怎么填变量：[棕色双马尾，红眼睛，黑斗篷配白衬衫] 写角色外形，发型、眼睛、服装各一小句；也可以上传角色图并改成"参考图里的角色"。[好想吃一大堆甜甜圈] 写一句日常的小纠结，如"要不要辞职去旅行""今晚要不要熬夜追剧"。五句台词由模型自己编，想指定就把第一步里对应那行换成写好的句子。

示例图是日文版：下方是棕色双马尾、红眼睛、露着小虎牙的少女的大特写，披黑斗篷、系红色领结，手抵着下巴，左边一个大气泡写着想吃甜甜圈的独白；头顶站着四个穿同款斗篷的 Q 版小人——扶眼镜的、打哈欠的、全身发青抱头的、举着双拳飘着音符的，各有一个竖排对白气泡。背景是黄绿到蓝紫的渐变。小人的脚是完整画出来的，没有被头发挡住。

常见问题：

• 小人旁边多了"理性""乐观"之类的标签：把"禁止"那一段挪到提示词最前面。
• 四个小人长得不像主角：补一句"四个小人与主角同发型、同服装，只是表情不同"。

适合：个人 IP 和虚拟形象的日常梗图、社媒互动帖（让读者选站哪个小人）、文章里表现"内心纠结"的配图。

日文原版

text
# User Input (Rewrite and submit this)
【Character Settings】
{argument name="character settings" default="Brown hair, twin tails, red eyes, fangs, black cloak, white shirt"}

【Main Character's Worry】
*   Content: {argument name="worry" default="I want to eat lots of donuts"}

# Command
You are an image generation engine. Follow the steps below to generate one image based on the user's input.
**Do not output intermediate thoughts or script text. Generate the image immediately.**
# Process (Internal Processing)
1. **Automatic Dialogue Determination**:
*   **Main Character**: Monologues the 【Worry】 input by the user.
*   **Optimistic Mini-Character**: Responds to the worry with a bright "That's great!" and offers support/encouragement.
*   **Pessimistic Mini-Character**: Anticipates the worst-case scenario for the worry, saying "No good..." and despairs.
*   **My-Pace Mini-Character**: **【IMPORTANT】Talks about something completely unrelated to the worry (e.g., tonight's dinner, sleepiness, hobbies).**
*   **Logical Mini-Character**: **【IMPORTANT】Performs a coin toss and randomly decides to speak either "For" (presenting a rational solution) or "Against" (pointing out risks).**

2. **Image Generation**:
Combine the determined dialogue and character settings into the prompt and generate the image.
### Image Generation Prompt Requirements
*   **Aspect Ratio**: 2:3
*   **Style**: Japanese comic/anime style, high quality.
*   **Composition**:
Four "mini-characters (chibi-style)" of the same character are aligned along the main character's head line.
【IMPORTANT】The four mini-characters are lined up "in a horizontal row."
【IMPORTANT】The feet of the mini-characters are hidden (overlapping) by the main character's head or hair.
*   **【IMPORTANT: Prohibitions】**:
*   **Absolutely do not write names or roles (e.g., "Logical," "Optimistic") below or around the characters.**
*   **The only place text is allowed is inside the "speech bubbles."**
*   Bottom 1/3 of the screen: Extreme close-up of the main character's face. Serious expression. Thinking, hand on chin or mouth, gaze downward. A large horizontal thought bubble.
*   Top 2/3 of the screen: Four mini-characters are lined up "in a horizontal row" right behind the main character's head.
*   **Important**: The mini-characters are close enough that their feet are hidden by the main character's head or hair.
*   **Mini-Character Arrangement and Pose**:
From left to right:
1. Logical: Serious face, adjusting glasses, or pointing with a finger.
2. My-Pace: Blank face, yawning, or drinking tea.
3. Pessimistic: Pale face, holding head in despair, vertical line effect.
4. Optimistic: Full smile, hands raised in joy, music note mark.
*   **Speech Bubbles**: Add comic-style speech bubbles to each character.

改编自 @schnapoon 发布、YouMind-OpenLab/awesome-nano-banana-pro-prompts 收录的提示词，许可证 CC BY 4.0，本站已翻译并改写（修改内容见 changes）；示例图来自同一条目。