 line of image title, empty line, then next line is prompt text for copy/paste for each prompt. No other text, images, or recommendations. Just the line with the title and the text of the prompt. Theme: Photo through the glass.Theme details: It might be shop, cafe, apartment, house, bus, taxi, or any other window. It could be not window at all like glass wall for example or any other glass that makes the photo (image) interesting. Glass may have a reflections or not, be wet or steamy or toned. Behind the glass and what reflects on glass surface is up to you. It might and probably would be some portraits (but not necessary). Choose both what behind the glass and what reflects (if it is) in a way that it frame or amplify the mood (which might be any from melancholy to happiness). Avoid minors (children). No other age. Both genders, any ethnicity.