ai-poster-workflow

어떤 스킬인가요?

reusable workflow for creating cinematic portrait posters, graduation visuals, social media avatars, and prompt-engineering briefs from user-provided themes, portrait references, school information, and style requirements. use when the task involves ai-assisted visual concept design, portrait-preserving poster prompts, iterative image-generation feedback, or structured creative production.

설치 방법

프로젝트 폴더에서 아래 명령어를 실행하고, 설치할 에이전트를 선택하세요.

설치 명령을 직접 실행해야 적용됩니다. 지원 에이전트와 필요한 권한·라이선스는 제작자의 안내를 확인하세요.

스킬 원문 파일

---
name: ai-poster-workflow
description: reusable workflow for creating cinematic portrait posters, graduation visuals, social media avatars, and prompt-engineering briefs from user-provided themes, portrait references, school information, and style requirements. use when the task involves ai-assisted visual concept design, portrait-preserving poster prompts, iterative image-generation feedback, or structured creative production.
---

# AI Poster Workflow

## Purpose

Use this workflow to turn a user's creative request into a clear, reusable visual-generation brief. The skill focuses on portrait-based posters, graduation commemorative artwork, social media avatars, and cinematic visual concepts.

## Core principles

1. Preserve the subject's identity when a portrait reference is provided.
2. Separate the prompt into clear sections: subject, composition, symbols, style, color, typography, and constraints.
3. Avoid clutter. Select only elements that reinforce the theme.
4. Treat each user revision as design feedback and update the brief precisely.
5. When likeness matters, avoid over-stylization and explicitly keep facial structure, face angle, hairstyle, expression, and clothing cues.

## Standard workflow

1. Collect inputs:
   - portrait or subject description
   - theme
   - school, major, city, or story context
   - preferred style
   - required format
   - text requirements

2. Build a concept:
   - define the central visual metaphor
   - choose 3 to 6 symbolic elements
   - define what should be prominent and what should be subtle

3. Write the prompt:
   - composition first
   - subject and identity constraints second
   - scene elements third
   - style and color fourth
   - typography and negative constraints last

4. Iterate:
   - identify what failed
   - correct only the relevant section
   - keep successful parts stable

## Portrait preservation guidance

When the user wants a result to resemble a real person, include constraints such as:

- keep original face identity
- do not alter facial structure
- preserve face angle, hairstyle, expression, and natural proportions
- use the portrait as the visual base rather than inventing a new face
- avoid generic beauty-face transformation
- avoid excessive smoothing or plastic skin texture

## Output format

Return one of the following depending on the user's need:

- full generation prompt
- short prompt
- revision prompt
- design brief
- checklist for judging the generated result

## Quality checklist

Before finalizing, check:

- Is the subject still recognizable?
- Is the theme clear within three seconds?
- Are there too many unrelated visual elements?
- Is the style consistent?
- Is the typography minimal and appropriate?
- Does the prompt include constraints that prevent common failures?

# Graduation Poster Prompt Example

## Scenario

Create a cinematic graduation poster based on a real portrait reference.

## Input fields

- Portrait reference: provided by user
- Theme: graduation, growth, future, memory
- Style: cinematic poster, soft watercolor, double exposure
- Output: vertical poster for social media and print

## Prompt template

Use the provided portrait as the visual base. Keep the original face identity and natural facial structure. Do not invent a new face. Create a vertical cinematic graduation poster with a large side-profile or half-profile silhouette. Inside the silhouette, build a double-exposure narrative world connected to campus life, reading, writing, future exploration, and personal growth.

Include symbolic elements such as books, handwritten notes, campus architecture, soft flowers, moonlight, butterflies, and a distant future city. Keep the composition minimal, poetic, clean, and balanced. Use soft watercolor texture, paper grain, gentle haze, warm vintage lighting, and restrained typography.

If text is needed, use minimal English title text only, such as:

To the Unwritten
Where words begin, the world unfolds.

## Constraints

- Preserve the subject's facial structure and expression.
- Avoid generic AI beauty-face transformation.
- Avoid cluttered collage composition.
- Avoid too many duplicated flowers or decorative props.
- Do not include other clear people.
- Use a soft, clean, collectible poster style.

## Review checklist

- Does the subject still look like the original person?
- Is the silhouette clearly related to the person?
- Is the graduation theme visible within three seconds?
- Is the image poetic rather than messy?

# AI Poster Agent Skill

A reusable AI workflow for cinematic graduation posters, portrait-based visual storytelling, and structured prompt optimization.

This repository documents a repeatable workflow that turns user-provided portraits, themes, school information, and design preferences into visual concepts and generation prompts.

