name

visual-prompt-mastery

description

Expert visual prompt engineering using Emily's proven methodology from 30 engagement-validated examples. Generates structured JSON prompts for image and video generation with technical precision, platform optimization, and intelligent MCP tool selection. Use when creating images, generating videos, writing visual prompts, optimizing prompts for nanobanana gpt-image-1 veotools fal-video, creating Instagram posts LinkedIn graphics Twitter images YouTube thumbnails TikTok Reels videos, or need high-engagement visual content. Includes pattern matching, quality validation, and automatic tool selection.

Visual Prompt Mastery

Overview

This skill transforms simple visual requests into phenomenally good structured prompts using Emily's proven methodology. Based on 30 engagement-validated examples with engagement ranging from 134 to 4,717 likes.

What This Skill Does:

Analyzes your request and matches to proven patterns
Generates comprehensive structured JSON prompts
Applies platform-specific optimizations
Selects optimal MCP tool (nanobanana, gpt-image-1, veotools, fal-video)
Validates prompt quality before generation
Ensures high-engagement visual outcomes

Source: Emily (@IamEmily2050) - 30 complete prompts for Nano Banana (images) and Veo3 (videos)

Core Methodology: Emily's JSON Structure

Every visual prompt follows this comprehensive structure:

For Images (10+ sections required):

{
  "scene": {
    "description": "detailed environment",
    "mood": "emotional tone",
    "environment": "physical setting"
  },
  "subject": {
    "type": "what/who is the focal point",
    "appearance": "specific details",
    "pose": "body position and framing",
    "wardrobe": "clothing and accessories"
  },
  "composition": {
    "framing": "how subject is positioned",
    "aspect_ratio": "4:5 for Instagram, 1:1 for LinkedIn, 16:9 for YouTube",
    "layout": "rule of thirds, symmetry, etc."
  },
  "lighting": {
    "source": "window, softbox, natural, etc.",
    "quality": "soft, harsh, diffused",
    "direction": "camera-left, overhead, etc.",
    "color_temp": "5200K exact Kelvin value"
  },
  "color_palette": {
    "primary": "#1E3A8A",
    "secondary": "#60A5FA",
    "background": "#F8F9FA"
  },
  "camera": {
    "focal_length_mm": 85,
    "aperture_f": 2.8,
    "iso": 400,
    "shutter_s": "1/200",
    "white_balance_k": 5600
  },
  "look": {
    "texture": "film grain, digital noise, smooth",
    "sharpness": "high, soft, selective",
    "effects": "vignette, bokeh, etc."
  },
  "negative_prompt": [
    "minimum 10 items",
    "watermark",
    "text errors",
    "distorted anatomy",
    "poor composition",
    "low quality",
    "blurry",
    "artifacts",
    "oversaturated",
    "cluttered"
  ]
}

For Videos (adds timeline and motion):

{
  "format": "shot type and camera feel",
  "primary_camera_motion": "single slow push-in OR dolly OR static",
  "lens": "24-28mm equiv smartphone OR 85mm portrait",
  "subject": {
    "character_string": "detailed description of who/what",
    "mood": "emotional tone"
  },
  "scene": {
    "location": "specific place",
    "time": "time of day",
    "lighting": "key + fill description",
    "weather": "environmental conditions"
  },
  "visual_details": {
    "beats": [
      {
        "time": "0.0-2.5s",
        "action": "what happens",
        "physics": "how things move naturally"
      },
      {
        "time": "2.5-5.0s",
        "action": "continuation",
        "physics": "fabric, hair, objects"
      },
      {
        "time": "5.0-8.0s",
        "action": "resolution",
        "physics": "settling, final movements"
      }
    ]
  },
  "negative_prompt": ["10+ items for videos too"]
}

Decision Tree: Pattern Selection

When a visual request comes in, match to the appropriate pattern category:

1. Social/Selfie Patterns (High Engagement: 1,478 - 4,717 likes)

Use When:

Instagram posts (casual, relatable content)
Social media profiles
Personal branding
Authentic, unpolished aesthetic
Mirror selfies, casual settings

Load Reference: reference/emily-examples/social-selfie/

Characteristics:

Full-body framing (HEAD-TO-SHOES visible)
Specific camera simulation (iPhone 4/5 era for Y2K)
Lo-fi texture (grain, JPEG artifacts, soft focus)
Everyday wardrobe (relatable, not styled)
Specific aesthetic trends (Y2K, K-Beauty, gaming corner)

Example Pattern - Y2K Mirror Selfie (4,717 likes):

{
  "scene": "casual 2010s phone selfie unintentional snapshot",
  "camera": {
    "device": "iPhone 4/5 era",
    "flash": "on harsh LED with mirror glare",
    "aspect_ratio": "4:3",
    "focus": "soft slight hand-shake blur"
  },
  "look": {
    "texture": "grainy digital noise mild JPEG artifacts",
    "sharpness": "low soft focus"
  },
  "pose": {
    "angle": "mirror selfie FULL BODY",
    "framing": "HEAD-TO-SHOES feet visible not cropped"
  }
}

2. Professional/Editorial Patterns (Engagement: 294 - 1,472 likes)

Use When:

LinkedIn content
Professional headshots
Corporate communications
Editorial photography
High-fashion content
Portfolio pieces

Load Reference: reference/emily-examples/professional-editorial/

Characteristics:

Technical camera precision (Canon EOS R5, exact specs)
Professional lighting setups (softbox, negative fill)
Exact color grading (hex codes, precise adjustments)
Composition rules (rule of thirds, negative space)
Post-processing notes (dodging, grading, grain)

Example Pattern - Glacial Couture (426 likes but highly detailed):

{
  "camera": {
    "system": "Canon EOS R5 full-frame 45MP 14-bit RAW",
    "lens": "RF 85mm f/1.2L use at f/3.5",
    "focal_length_mm": 85,
    "aperture_f": 3.5,
    "iso": 100,
    "white_balance_k": 5600
  },
  "composition": {
    "framing": "subject on right third low horizon negative space behind head"
  },
  "color_palette": {
    "notes": "true blacks vs icy whites with selective warm gold"
  }
}

3. Creative/Artistic Patterns (Engagement: 13 - 1,143 likes)

Use When:

Experimental content
Artistic projects
Viral content attempts
Unique compositions
Cultural mashups (Cat Bus, Kawaii, Tokyo scenes)
Conceptual visuals

Load Reference: reference/emily-examples/creative-artistic/

Characteristics:

Unique subjects (cat bus, banana theme, kawaii illustrations)
Cultural references (Ghibli, K-culture, Y2K nostalgia)
Innovative compositions (POV, unusual framing)
Selective color techniques
Storytelling elements

Example Pattern - Tokyo Rain Cat Bus (1,143 likes):

{
  "view": "POV from inside a car dash-cam perspective",
  "subjects": [
    {
      "id": "cat_bus",
      "style_hint": "Ghibli-inspired soft friendly forms"
    },
    {
      "id": "girl",
      "pose_action": "leans down and kisses the corgi"
    }
  ],
  "cinematography": {
    "lens": "anamorphic look",
    "effects": ["cinematic lens flares", "chromatic aberration"],
    "grade": "high-contrast teal–orange with cyber-pink accents"
  }
}

4. Cinematic/Video Patterns (Engagement: 134 - 1,245 likes)

Use When:

Video generation (TikTok, Reels, YouTube)
Cinematic sequences
Documentary style
Narrative content
Multi-shot projects

Load Reference: reference/emily-examples/cinematic-video/

Characteristics:

Timeline beats (0-2.5s, 2.5-5s, 5-8s structure)
Single camera motion (push-in, dolly, static)
Physics rules (how things move naturally)
Shot composition language (wide, medium, close)
Exact cinematography specs

