Canvas Design Skill

Visual art creation for print and digital graphics. Communication through form, space, and color — not text.

Process

1. Understand the intent — What feeling should this evoke?
2. Create design philosophy (.md) articulating aesthetic movement
3. Express visually on canvas (.pdf/.png)
4. Generate 3-5 distinct variations exploring different aesthetic directions
5. Iterate on user favorites

Tools & Stack

Primary (Recommended)

Secondary

Typst Quick Reference

// Basic document
#set page(
  width: 1200pt,
  height: 630pt,
  margin: 0pt,
  fill: rgb("#0a0f1a")
)

#set text(font: "Outfit", fill: white)

// Positioned element
#place(
  top + left,
  dx: 100pt,
  dy: 100pt,
  text(size: 72pt, weight: "bold", "Your Title")
)

// Rectangle
#place(
  bottom,
  rect(width: 100%, height: 80pt, fill: rgb("#00d4aa"))
)

What "Distinct" Means

Variations must differ in AT LEAST 3 dimensions:

NEVER submit variations with same layout, different colors. That's lazy.

Aesthetic Movements to Draw From

Minimalism

• Maximum restraint
• Abundant white (or negative) space
• One or two focal elements
• Typography as hero

┌─────────────────────────────────────┐
│                                     │
│                                     │
│            WORD                     │
│                                     │
│                                     │
└─────────────────────────────────────┘

Swiss/International

• Grid-based, mathematical
• Sans-serif typography (Helvetica roots, but use Outfit/Manrope now)
• High contrast
• Objective, rational

Bauhaus

• Primary colors + black/white
• Geometric shapes
• Form follows function
• Bold, simple geometry

Japanese Minimalism (Ma)

• "Negative space as form"
• Asymmetric balance
• Muted palette
• Implied rather than stated

Brutalism

• Raw, unpolished
• Exposed structure
• Monospace type
• Intentionally jarring

Neo-Retro

• 70s/80s color palettes
• Grain, noise, analog textures
• Warm earth tones
• Serif display fonts

Color Theory (Quick Reference)

Monochromatic

One hue, multiple values. Safe, elegant.

--base: #0d9488;
--light: #5eead4;
--dark: #115e59;

Complementary

Opposite on wheel. High contrast, vibrant.

--primary: #0d9488; /* teal */
--accent: #f97316;  /* orange (opposite) */

Analogous

Adjacent on wheel. Harmonious, calm.

--a: #0d9488; /* teal */
--b: #0ea5e9; /* blue */
--c: #10b981; /* green */

Triadic

Evenly spaced. Balanced, dynamic.

--a: #dc2626; /* red */
--b: #0d9488; /* teal */
--c: #c9a227; /* gold */

Composition Techniques

Rule of Thirds

Divide canvas into 3x3 grid. Place focal points at intersections.

┌───────┬───────┬───────┐
│       │       │   X   │  ← focal point
├───────┼───────┼───────┤
│       │       │       │
├───────┼───────┼───────┤
│       │       │       │
└───────┴───────┴───────┘

Golden Ratio

Spiral from corner. Place key elements along the curve.

Edge Tension

Place elements close to (but not touching) edges. Creates visual tension.

Asymmetric Balance

Heavy element + light element with more space = balanced.

Overlap

Elements breaking grid boundaries create depth.

Typography as Visual Element

Type as Art

When text IS the design, not just on the design:

#place(
  center + horizon,
  rotate(-5deg,
    text(
      size: 200pt,
      weight: "black",
      tracking: -0.05em,
      fill: rgb("#1a1a1a").lighten(90%),
      "WORD"
    )
  )
)

// Smaller text on top
#place(
  center + horizon,
  text(size: 48pt, weight: "bold", "Your Message")
)

Hierarchy Through Scale

Display: 72-200pt Headline: 36-48pt Body: 14-18pt Caption: 10-12pt

Output Specifications

Sharp Export Examples

// High-quality PNG
await sharp(input)
  .resize(1200, 630, { fit: 'cover' })
  .png({ quality: 100 })
  .toFile(output);