## What this project does

- Builds structured prompts for graduation posters and commemorative portrait artworks.
- Prioritizes portrait likeness by preserving original facial structure and reference details.
- Produces cinematic double-exposure poster concepts with controlled composition, typography, color, and symbolism.
- Supports iterative refinement around likeness, layout, clutter, and theme adjustment.
- Provides reusable examples for social media avatars, graduation posters, and narrative visual design.

## Typical use cases

1. Graduation poster generation
2. Portrait-based commemorative artwork
3. Social media avatar concept design
4. Prompt engineering for image generation
5. Multi-round creative refinement

## Repository structure

- README.md
- SKILL.md
- examples/graduation-poster-prompt.md
- examples/portrait-preservation-prompt.md
- examples/social-avatar-prompt.md
- outputs/sample-result-description.md

## Example workflow

1. Collect user inputs: portrait, school, major, theme, preferred style, and output format.
2. Analyze identity-sensitive details such as face angle, hairstyle, clothing, and expression.
3. Generate a structured prompt with sections for composition, subject, visual symbols, style, color, typography, and constraints.
4. Iterate based on user feedback, especially around likeness, layout, clutter, and theme relevance.
5. Produce final prompt or concept brief for image generation.

## Notes

This project focuses on workflow design, prompt structure, and repeatable AI-assisted creative production. It does not include private user images or third-party assets.

# Portrait Preservation Prompt Example

## Scenario

Generate an image while preserving the identity of a real person.

## Prompt template

Use the provided portrait as the base reference. Keep original face identity, do not alter facial structure, preserve face angle, hairstyle, expression, and proportions. Avoid generating a generic or beautified face. The output should resemble the real person as closely as possible.

Apply a controlled visual style (e.g., cinematic, soft light, minimal watercolor), but ensure that the style does not override identity. The face should remain realistic and recognizable.

## Constraints

- Do not change facial proportions
- Do not over-smooth skin
- Avoid plastic or artificial texture
- Avoid exaggerated anime or illustration transformation

## Use case

This prompt is used when identity accuracy is more important than style, such as graduation posters, commemorative portraits, or profile images.

## Review checklist

- Is the person still recognizable?
- Is the face structure unchanged?
- Is the style applied subtly rather than dominating the face?

# Social Media Avatar Prompt Example

## Scenario

Generate a stylized but identity-consistent avatar for social media.

## Input

- Portrait reference
- Style preference (clean, cute, minimal, stylized)

## Prompt template

Use the provided portrait as the base reference. Preserve the original identity and facial structure. Create a clean and expressive social media avatar with simplified background and controlled stylization.

Apply a soft stylized treatment (illustration-like or semi-realistic), but keep the face recognizable. Use minimal composition, centered framing, and subtle lighting. Background should be simple (solid color, soft gradient, or light texture).

Optional elements: small icons, subtle glow, soft shadow.

## Constraints

- Avoid over-stylization
- Avoid distortion of face proportions
- Avoid cluttered background
- Keep composition centered and clean

## Output

Square image suitable for profile use.

## Review checklist

- Is the avatar recognizable as the original person?
- Is the composition clean and readable at small size?
- Is the style consistent and not excessive?

# Token Usage Estimate

This document provides a rough estimate of token usage during typical workflows.

## Typical task: graduation poster generation

### Stage 1: requirement analysis
- ~300–600 tokens

### Stage 2: prompt drafting
- ~500–1200 tokens

### Stage 3: iteration (2–5 rounds)
- ~1000–3000 tokens

### Stage 4: refinement and final output
- ~300–800 tokens

## Total per task

Approximate range:
- 2000 – 5000 tokens per complete workflow

## Notes

- Complex identity-preserving tasks require more iterations.
- Visual clarity and prompt precision reduce total token usage over time.
- Structured prompts help maintain consistency and reduce rework.

This estimate is based on repeated usage in creative design scenarios such as poster generation and portrait-based visual concepts.

# Workflow

This document describes the practical workflow used in this repository for AI-assisted visual creation.

## Step 1: Requirement definition

Collect structured input:
- portrait reference (optional but recommended)
- theme (graduation, growth, memory, etc.)
- context (school, major, story)
- preferred style (cinematic, watercolor, minimal, etc.)
- output format (poster, avatar, etc.)

## Step 2: Concept design

- Define a core visual idea (e.g., silhouette double exposure)
- Select limited symbolic elements (3–6 items)
- Decide hierarchy (main vs supporting elements)

## Step 3: Prompt construction

Organize prompt into sections:
- composition
- subject and identity constraints
- scene elements
- style and lighting
- color palette
- typography (if needed)
- negative constraints

## Step 4: Generation and review

Evaluate output based on:
- likeness
- composition clarity
- thematic relevance
- visual clutter

## Step 5: Iteration

- adjust only necessary parts
- preserve working components
- refine prompt progressively

## Step 6: Final output

Deliver:
- final prompt
- description
- optional design notes

This workflow is designed for repeatable use and can be applied across different creative tasks.