Example Pattern - CCTV Donkey Baboon (1,245 likes):

{
  "shot": {
    "composition": "static ultra-wide shot high-mounted CCTV",
    "camera_motion": "none locked",
    "frame_rate": "24 fps"
  },
  "timeline": [
    { "time": "0–2s", "action": "subject enters frame" },
    { "time": "2–6s", "action": "main action with physics" },
    { "time": "6–8s", "action": "exit and hold" }
  ],
  "subject": {
    "donkey": {
      "physics": "hooves strike concrete realistic cadence"
    }
  }
}

Instructions: Generating Optimal Prompts

Follow this process when creating any visual prompt:

Step 1: Analyze the Request

Extract key information:

Platform: Instagram, LinkedIn, Twitter, YouTube, TikTok
Content Type: Photo, illustration, diagram, video, thumbnail
Aesthetic: Professional, casual, creative, cinematic
Text Rendering: Does it need readable text?
Quality Need: Social (good enough) vs Professional (exceptional)

Step 2: Match to Pattern Category

Based on Step 1, select pattern category:

Request Type	Pattern Category	Reference to Load
Instagram post, casual photo	Social/Selfie	reference/emily-examples/social-selfie/
LinkedIn graphic, professional	Professional/Editorial	reference/emily-examples/professional-editorial/
Unique concept, artistic	Creative/Artistic	reference/emily-examples/creative-artistic/
Video, cinematic scene	Cinematic/Video	reference/emily-examples/cinematic-video/

Step 3: Load and Adapt Pattern

Read 1-2 relevant examples from matched category
Extract structural pattern (which sections, what level of detail)
Adapt pattern to user's specific request
Maintain Emily's technical precision standards

Example:

User: "Create Instagram post about productivity"
→ Platform: Instagram
→ Type: Photo/illustration
→ Category: Social or Professional (depends on tone)
→ If casual tone: Load social-selfie patterns
→ If professional tone: Load professional-editorial patterns
→ Adapt pattern with productivity-specific elements

Step 4: Populate All Required Sections

Generate comprehensive JSON with:

Minimum Required Sections:

scene (description, mood, environment)
subject (type, appearance, pose, details)
composition (framing, aspect_ratio, layout)
lighting (source, quality, color temperature in Kelvin)
color_palette (hex codes in #RRGGBB format)
camera (focal length, aperture, ISO, white balance)
look or aesthetic (texture, sharpness, effects)
negative_prompt (10+ items minimum)

Optional but Recommended:

wardrobe (for subjects with clothing)
props (environmental elements)
post_processing (for editorial quality)
For videos: timeline beats, physics, camera_motion

Step 5: Apply Platform Optimizations

Platform-specific requirements (see reference/platform-specs.md for full details):

Instagram:

Aspect ratio: 4:5 (1080x1350) for posts, 9:16 for Reels
Style: Creative, vibrant, trend-aware
Text: Minimal, use captions instead
Tool preference: nanobanana (creative + cost-effective)

LinkedIn:

Aspect ratio: 1:1 (1200x1200) or 1.91:1 for articles
Style: Professional, clean, trustworthy
Text: Can include quotes, statistics
Tool preference: gpt-image-1 (professional + text rendering)

YouTube:

Aspect ratio: 16:9 (1280x720)
Style: Bold, high contrast, CTR-optimized
Text: Large, readable at thumbnail size
Tool preference: gpt-image-1 (text rendering critical)

Twitter:

Aspect ratio: 16:9 (1200x675) optimal
Style: Eye-catching, scroll-stopping
Text: Optional but powerful
Tool preference: nanobanana (speed + volume)

TikTok/Reels:

Aspect ratio: 9:16 vertical video
Style: Hook in first 3 seconds
Motion: Single camera move preferred
Tool preference: veotools or fal-video

Step 6: Select Optimal MCP Tool

Use this decision matrix (see reference/mcp-tool-selection.md for full analysis):

