SuperAPI AI Directory

graphic-designer

Design & creation

View repository

Updated 10-15 23:57

About this skill

Graphic Designer

Create effective visual communication through research-backed design principles.

Design = Communication + Aesthetics — Good design is invisible: it guides the eye, conveys the message, and feels "right" without effort.


Quick Workflow

1. PURPOSE   — What should viewer DO after seeing this?
2. AUDIENCE  — Who? What culture? What device?
3. HIERARCHY — What's #1, #2, #3 in importance?
4. LAYOUT    — Sketch placement (Z or F pattern)
5. COLORS    — 60-30-10 rule (check cultural meaning!)
6. TYPE      — Max 2 fonts (1 display + 1 body)
7. ELEMENTS  — Add graphics, icons, photos
8. REFINE    — Remove until it breaks, then add back
9. CHECK     — Squint test, mobile test, contrast check
10. REVIEW   — Self-critique and iterate (optional loop)

Self-Review Loop (Auto-Improvement)

Once the work is created, you can turn on the loop to analyze and improve automatically.

When should you ask the User

Ask the user before starting the loop:

"Would you like me to turn on the Auto-Improvement Loop? I will analyze the results and refine repeatedly until good quality is achieved."

Options:

  1. Quick Review — 1 round of analysis + fixes if necessary
  2. Full Loop — Re-analyze until it passes every item in the Quality Checklist
  3. Skip — Deliver right away, no review needed

Review Loop Workflow

┌─────────────────────────────────────────────────────┐
│  1. CREATE — สร้างผลงาน (slides, graphics, etc.)    │
└─────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────┐
│  2. ANALYZE — วิเคราะห์โครงสร้าง                    │
│     • อ่าน XML/code ที่สร้าง                        │
│     • ตรวจสอบ positions, sizes, colors             │
│     • สร้าง thumbnails (ถ้าเป็น PPTX)              │
└─────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────┐
│  3. CRITIQUE — วิจารณ์ตาม Design Principles         │
│     ใช้ Review Checklist (ด้านล่าง)                 │
└─────────────────────────────────────────────────────┘
                          ↓
              ┌───────────────────┐
              │  ผ่านทุกข้อ?      │
              └───────────────────┘
               ↙ NO          YES ↘
┌──────────────────┐      ┌──────────────────┐
│  4. FIX          │      │  5. DELIVER      │
│  แก้ไขปัญหา       │      │  ส่งมอบผลงาน     │
│  ที่พบ           │      │                  │
└──────────────────┘      └──────────────────┘
          ↓
    กลับไป Step 2
    (max 3 iterations)

Review Checklist (Score 0-10)

Category Check Weight
Hierarchy #1 element clearly dominant 2x+ over the rest? High
Contrast WCAG AA (4.5:1)? High
Alignment Elements align to grid? Medium
White Space ≥20% empty space? Medium
Color Count ≤4 colors? Medium
Font Count ≤2 fonts? Low
Visual Impact Is there a focal point that draws the eye? High
Brand Consistency Does it match the brand guidelines? High

Pass threshold: Average ≥ 7/10 and no High-weight items below 6

Visual Impact Boost

If the Visual Impact score is low, you can use these skills to help:

Need Skill Use Case
Hero images /art-director Gen prompt for an AI image with good composition
Local image gen /comfyui-user Gen images via the ComfyUI server on the local machine
Decorative elements /geometric-elements Create corners, lines, patterns

Example integration:

1. สร้าง slide แล้วพบว่า visual impact ต่ำ
2. ใช้ /art-director สร้าง prompt สำหรับ background image
3. ใช้ /comfyui-user gen รูป (หรือ cloud API)
4. ใส่รูปใน slide แล้ว re-analyze

Iteration Limits

  • Max iterations: 3 rounds (prevent infinite loop)
  • Stop early if: The User says to stop, or the score does not increase for 2 consecutive rounds
  • Report: Summarize the changes made in each iteration

Design Principles (Summary)

CRAP Principles

Principle What How
Contrast Make differences obvious Size, color, weight
Repetition Create consistency Reuse colors, fonts
Alignment Connect visually Grid, edges
Proximity Group related items Spacing

→ Details: references/gestalt.md

Visual Hierarchy (order of impact)

  1. Size — Larger = more important
  2. Color/Contrast — Bright catches eye first
  3. Position — Top-left (Western), top-right (RTL)
  4. White Space — Isolation creates emphasis
  5. Weight — Bold stands out

Reading Patterns

Pattern Best For Flow
Z-Pattern Visual/marketing Top-L → Top-R → Bottom-L → Bottom-R
F-Pattern Text-heavy Horizontal scans + vertical down left

Color System

60-30-10 Rule

% Role Example
60% Dominant Background
30% Secondary Containers, cards
10% Accent CTAs, highlights

Quick Palettes

Mood Colors
Professional Navy + White + Gold
Energetic Orange + Black + White
Calm Blue + Light Gray + White
Premium Black + Gold + White
2025 Trend Dark + Neon accent

Cultural Color Meanings (Check First!)

