如何 ComfyUI /MJ 给博客配图 cover image

如何 ComfyUI /MJ 给博客配图

场景 / 角色 描述 Prompt
🧑‍🚀 主控台宇航员 写实风格,主控台内,手捧金色地球,深蓝配金色,科技感 a realistic astronaut standing inside a futuristic space control station, holding a glowing golden planet in his hands, surrounded by holographic interfaces and star charts, cosmic blue color scheme, detailed, cinematic lighting, ultra wide shot
🐝 幻彩小蜜蜂邮差 拟人化卡通风格,穿邮差帽,有翅膀,拿信封,可爱扁平 a cute flat-style cartoon of a humanoid bee girl postwoman, wearing a small postal cap, holding envelopes, with colorful wings and a mailbag, cheerful expression, soft pastel color palette, minimal background, children's book illustration style
☁️ 云端邮局 浮空建筑,周围漂浮信封,梦幻云雾,柔光风格,像宫崎骏 a whimsical floating post office in the clouds, surrounded by floating envelopes and glowing message lines, pastel color nebula sky, dreamy atmosphere, warm soft light, Ghibli-inspired architecture, mix of steampunk and fantasy, medium distance shot
🌌 银河总览图 宇航员远眺,小蜜蜂在宇宙中飞行,主控台与邮局同框,故事氛围 a scenic panoramic view of a futuristic galaxy station with a floating post office nearby, an astronaut stands at the control center, watching a small bee-like postwoman fly with glowing envelopes, deep space background, colorful planets, soft glow, storytelling vibe, anime-style digital art --v 6 --ar 21:9 --style scenic

视觉素材收集

favicon

社交卡片

首页 banner 动图用图的 prompt,

逐步建立一个视觉统一的“幻彩小蜜蜂宇宙”。🪐🐝


封面图明确👌喜欢的风格之一,扁平的好处是,对图片的压缩会比较好,虽然看起来有点笨拙,但也不失专业性:

  • 人物偏 Q 萌 / 简笔画风格(可爱不复杂)
  • 画面整体简洁、留白多
  • 背景干净,线条轻松,风格更轻巧现代一些

A simple and cute illustration of a girl using Tailscale to remotely control her home computer. The girl is drawn in a chibi or minimal line-art style, sitting at a laptop. A dotted line or whimsical arrow connects her to a second computer far away, showing the remote connection. The two devices are connected through a floating tail-shaped icon (Tailscale). The background is clean and minimal, with soft colors or white space. Everything looks light, modern, and friendly. The style is flat, with Q-style character, simple lines, and no heavy shading. No clutter, no extra background details. Calm, tech-inspired but playful.

内容 描述
🎨 风格 Q版人物、简笔插画、极简背景
👧 主体 一个女孩正在远程连接家里的电脑
🧠 元素 虚线、箭头、小图标代表连接
💡 整体感觉 轻松、有趣、现代、好理解

  • “女孩坐在阳台远程连接家中台式机”
  • “手上有手机,远程控制远方电脑开机”
  • “两个设备之间飘着 Tailscale 的小尾巴”

这些我都可以考虑加进去,用 ComfyUI 的节点控制也比较明确。


还是这个风格,主题改成:从上海虹桥出发到南京的两日游行程规划

继续保持喜欢的「Q 版简笔画 + 干净留白 + 温柔又轻松」的风格,把主题改为:

🧳 从上海虹桥出发 → 南京,两日游行程规划

A simple and cute travel-themed illustration showing a girl preparing for a two-day trip from Shanghai Hongqiao to Nanjing. The girl is drawn in a minimal, chibi-style with a suitcase. There is a train (suggesting high-speed rail) connecting the two cities, marked with playful dotted lines or arrows. Small icons around her show trip planning elements: food, hotel, map, camera, and famous landmarks like Nanjing’s city wall or Confucius Temple. The background is clean and bright, with light pastel colors and soft lines. The style is flat, whimsical, and modern, with no clutter. A friendly, relaxed mood perfect for a casual travel itinerary. Anime or doodle style, simple line art, only one character in focus.

  • 👧 主角:Q版女孩,带着包或行李箱;
  • 🚄 场景:虹桥站出发 → 南京(可以用虚线或小火车连接两个地名);
  • 📍 元素:南京的地标图标(城墙、夫子庙、鸭血粉丝汤、秦淮河);
  • 🧭 工具类:地图、攻略本、手机App、小闹钟等简图标;
  • ☁️ 背景干净留白,主视觉集中在“人物+路线+少量icon”。

flat illustration, travel planning, pastel colors, minimal background, chibi girl, suitcase, train line, tourist icons, cozy mood, day trip, whimsical map style

如果打算分成多个节点(比如第1天 / 第2天路线图),我也可以帮拆成多个 prompt 场景。例如:

  1. 出发前(站在虹桥站前,手机看地图)
  2. 南京景点游览(城墙边拍照 or 晚上坐秦淮河游船)
  3. 旅馆休息(在 Q 版小房间泡脚看电视)
  4. 第二天吃美食(吃鸭血粉丝汤、盐水鸭)
  5. 准备返程(背影牵着行李箱走进站)

很多 AI 插画现在的共性问题

构图可以很聪明、氛围也对,但一到细节的小图标、小物件,就容易画崩、变形或者辨识度低。

  1. AI 在处理“低对比细节+小尺寸元素”时识别力弱,像图标、路牌、手机屏幕这类内容;
  2. *没有设定好“图标需简洁、清晰、分明”**的关键词;
  3. 图标本身在 prompt 中是“配角”,AI 可能随意处理;
  4. 如果是图示类插画风格(地图、计划页),需要特别强调“icon-like clarity”。

把那一段加上这类关键词来明确风格需求:

with small, clearly drawn icons around her, flat icon style, clean vector-like design, minimal iconography, high contrast, legible small elements, doodle icons, consistent line weight

✅ 示例重写:

...with small, clearly drawn travel icons such as a hotel, food, camera, and train, drawn in clean, flat icon style with good clarity and consistent line weight.

让 AI 生成一个 主图 + 空白标注框 的版本(构图不变),然后自己用画笔或者 Notion / Canva / Figma 贴上:

  • 更清楚的 icon PNG;
  • 自己画的简单图案;
  • 写上 “DAY 1 - 吃鸭血粉丝汤” 这样的手写标题~

这样图片构图保留,信息表达又完整,超级适合喜欢内容整洁但有细节的人 💡


角色的脸因为太小也画崩,可惜了一些好图,我还挺喜欢的

那种整体氛围对了、色调对了、动作也有趣,偏偏角色脸糊掉/崩掉的图,真的是“心动+心碎”混合体。特别是喜欢的简笔画风/Q 版风,角色小,一旦脸歪了,整个图的情绪就被毁一半。


原因 说明
🤏 角色太小 面部像素本就少,AI 渲染时容易模糊或乱搞细节
🌀 prompt 未锁定面部重点 模型不清楚“脸是核心”,可能当成背景处理了
🖍️ 风格偏极简时更难控脸 简笔画风的脸部其实容错率更低,一个眼睛错位都难看

这句非常有用,可以放到喜欢的 prompt 末尾:

the girl's face is clear, expressive, and symmetrical, drawn in a clean minimal style

可选扩展写法:

face is big enough to show expression, no distortion, cute and clean chibi style

🌟 这句话让 AI 把“脸部质量”当成优先项,有时哪怕整体尺寸小,也会努力把脸画得精致些。


例如:

  • 改成“女孩+行李箱+背景线条地图”构图,而不是整张地图上小小的她;
  • 或者让女孩坐在画面中间、背景元素围绕她排布(让人脸更大);
  • 实在喜欢全景构图的,可以试试“女孩举起一张小地图/手机在前景”,这样脸自然大一些。

可以直接:

  • 脸部 inpaint 修复(如果用的是 MJ / ComfyUI / Photoshop 搭配 SD)
  • 或者把这张图当成“半成品”,生成相似图,然后批量挑“脸好看的版本”

这里设计一个综合版 prompt 模板,兼顾构图、角色大小和细节刻画:


综合版 Prompt 示例:

A simple and cute illustration showing a girl prominently in the foreground, drawn in a chibi or minimal line-art style. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. She is sitting at a laptop, connected via a dotted line and whimsical arrows to a second computer in the background. Around her are small, clearly drawn, flat-style travel icons (or tech icons) such as a map, camera, train, shield, and suitcase. The background is minimal and clean with soft pastel colors and white space, emphasizing clarity and simplicity. The overall style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, clean vector-like icons, good contrast, legible small elements, consistent line weight. Anime style (optional).

小贴士:

  • “girl prominently in the foreground” 让角色更大更突出;
  • “face is clear, expressive, and symmetrical” 重点保证脸部质量;
  • “small, clearly drawn, flat-style … icons” 让图标更简洁明了;
  • “background is minimal and clean” 保持画面清爽不杂乱。

把这段 prompt 拆成几个模块,方便替换主体或动作,只保留风格和整体氛围的描述,快速套用生成不同封面图。


A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. [主体动作或场景描述:例如“she is reading a book” / “she is working on a laptop” / “she is thinking”]. Above her head floats a large question mark (❓), symbolizing curiosity and eagerness to learn.


The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity.


The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).


如果想生成“女孩正在读书”的封面图,套用模板:

A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. She is reading a book. Above her head floats a large question mark (❓), symbolizing curiosity and eagerness to learn. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).

想固定风格,让 AI 自由发挥女孩在不同生活或工作场景的表现,生成多样化图像。


  • 固定风格描述,明确画面基调;
  • 主体“女孩”保留,但不限定动作;
  • 用宽泛的、开放式的场景提示,鼓励多样表现;
  • 避免限制性动作词,改用模糊或多选描述。

A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. She is depicted in various possible daily life or work scenes, engaging in activities such as reading, working, thinking, relaxing, or exploring. Above her head floats a large question mark (❓), symbolizing curiosity and eagerness to learn. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).

