
原始文档
GPT Image 2 · en
A glossy, soft pink anime banner in a wide, cinematic layout features a cute, romantic, and dessert-themed design. On the right is a confident anime girl, showing her upper body, sporting a fluffy short bob and hair color {argument name="hair color" default="Soft pink"} She has large, pinkish-purple eyes, a gentle smile, and crossed arms. She wears a Japanese school uniform: a brown blazer, a white shirt with a red bow at the collar, and a dark navy blue and purple plaid skirt. Two red ribbon hair accessories are added, a larger bow on the side, and a smaller ribbon decoration. On the left side, her name is presented in bold, glossy 3D cursive script.{argument name="character name" default="Natsuki"} The background should be a layered scrapbook collage in blush pink tones, with a bright pink outline, soft bevels, subtle shadows, shimmering effects, and a small heart decoration within the font. The background should be a layered scrapbook collage in blush pink tones with notebook paper textures, a subtle grid, and torn paper details, scattered with doodle hearts, petals, glitter, and cute baking elements. In the upper right and center-right corners, add four sketch-style portrait cards of the girl behind her, arranged like overlapping Polaroid photos. In the lower left and lower-left-center foreground, add two cupcakes, both adorned with pink frosting, striped wrapping paper, and miniature heart decorations or candy embellishments. Decorate the composition with flowing ribbons and bows: four main ribbon elements are visible, including one bow near the upper left, one near the lower left, and two long, rolled ribbons spanning the top and right edges. It adopts a soft and highly detailed anime illustration style, combined with exquisite lighting, dreamy halo, romantic Valentine's Day color scheme, delicate texture, and simple yet visually impactful thumbnail composition.使用方法
- 复制完整原始提示词,并保留其中的参数。
- 按需求替换主体、文字和构图要求。
- 在兼容的外部工具中使用,核对当前使用条件并检查结果。
开始前需了解
- 先确认是否需要参考图。标注兼容模型,不代表该示例由这个模型生成。
- 核对名称、数字与画面文字;不同模型版本的结果可能不同。