Color Western East Asia Thai Context
Red Danger, urgency Luck, joy Auspicious
White Pure, clean Mourning Formal/Mourning
Yellow Optimism Sacred Royal
Gold Luxury Prosperity Premium

→ Full guide: references/color-theory.md

Accessibility (WCAG)

Standard Normal Text Large Text (18pt+)
AA (Minimum) 4.5:1 3:1
AAA (Enhanced) 7:1 4.5:1

Tool: WebAIM Contrast Checker


Typography

Quick Rules

  • Max 2 fonts — 1 display + 1 body
  • Hierarchy via size — Not font changes
  • Line height — 1.4-1.6 for body, 1.1-1.2 for headlines

Safe Font Pairs

Display Body Mood
Montserrat Bold Open Sans Modern
Playfair Display Lato Elegant
Kanit Bold Sarabun Thai-friendly

→ Full guide: references/typography.md


Layout

8px Spacing System

Size Use
8px Within groups
16px Between elements
24-32px Sections
48px Page margins

Social Media Dimensions

Platform Ratio Size
YouTube Thumbnail 16:9 1280×720
Instagram Post 1:1 1080×1080
Instagram Story 9:16 1080×1920
Facebook/LinkedIn 1.91:1 1200×630

→ Layout templates: references/layouts.md


Presentation Slides

Core Rules

Rule Guideline
One idea per slide Single focused message
Rule of 4 Max 4 bullets, 4 words each
Don't compete Audience can't read AND listen

Font Sizes

Context Titles Body Captions
Large room 60pt+ 40pt+ 24pt+
Virtual 44pt+ 32pt+ 20pt+

→ Full guide: references/presentation-design.md


YouTube Thumbnails

Element Recommendation
Faces Use expressive faces (+20-30% CTR)
Text Minimal, bold, curiosity
Colors High contrast, 3-4 max
Mobile Readable at small size

Layout: Face 40%+ height, eye contact, blur background


Common Mistakes

Mistake Fix
Too many fonts Max 2
No hierarchy Make #1 thing 2x bigger
Centered everything Use left-align + proximity
Rainbow colors Pick 1 accent color
Filled every space Add 20% more white space
Text on busy photo Add overlay or blur
Ignoring culture Check color meanings

Checklists

Before Designing

  • What's the ONE message?
  • Who's the audience? (culture, device)
  • What emotion should it evoke?

Quality Check

  • Clear hierarchy? (squint test)
  • Readable at target size?
  • Max 3-4 colors, 2-3 fonts?
  • Contrast 4.5:1+? (WCAG AA)
  • Aligned to grid?
  • Enough white space?

Tools Integration

/geometric-elements — Decorative Assets

python scripts/generate.py shape --style star --color "#D4A84B" --size 100
python scripts/generate.py corner-accent --color "#D4A84B" --size 150
python scripts/generate.py line-divider --color "#D4A84B" --width 800

/art-director — Visual Impact (AI Image Prompts)

When you want an image with good composition and visual impact:

1. Invoke /art-director พร้อมบอก context (slide theme, mood, brand)
2. ได้ prompt ที่พร้อมใช้กับ AI image generators
3. Gen รูปผ่าน /comfyui-user หรือ cloud API

AI Slide Backgrounds (Best Practices)

Workflows that work well:

1. ขอดู REFERENCE — ให้ user แชร์ตัวอย่าง slides ที่ชอบ
2. วิเคราะห์ PATTERNS — สรุป style (colors, elements, layout)
3. Gen ทั้ง BACKGROUND — ไม่แยกชิ้น, ไม่มี text
4. ITERATE — ปรับ size/position ตาม feedback
5. User วาง TEXT เอง — ยืดหยุ่นกว่า

✅ Do:

Technique Why
Gen complete background Gets a more cohesive, beautiful image than separate pieces
Use --edit with the logo as a ref The AI sees the actual shape, no need to guess
Say "NO TEXT" Text from AI is often wrong/ugly
White background Nano Banana Pro cannot make transparent
Start simple A small logo in one corner is better than a lot in every corner

❌ Don't:

Technique Problem
"TRANSPARENT BACKGROUND" Gets a fake checkerboard
Describe the logo shape yourself If the AI misinterprets, use --edit instead
Gen pieces separately then assemble Elements do not match each other
Too much decoration Too cluttered, not professional

Prompt Template:

Professional presentation [TYPE] slide background,
16:9 aspect ratio. NO TEXT. [STYLE] STYLE.
BACKGROUND: [color, grid, gradient]
DECORATIVE: [small/subtle elements, specific corners]
LAYOUT: Leave [area] empty for [content]
COLORS: [hex codes]

/comfyui-user — Local Image Generation

For gen images on the local machine via ComfyUI:

1. Invoke /comfyui-user พร้อมบอกว่าต้องการรูปแบบไหน
2. เลือก workflow (turbo สำหรับ gen ใหม่, edit สำหรับแก้รูปเดิม)
3. ได้รูปพร้อมใช้ใน slides

Integration Skills

Need Skill When to Use
Decorative elements /geometric-elements corners, lines, patterns, shapes
AI image prompts /art-director hero images, backgrounds that need good composition
Local image gen /comfyui-user gen images via the ComfyUI server
PowerPoint slides /pptx Create/Edit PPTX files
ThepExcel brand /thepexcel-brand-guidelines brand colors, fonts, logo usage

References

Topic File
Color theory references/color-theory.md
Typography references/typography.md
Layouts references/layouts.md
Presentation design references/presentation-design.md
Gestalt principles references/gestalt.md

Related Skills

  • /geometric-elements — Generate decorative assets for designs
  • /thepexcel-brand-guidelines — Apply brand colors and typography
  • /art-director — For photography/cinema prompts (not layout)
  • /pptx — Design presentation slides