哈哈哈哈哈哈这个有意思

@formulasearch·May 5, 2026·0

将任意输入图重绘为白底超极简抽象 meme 海报。只借用原图的构图骨架：主体位置、大小关系、前后层级、动作方向、视觉节奏和情绪对比。删除所有文字、UI、logo、水印、表情、五官、服装、纹理、背景细节和身份信息。人物、动物、物体全部转译为圆、椭圆、圆角块、胶囊形、软剪影和少量点状符号。保留“谁在何处、谁大谁小、谁与谁形成关系”的结构感，但不保留具体内容。低饱

Prompt

Try this Prompt

Logo / Icon Social Media Post Marketing Campaign Blog Illustration

How to Use This Prompt

Copy the full prompt text with the copy button above.
Replace any [bracketed] placeholders with your own subject, scene, or details.
Paste it into Gemini / Nano Banana Pro — or hit "Try this Prompt" above to generate here for free — then tweak the wording against the example images.

Style Notes

Logo / Icon: Symbol-first design — simple geometry that survives being scaled down, flat color, strong silhouette. Expect to vectorize the output before real use.
Social Media Post: Composition and aspect ratio tuned for feeds (Instagram, X, Xiaohongshu): bold focal subject, room for overlay text, colors that survive small screens.
Marketing Campaign: Ad-ready imagery: clear product or message hierarchy, space reserved for copy and logos, and a polished commercial finish.
Blog Illustration: Editorial spot art: a clear concept metaphor, friendly palette, and enough negative space to sit beside an article headline.

FAQ

How do I use this prompt with Gemini (Nano Banana Pro)?

Copy the full prompt with the copy button above, paste it into Gemini or any Nano Banana Pro-powered tool, and compare your result with the example images on this page. You can also click "Try this Prompt" to generate directly on this site — free daily generations, no sign-up needed.

Can I customize "哈哈哈哈哈哈这个有意思"?

Yes. Anything in [brackets] is a placeholder meant to be replaced with your own subject or details, and you can freely adjust subject, lighting, or style keywords — small wording changes often produce noticeably different results.

Which AI models does this prompt work with?

It was originally shared for Google's Gemini image models (including Nano Banana Pro). Most text-to-image models will understand it too, though composition and text rendering quality vary by model.