Choose gpt-image-1 (OpenAI) when:

Professional/corporate content (LinkedIn)
Text rendering required (thumbnails, quotes)
Photorealism critical (headshots, products)
Budget allows (~$0.04-0.08 per image)
Quality over speed

Choose nanobanana (Gemini) when:

Social/creative content (Instagram, Twitter)
High volume needed (carousels, variants)
Speed important (iterations, testing)
Budget-conscious (~$0.039 per image)
Image editing or multi-image blending
Creative/artistic freedom preferred

Choose veotools (Google Veo 3) when:

Video generation (8-30 seconds)
Cinematic quality needed
Audio/dialogue sync required
Professional video content

Choose fal-video when:

Quick video prototypes
Multiple model comparison needed
Custom model training (Sid's LoRA)
Specific model requirements (Kling, Luma, etc.)

Step 7: Apply Technical Precision Standards

Emily's prompts achieve high engagement through technical precision:

Hex Color Codes:

Always use #RRGGBB format: #0B0B0B (not "black"), #FF4FA3 (not "pink")
Specify primary, secondary, accent colors
Include background color

Camera Specifications:

Focal length: 24mm, 35mm, 50mm, 85mm (exact values)
Aperture: f/1.8, f/2.8, f/5.6 (exact f-stops)
ISO: 100, 400, 800, 1600 (exact values)
White balance: 4200K, 5200K, 5600K (Kelvin exact)
Shutter speed: 1/125s, 1/200s (exact values)

Negative Prompts (Mandatory 10+ items):

Prevent common failures
Platform-specific exclusions
Quality gates
Style anti-patterns

Example negative prompt list:

"negative_prompt": [
  "watermark",
  "text errors",
  "distorted anatomy",
  "extra limbs",
  "poor composition",
  "low quality",
  "blurry",
  "artifacts",
  "oversaturated",
  "harsh shadows",
  "cluttered background",
  "amateur lighting"
]

Step 8: Validate Prompt Quality

Before finalizing, validate against Emily's standards:

Quality Checklist:

All 10+ sections populated
Hex codes in #RRGGBB format
Camera specs with exact values (focal, aperture, ISO)
Color temperature in Kelvin
Negative prompt has 10+ items
Platform aspect ratio correct
Subject description specific (not vague)
Lighting source and quality defined
Composition framing clear

Scoring Framework:

10/10: Emily-level quality (all checks pass, engagement-optimized)
8-9/10: Professional quality (all checks pass)
6-7/10: Good (minor issues, still usable)
<6/10: Needs revision (missing required elements)

If score < 8, revise prompt before returning.

Pattern Templates (Factory Functions)

These are conceptual templates - use them as mental models for generating prompts:

Template 1: Social Selfie (Instagram, Casual)

Use For: Relatable Instagram content, personal branding, casual aesthetics

Key Elements:

Full-body framing (HEAD-TO-SHOES visible)
Specific aesthetic era (Y2K 2010s, K-Beauty studio, gaming corner)
Natural imperfections (grain, soft focus, mild blur)
Everyday wardrobe (relatable, not overly styled)
Environmental storytelling (bedroom, bathroom, PC corner)

Reference Examples:

09-y2k-mirror-selfie - 4,717 likes (nostalgic lo-fi)
12-winking-sofa-k-beauty - 3,229 likes (studio K-Beauty)
11-blue-pc-gaming-corner - 1,478 likes (themed environment)

Adaptation Steps:

Choose aesthetic sub-type (Y2K, K-Beauty, themed)
Load matching example from reference/emily-examples/social-selfie/
Extract camera settings, look/texture, pose structure
Replace subject details with user's requirements
Maintain grain/texture/imperfections (critical for engagement)

Template 2: Professional Editorial (LinkedIn, Corporate)

Use For: Professional content, LinkedIn posts, corporate communications, portfolio

Key Elements:

Precise camera specifications (Canon EOS R5, exact lens)
Professional lighting (softbox, fill, rim)
Clean composition (rule of thirds, negative space)
Color grading notes (cool/warm, selective saturation)
Post-processing guidance (micro-dodge, grain)

Reference Examples:

02-glacial-couture - 426 likes (high-fashion precision)
13-gallery-hallway - 1,472 likes (urban editorial)
03-editorial-red-dress - 294 likes (studio minimalism)

Adaptation Steps:

Define technical requirements (headshot, 3/4 body, environmental)
Load matching editorial example
Extract camera system, lighting setup, composition
Apply to user's subject
Include post-processing notes for quality

Template 3: Creative Artistic (Experimental, Viral)

Use For: Unique concepts, viral attempts, artistic projects, cultural mashups

Key Elements:

Unique subject combinations (cat bus + corgi, banana aesthetic)
Cultural references (Ghibli, kawaii, specific eras)
Innovative compositions (POV, unusual framing, selective color)
Storytelling through environment
Trend-aware aesthetics

Reference Examples:

07-tokyo-rain-cat-bus - 1,143 likes (Ghibli mashup)
10-kawaii-cat-notebook - 610 likes (illustration style)
15-black-white-blue-lingerie - 2,986 likes (selective color)

Adaptation Steps:

Identify the creative concept
Find nearest creative example (subject type, style approach)
Extract compositional innovation
Adapt cultural/aesthetic references
Maintain narrative elements

Template 4: Cinematic Video (TikTok, Reels, YouTube)

Use For: Video content, cinematic scenes, narrative sequences, documentary

Key Elements:

Timeline beats (0-2.5s, 2.5-5s, 5-8s structure)
Single camera motion (push-in, dolly, static - ONE move only)
Physics specifications (how things move naturally)
Shot language (wide, medium, close)
Audio considerations (for Veo3)

Reference Examples:

13-cctv-donkey-baboon - 1,245 likes (static surveillance)
09-italian-siblings - 579 likes (dialogue sync)
08-blue-alpine-gentian - 416 likes (macro dolly-in)

Adaptation Steps:

Define shot type (static, push-in, dolly)
Load matching cinematic example
Extract timeline beat structure
Specify physics for each beat
Include camera technical specs

Composition Patterns: Real Examples

Example 1: Simple Instagram Post Request

User Request: "Create Instagram post about my morning coffee routine"

Process:

Analyze: Platform=Instagram, Type=casual photo, Aesthetic=relatable
Match: Social/Selfie category
Load: Y2K or casual aesthetic pattern
Generate:

{
  "scene": "cozy morning kitchen natural light streaming through window",
  "subject": {
    "type": "flat lay composition",
    "elements": "ceramic coffee mug with latte art notebook with handwritten notes smartphone showing time",
    "arrangement": "rule of thirds with coffee at intersection"
  },
  "composition": {
    "aspect_ratio": "4:5",
    "framing": "overhead flat lay",
    "layout": "diagonal composition from top-left to bottom-right"
  },
  "lighting": {
    "source": "window light from left",
    "quality": "soft diffused morning light",
    "color_temp": "5200K warm daylight"
  },
  "color_palette": {
    "primary": "#8B7355",
    "secondary": "#F5DEB3",
    "background": "#FAFAFA",
    "accent": "#2C1810"
  },
  "camera": {
    "angle": "directly overhead 90 degrees",
    "focal_length_mm": 35,
    "aperture_f": 2.8,
    "iso": 400
  },
  "look": {
    "texture": "light film grain warm cozy feel",
    "sharpness": "high on coffee cup selective on notebook",
    "effects": "soft vignette natural depth of field"
  },
  "negative_prompt": [
    "cluttered",
    "harsh shadows",
    "artificial lighting",
    "watermark",
    "text overlay",
    "messy",
    "dark mood",
    "cold tones",
    "people visible",
    "low quality",
    "blurry",
    "distorted"
  ]
}

