SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillscomic-explainer-illustration

comic-explainer-illustration

Design & creation

View repository

Updated 10-15 20:12

About this skill

Comic-style explanatory image

Effect preview

Effect preview

Turn complex content into clear, lightweight, comic-feeling explanatory image prompts; if the user requests direct image generation, then organize the final prompt and call imagegen skill's image_gen tool.

Top metadata

  • Default ratio：16:9
  • Default language: Chinese
  • Default art style: white background, black lines, simple expressions, clear actions, a small amount of accent color
  • Default content strategy: follow the Pareto principle, keep only the most important 20% of information to express 80% of the meaning
  • Direct image generation: when the user says "generate image", "directly generate image", "draw for me", "output image", or "generate image", first organize the final prompt, then call image_gen
  • Continuous dialogue default: if the previous image or previous version has already used this skill, and the user has not explicitly changed characters, then continue using the same set of character settings, clothing, hairstyle, line style, and accent color

Core Workflow

Step 1: Determine the deliverable

  • User only wants the prompt: output the final prompt.
  • User wants an image: organize the final prompt and call image_gen。
  • User provided long text, an article, an opinion, or a process: first compress it into visual key points; do not copy the original text.

Step 2: Extract content using the Pareto principle

Extract only the small amount of information that best explains the topic:

  • One central idea: what this image wants people to remember.
  • Two to four key elements: characters, objects, actions, arrows, speech bubbles, labels.
  • At most one relationship path: cause and effect, comparison, steps, choices, misconception to correct solution.
  • A small amount of text: 1 title, 2 to 5 keywords, 1 to 3 short bubbles.

It is strictly forbidden to put long paragraphs into the image. If the user's input is very long, first translate it into image actions, simple symbols, and short labels.

Step 3: Choose the comic structure

Choose one structure based on the content; do not mix too many:

  • Single scene: one or two people use actions to explain one concept.
  • Left-right comparison: wrong approach vs correct approach, or complex state vs simplified state.
  • Three-panel storyboard: problem appears, key turning point, result presented.
  • Dialogue explanation: two people have a brief conversation, using bubbles to present viewpoint conflict or a solution.
  • Process path: characters move along arrows and pass through no more than 3 nodes.

Visual Style

  • The image is horizontal 16:9, white or very light background, with plenty of blank space.
  • Use black hand-drawn lines; the lines should be clean and relaxed, not heavy and messy.
  • Facial expressions should be simple: smile, confusion, surprise, seriousness, sudden realization are enough.
  • Character actions should be clear: pointing, handing over, pushing away, connecting, observing, discussing, holding up a sign, walking toward a goal.
  • Use only 1 to 2 accent colors, for example blue, yellow, red, or green, for arrows, key labels, speech bubble borders, and small props.
  • Text must be minimal, short, and readable, suitable for Chinese; do not use dense layout, do not use long sentences.
  • The composition should look like a "comic-style explanatory diagram," not a poster, advertising KV, complex commercial illustration, realistic photography, or 3D rendering.

Character Consistency

In multi-turn dialogue, series images, or multi-image tasks, character consistency constraints must be explicitly written in:

  • Keep the same character's gender/age impression, hairstyle, facial proportion, clothing silhouette, main-color small accessories, and line style.
  • Only adjust actions, poses, props, and expressions according to the new theme.
  • If there are two fixed characters in the image, describe them with stable code names, for example "Character A: short hair, round face, blue top; Character B: long hair, yellow scarf."
  • Do not let characters suddenly become different ages, belong to different clothing systems, or switch to different art styles within the same image set.

Prompt Writing Rules

  • First write "overall style," then "theme distillation," then "image composition," and finally "text and constraints."
  • Translate abstract viewpoints into actions and props; do not just pile up text boxes.
  • Explicitly state in the prompt: "follow the Pareto principle, extract only the key points."
  • Specify "no large blocks of text, no dense small text, no complex background."
  • Specify "simple facial expressions, clear body actions, black line art, minimal accent colors".
  • If the user does not provide a title, generate a short Chinese title of 4 to 10 characters.
  • If the user provides very complex content, output the clearest single option by default; only provide 2 to 3 directions when the user asks for multiple options.

Default Output Template

When only the prompt is needed, output this structure:

# 漫画式说明图提示词

## 最终提示词
[可直接用于出图的完整提示词]

The final prompt can be naturally rewritten according to the following skeleton:

请生成一张 16:9 横向漫画式说明图，中文，白色或极浅背景，大量留白。整体风格是简洁手绘漫画说明图：黑色线条为主，人物表情简单但明确，动作清晰，使用 1 到 2 种少量强调色点出重点。画面遵循帕累托法则，只提取最重要的少量信息来解释主题，不要出现大段文字、密集小字、复杂背景或装饰性堆叠。

主题：[提炼后的主题]
核心观点：[一句话核心观点]

画面构图：
- [漫画结构：单幅场景 / 左右对比 / 三格分镜 / 对话说明 / 流程路径]
- [人物、动作、道具、箭头或气泡如何表达主题]
- [视觉重点如何落在 1 到 3 个关键元素上]

人物设定：
- [角色 A 的稳定特征]
- [角色 B 的稳定特征，如需要]
- 如果这是系列图或持续对话，沿用上一张/上一版同一人物的年龄感、发型、脸型比例、服装轮廓、主色配件和线条风格，只调整动作、姿势、道具和表情。

图中文字只保留短标题和少量关键词，可包含：
- 标题：[4 到 10 字中文标题]
- 关键词：[2 到 5 个短词]
- 对话气泡：[1 到 3 句短句，每句尽量不超过 12 个字]

限制：不要写长段落；不要把用户原文整段放进图里；不要写实摄影；不要 3D 渲染；不要复杂渐变；不要密集图表；不要让文字遮挡人物动作。图片比例 16:9。

Output Modes

Only the prompt

If the user explicitly requests "only the prompt," "no image generation," or "give me prompt," output only the final prompt.

Direct image generation

If the user requests image generation, drawing, or image creation:

  1. Internally complete content compression and the final prompt.
  2. Call image_gen to generate the image.
  3. After generation, keep the reply brief and do not output long analysis.

Multiple options

If the user says "give me several directions" or "give me 3 versions," output at most 3 options. If the user requests image generation, then call according to the user's selection or by default prioritize the first option image_gen。

Do / Don't

Do

  • Compress the content first, then visualize it.
  • Use actions, expressions, props, and arrows to carry meaning.
  • Make every speech bubble short and useful.
  • Maintain character consistency across series images.
  • Default 16:9.
  • When generating images, use image_gen。

Don't

  • Do not stuff a complex article directly into a screen full of text.
  • Do not use too many characters, too many panels, or too many colors.
  • Do not make it into a PPT infographic, realistic poster, commercial advertisement, or 3D image.
  • Do not let characters only pose without an explanatory function.