PROMPT RADAR / IMAGE平面设计

社交媒体帖子 - Dark Mode AI Model Leak Tweet

Generates a realistic X/Twitter screenshot announcing a leaked AI image model with two side-by-side embedded example images.

@ou_zhen599 0 热度 0 浏览
社交媒体帖子 - Dark Mode AI Model Leak TweetORIGINAL VISUAL / 来源见文末
编辑先看

它为什么值得进入灵感库

先别急着照抄整段 Prompt:真正有用的是主体、空间和镜头之间的约束关系。 先规定信息层级和文字安全区,再谈视觉风格;无法落版的漂亮画面不是完整设计。

完整视觉 Prompt

Goal: Create a realistic dark-mode social media post screenshot about a leaked AI image model, styled like an X/Twitter post. Canvas: Square 1024×1024 image, black background, with a single tweet occupying the full frame. Use authentic X/Twitter dark mode spacing, white text, blue hashtag/link color, and slightly rounded embedded media corners. Tweet header: Top left circular profile photo of a light-skinned middle-aged man in a white shirt. Display name is {argument name="display name" default="White"} in bold white text with a blue verified checkmark, and handle {argument name="username" default="@White1637402"} in gray beneath or beside it. Top right has two small UI icons: a stylized X/Twitter action icon and a three-dot menu. Tweet text: Use the exact visible post copy: “A new GPT image model just appeared on Arena called mona-lisa-1. It's extremely better than Image 2. 💪💪” followed by a blank line and the blue hashtag {argument name="hashtag" default="#ChatGPT"}. Keep line breaks and spacing like a real tweet. Embedded media layout: Below the tweet text, show exactly 2 side-by-side image cards in one row, separated by a thin black gutter, both with rounded outer corners. Left media card: A glossy 3D illustration of a cute white-and-black robot mascot in a classroom or chalkboard setting, waving with its right hand and holding a paper sign. The robot has a rounded black face screen with two glowing cyan smiling eyes, white helmet-like head shell, white body panels, black joints, and a friendly pose. Behind it is a chalkboard covered with handwritten educational text and doodles. Large chalk text reads “Hi there! I’m ChatGPT” on the left, and “That’s me!” with an arrow pointing toward an OpenAI swirl logo on the upper right. The robot’s paper sign reads “I was created by OpenAI” with smaller text underneath. Add realistic 3D lighting, slight depth of field, and toy-like materials. Right media card: A dramatic cinematic poster-like fantasy/business scene showing exactly 3 human figures: two giant suited men looming in the stormy cloudy sky above, and one small caped figure seen from behind in the foreground. The two giant men resemble tech executives, wearing dark business suits and looking down sternly; one has glasses and gray hair, the other has curly hair. The foreground figure wears a bright royal-blue cape with a white logo and the word “claeposeok” or similarly misspelled text on the back, standing on rocky ground and holding a sword at their side. The sky is dark, smoky, and epic with strong rim lighting, blue-gray clouds, and heroic scale. Visible text count: Include exactly 7 main visible text groups: 1 display name “White”, 2 handle “@White1637402”, 3 tweet body sentence, 4 hashtag “#ChatGPT”, 5 chalkboard phrase “Hi there! I’m ChatGPT”, 6 chalkboard phrase “That’s me!”, 7 paper sign “I was created by OpenAI”. Minor chalk scribbles may be present but should be small and not dominant. Visual style: Photorealistic screenshot compositing with crisp UI text, realistic embedded image thumbnails, high contrast, and no extra watermark. Keep the overall look like a viral tech leak tweet screenshot discussing {argument name="model name" default="mona-lisa-1"}.
查看原作者内容

不要一次改完所有变量

用 GPT Image 2 复刻时,先固定主体和构图,再逐次替换光线、色调或材质中的一个变量,这样更容易判断哪项描述真正起作用。

  1. 主体:Generates a realistic X/Twitter screenshot announcing a leaked AI image model with two side-by-side embedded example images.
  2. 视觉方向:社区实测摄影提示词;按原提示词的光线描述执行
  3. 镜头与构图:按原提示词的镜头与景别执行;按原提示词的构图与画幅执行
  4. 交付约束:auto,遵循原提示词中的避免项;不要水印、乱码、低清、畸形手、错误透视或主体漂移。
负向约束

遵循原提示词中的避免项;不要水印、乱码、低清、畸形手、错误透视或主体漂移。

模型建议

GPT Image 2 / Grok Imagine / Midjourney / Flux。先用原始 Prompt 建立基准结果,再根据模型对自然语言、镜头运动或文字排版的能力做局部改写。

来源与编辑原则

保留作者,也保留判断。

本页收录公开案例并提供中文索引、结构化拆解和复刻建议,不冒充原作者作品,也不替代原帖上下文。

作者:Loriel.AI · 收录于 2026年8月11日

访问原始来源

沿着这条线继续看