  • depicted in various possible daily life or work scenes
  • engaging in activities such as reading, working, thinking, relaxing, or exploring

    这里用“such as”列举了几个示例,但不会限定AI只画这些动作;

  • 这样 AI 会尝试在给定风格下自由发挥动作和场景细节。

可以再加些喜欢的活动词,让它更丰富:

  • cooking, typing, chatting, drawing, drinking coffee, walking, daydreaming 等等

A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. She is depicted in various possible daily life or work scenes, engaging in activities such as [activity list]. Above her head floats a large question mark (❓), symbolizing curiosity and eagerness to learn. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).

整理一份丰富的「女孩日常&工作动作示例清单」,方便快速替换进 prompt 里,多场景多变化,轻松出图!


👧 女孩日常&工作动作示例清单

  • reading a book
  • working on a laptop
  • thinking thoughtfully
  • relaxing with a cup of coffee
  • drawing or sketching
  • cooking in the kitchen
  • chatting on the phone
  • walking in the park
  • daydreaming or gazing into the distance
  • typing on a keyboard
  • holding a smartphone
  • writing in a notebook
  • drinking tea or juice
  • organizing papers or documents
  • listening to music with headphones
  • browsing on a tablet
  • watering plants
  • packing a suitcase
  • shopping with a bag
  • riding a bicycle
  • stretching or doing light exercise
  • taking photos with a camera
  • playing a musical instrument
  • meditating calmly
  • laughing or smiling brightly
  • looking at a map or travel guide
  • cleaning or tidying up
  • painting on a canvas
  • setting up a video call
  • enjoying a snack or dessert

把它直接放到 prompt 的 [activity list] 位置里,或者挑几个喜欢的组合成新的版本。


现在就是:

🎨「多场景视觉导演」+ 🖌️「灵活prompt大师」

但我发现问题了,这种替换比如:A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail. She is depicted in various possible daily life or work scenes, engaging in activities such as cleaning or tidying up. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).会降低出图的质量

主要原因是替换动作后 prompt 变得“更宽泛”或“上下文不连贯”,导致 AI 模型生成时难以聚焦,质量受影响。这里给几点实操建议,提升替换后图像质量:


直接替换单词时,句子结构可能出现不够顺畅或语义不完整,模型难以精准理解。

比如“engaging in activities such as cleaning or tidying up.”这句话太简单,可以稍微丰富成:

“engaging in daily chores such as cleaning and tidying up with focused expression.”


动作越具体,AI越容易还原细节,生成质量越高。

比如:

  • 不只是“cleaning”,而是“carefully dusting a bookshelf”
  • 不只是“working”,而是“typing rapidly on a laptop with concentration”

“activities such as A, B”太干,容易变得机械。建议用完整句子或多分句描述动作和场景。

在替换动作时,确保“风格”和“角色特征”的关键词仍保持完整清晰,不能删减。

比如“expressive and symmetrical face”可以加上“showing gentle smile”或“focused eyes”,这样更生动。

A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail, with a gentle smile. She is carefully dusting a bookshelf, showing focus and care in her movements. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).

  • 多尝试几版动作描述,从简到丰富,找平衡点。
  • 用“and”连接动作和表情细节,增强连贯性。
  • 若条件允许,用 ComfyUI 或 Midjourney 的变体或重试功能,选择最佳版本。

把“动作替换”那部分写得更详细、更生动,直接替换进 prompt 里,既保留风格,也提升生成质量。


She is carefully dusting a bookshelf, showing focus and care in her movements, her eyes gently concentrating on the task.

She is typing rapidly on a laptop, with a focused and determined expression, fingers moving swiftly across the keyboard.

She is reading an open book attentively, occasionally nodding as she absorbs information, with a calm and curious look.

She is sitting relaxed with a cup of coffee in hand, smiling softly as she takes a short break.

She is drawing on a sketchpad, her face showing creative excitement and concentration.

She is chatting happily on the phone, eyes bright and full of energy.

She is walking in a park with a gentle smile, enjoying the fresh air and surroundings.

She is watering plants carefully, with a nurturing and peaceful expression.

She is writing in a notebook thoughtfully, occasionally pausing to think deeply.

She is listening to music with headphones, eyes closed and immersed in the melody.


A simple and cute illustration of a girl drawn in a chibi or minimal line-art style, prominently in the foreground. Her face is clear, expressive, and symmetrical, large enough to show emotion and detail, with a gentle smile. She is carefully dusting a bookshelf, showing focus and care in her movements, her eyes gently concentrating on the task. The background is minimal and clean with soft pastel colors and plenty of white space, emphasizing clarity and simplicity. The style is light, modern, and playful, with no heavy shading or clutter. Masterpiece, best quality, flat illustration, Q-style character, anime style (optional).

💡
欢迎分享文章,或是 来信 与我交流