Based on this image, reverse engineer a Chinese prompt suitable for the SEEDREAM text-to-image model. Use coherent natural language, not a list of tags. Start by stating the type of photo, then sequentially describe: the model's age, hairstyle, makeup, clothing color/fit/material, the ongoing mid-action and eye expression, life-like details in the environment, the light source and time, the realistic texture of skin and clothes, the composition and shot size along with natural minor flaws, and the casual snapshot atmosphere. Aim for a realistic smartphone photo feel, retaining skin texture, clothing wrinkles, and environmental clutter. Avoid words like '8K', 'ultra-high-definition', 'extreme detail', 'perfect', 'cinematic'. Output only the prompt.