// Print-ready (300 DPI at 4x6")
await sharp(input)
  .resize(1200, 1800) // 4*300 x 6*300
  .png({ quality: 100 })
  .toFile('print-ready.png');

Iteration Workflow

1. User describes intent — "I need a thumbnail for a video about AI agents"
2. Create 3-5 variations exploring different directions
3. User picks favorites — "I like #2 and #4"
4. User explains what resonated — "The bold type, the dark background"
5. Generate 3 more inspired by those elements
6. Final should feel like 10th iteration, not first draft

Principles

1. Text as visual accent, not explanation — If you need words to explain the design, the design failed
2. Communication through form, space, color — Visual language, not verbal
3. Intentionality over safety — Commit to a bold direction
4. One idea per canvas — Don't dilute with multiple messages
5. Hierarchy is everything — What do you see first? Second? Third?

BANNED

Elements (Never Use):

• Purple/pink/blue gradients
• Helvetica, Arial, system fonts
• Centered-everything without visual tension
• Lazy noise overlays (unless intentional aesthetic)
• Fake CLI/terminal aesthetics
• Editorial/newspaper layouts
• Perfect bilateral symmetry
• Floating shapes with no focal point
• Geometric blob SVGs
• Stock photo vibes

Behaviors (Never Do):

• Submit first draft as final
• Use same layout with different colors
• Add text to "explain" the visual
• Default to safe, boring choices

Example Workflow: Thumbnail

User: Create a thumbnail for "Building AI Agents in 2026"

1. Philosophy: Bold, technical, futuristic. Dark mode. Agent = helper.

2. Variations:
   /1 - Minimal: Large "AI AGENTS" text, subtle circuit pattern
   /2 - Conceptual: Abstract network nodes, no text
   /3 - Character: Robot icon + human silhouette partnership
   /4 - Technical: Code fragment + glow effect
   /5 - Bold type: "2026" massive, "AI AGENTS" smaller

3. User: "I like #1 and #4 — the minimal approach with technical hints"

4. Generate 3 more:
   /6 - Merge: Large text + subtle code fragments
   /7 - Variation: Monospace type + accent color bar
   /8 - Refined: #4 with better hierarchy

5. User: "#8 is close, make the glow less intense"

6. Final adjustments → export at 1280x720 PNG

Templates by Use Case

Thumbnail (YouTube)

#set page(width: 1280pt, height: 720pt, margin: 0pt)
#set text(font: "Bebas Neue", fill: white)

#place(
  center + horizon,
  text(size: 120pt, "YOUR TITLE")
)

#place(bottom, rect(width: 100%, height: 4pt, fill: rgb("#00d4aa")))

OG Image

#set page(width: 1200pt, height: 630pt, margin: 0pt)

// Background gradient (not purple!)
#place(
  rect(
    width: 100%,
    height: 100%,
    fill: gradient.linear(
      rgb("#0a0f1a"),
      rgb("#1a1a2e"),
      angle: 45deg
    )
  )
)

// Title
#place(
  left + horizon,
  dx: 80pt,
  text(size: 64pt, weight: "bold", fill: white, "Title Here")
)

Poster (Minimal)

#set page(paper: "a3", margin: 2cm)
#set text(font: "Outfit", size: 14pt)

#place(top + left, text(size: 120pt, weight: "black", "WORD"))

#place(bottom + right, text(size: 14pt, "supporting text"))

Output Format

After generating visuals, output:

✅ SUCCESS: Generated [N] visual variations
📊 METRIC: variations=[count]
📊 METRIC: aesthetic=[movement name]
📊 METRIC: format=[PDF/PNG]
📊 METRIC: dimensions=[WxH]

Files:
- /1: [description] → [path]
- /2: [description] → [path]
...

Ready for feedback — which directions resonate?

After iteration:

✅ SUCCESS: Refined design based on feedback
📊 METRIC: inspired_by=[#2, #4]
📊 METRIC: preserved=[bold type, dark background]
📄 OUTPUT: [final file path]

For single outputs:

✅ SUCCESS: Created [asset type] in [aesthetic] style
📊 METRIC: format=[PDF/PNG]
📊 METRIC: size=[dimensions]
📄 OUTPUT: [file path]

Communication through form, space, and color. Not text.