four-panel-comic-explainer-illustration

Four-Panel Comic Explanatory Diagram

Effect Preview

Convert the user's given topic, viewpoint, article passage, concept, process, conflict, or before-and-after change into a 16:9 four-panel comic-style explanatory diagram: continuous small scenes explain one concept, character expressions are simple, actions are clear, black lines, with a small amount of accent color.

Top-Level Conventions

Default ratio：16:9
Default language: Chinese
Default output: directly generate the image
Underlying tool: use the system imagegen skill's default built-in image_gen tool
Default structure: four continuous panels, read in the order top-left, top-right, bottom-left, bottom-right
Core principle: follow the Pareto principle, extract only the key points, and do not include large blocks of text
Prompt only: only when the user explicitly says "only the prompt / only give the prompt / do not generate an image" should you output the complete prompt without calling the image generation tool

Applicable Scenarios

Prefer for:

Using a four-panel comic to explain an abstract concept, product feature, methodology, or technical change
Telling a short story to express the change from confusion to understanding, from problem to solution, or from the old way to the new way
Making lightweight explanatory diagrams for articles, PPTs, courses, or WeChat public account graphics and text
Knowledge illustrations that need character actions, expressions, conflict, and twists to help understanding

If the user wants realistic photography, commercial posters, 3D rendering, complex information architecture diagrams, serious whiteboard flowcharts, pure UI screenshots, or Xiaohongshu covers, do not use this skill.

Reference Files

references/visual-style.md: fixed visual specifications, four-panel narrative structure, and prompt template. Must be read before writing the final imagegen prompt or generating an image.

Core Workflow

Step 1: Determine the output mode

Default mode: organize the final image generation prompt and directly call image_gen
Only when the user explicitly says "only the prompt," "do not generate an image," or "give me the prompt" should you output only the final prompt
If the user provides content and says "draw an image / generate / add an illustration / four-panel comic," do not repeatedly ask follow-up questions; directly do the visual translation and generate the image

Step 2: Compress the content into a four-panel narrative

First distill the most important 20% of the information, then break it into four panels. Each panel retains only one core action, conflict, or change point.

Common four-panel structures:

Problem to solution: confusion -> attempt -> discover the key -> result improves
Before-and-after comparison: old way -> pain point exposed -> new way intervenes -> new result
Concept explanation: misunderstanding -> analogy -> mechanism -> one-sentence summary
Process change: input -> processing -> turning point -> output

Text rules in the image:

Keep the total text to 4 to 8 short Chinese labels
Each panel has 0 to 2 short labels, each label should be 2 to 8 characters if possible
Do not copy the user's original text, do not include body paragraphs, do not include long explanatory sentences

Step 3: Read the fixed art style

Before writing the final imagegen prompt, read references/visual-style.md, and merge its composition, lines, characters, colors, text, and prohibitions into the prompt.

Step 4: Write the final imagegen prompt

The prompt must include:

Four-panel comic-style explanatory image, 16:9
Continuous small scenes explain one concept
The four panels are read top-left, top-right, bottom-left, bottom-right, with clear story logic
Character expressions are simple, actions are clear, conflict is obvious
Mainly black lines, with a small amount of accent color
Each panel keeps only one core action or conflict
Follow the Pareto principle, extract only the key points, no large blocks of text
Chinese short labels are accurate and clear, with no typos, garbled characters, or extra text

Step 5: Call imagegen

When direct image generation is needed:

Use the organized final prompt to call the built-in image_gen
Generate only the image the user requested at one time; if the user requests multiple images, generate them separately
After generation, do not output a long explanation; keep it brief

Output Modes

Generate the image directly by default

Call image_gen to generate the image. Keep the reply brief and do not output long analysis.

Prompt only

If the user explicitly requests only the prompt, output:

# 四格漫画说明图提示词

## 四格拆解

- 第 1 格：[核心动作/冲突]
- 第 2 格：[核心动作/冲突]
- 第 3 格：[核心动作/冲突]
- 第 4 格：[核心动作/冲突]

## 最终提示词

[可直接用于 imagegen 的完整提示词]

Text Stability Fallback

If the user reports typos, garbled characters, missing characters, or unreadable text in the image, prioritize the following next time:

Keep only 0 to 1 short label per panel
Shorten labels to 2 to 5 Chinese characters
Delete all auxiliary text and keep only key node words
Emphasize "text as an independent typography layer, Chinese is clear and sharp, do not use sloppy handwriting"

Do / Don't

First understand the content, then convert it into a four-panel narrative
By default, directly call imagegen to generate the image
Make the four panels have clear introduction, development, turn, and conclusion
Use character actions and expressions to express changes, rather than relying on large blocks of text explanation
Keep 16:9, suitable for placing in PPTs and articles
Use black lines and a small amount of accent color, with a clean image and white space

Don't

Do not put entire paragraphs of the user's original text into the image
Do not make dense text, small-text annotations, or complex tables
Do not use 3D, realistic photography, heavy shadows, complex gradients, or commercial poster style
Do not cram multiple events into each panel
Do not add logos, watermarks, real App screenshots, or irrelevant decorations