Tool selection: nanobanana (Instagram + casual + no text)
Quality estimate: 8/10

Example 2: Professional LinkedIn Graphic

User Request: "Create LinkedIn post graphic about AI productivity with quote"

Process:

Analyze: Platform=LinkedIn, Type=professional graphic, Text=YES
Match: Professional/Editorial category
Load: Clean editorial pattern
Generate:

{
  "scene": "minimal modern office background clean lines neutral tones",
  "subject": {
    "type": "typography-focused composition",
    "main_text": "AI Doesn't Replace You. It Amplifies You.",
    "attribution": "— Your Name",
    "visual_elements": "abstract geometric shapes subtle tech motifs"
  },
  "composition": {
    "aspect_ratio": "1:1",
    "layout": "center-weighted with breathing room",
    "hierarchy": "quote dominant attribution secondary"
  },
  "typography": {
    "quote_font": "Inter Bold",
    "quote_size_pt": 56,
    "attribution_font": "Inter Regular",
    "attribution_size_pt": 24,
    "color": "#0B0B0B"
  },
  "lighting": {
    "quality": "soft even lighting no harsh shadows",
    "mood": "professional trustworthy"
  },
  "color_palette": {
    "primary": "#1E3A8A",
    "secondary": "#60A5FA",
    "background": "#F8F9FA",
    "text": "#0B0B0B"
  },
  "negative_prompt": [
    "cluttered",
    "too many colors",
    "unprofessional",
    "casual aesthetic",
    "low contrast",
    "hard to read",
    "watermark",
    "stock photo feel",
    "cheesy",
    "outdated design",
    "poor typography",
    "text errors"
  ]
}

Tool selection: gpt-image-1 (LinkedIn + professional + text rendering)
Quality estimate: 9/10

Example 3: Cinematic Video Scene

User Request: "Create cinematic video of entrepreneur walking through office at dawn"

Process:

Analyze: Type=video, Style=cinematic, Mood=inspiring
Match: Cinematic/Video category
Load: Dolly/push-in pattern
Generate:

{
  "format": "cinematic office portrait at golden hour",
  "primary_camera_motion": "single slow dolly-in following subject from medium to close-up",
  "lens": "35mm spherical cinematic feel",
  "subject": {
    "character_string": "Entrepreneur in business casual walking through modern office warm confident demeanor holding coffee mug reviewing documents on iPad",
    "mood": "focused determined optimistic"
  },
  "scene": {
    "location": "modern open-plan office floor-to-ceiling windows",
    "time": "dawn golden hour warm sunlight streaming through glass",
    "lighting": "key: natural window light from left fill: office ambient overhead rim: backlight from window",
    "atmosphere": "quiet empty office dust motes in sunbeams"
  },
  "visual_details": {
    "beats": [
      {
        "time": "0.0-2.5s",
        "action": "Establish wide shot of office entrepreneur enters from left walking steadily camera begins slow dolly-in",
        "physics": "natural walking gait coffee liquid gentle sway iPad screen readable"
      },
      {
        "time": "2.5-5.0s",
        "action": "Follow subject through office desks visible in background subject glances at iPad continues walking",
        "physics": "hair slight movement from air conditioning fabric drapes naturally"
      },
      {
        "time": "5.0-8.0s",
        "action": "Close to medium close-up subject stops at window looks out with confident expression holds gaze",
        "physics": "breathing subtle coffee steam rises iPad screen reflects window light"
      }
    ]
  },
  "cinematography": {
    "aperture_look": "f/2.8 shallow DOF background softly blurred",
    "shutter_look": "1/48s cinematic motion blur",
    "color_grade": "warm golden tones with lifted shadows"
  },
  "negative_prompt": [
    "shaky camera",
    "multiple camera moves",
    "artificial movement",
    "harsh lighting",
    "cold tones",
    "cluttered",
    "low quality",
    "compression artifacts",
    "unrealistic physics",
    "teleporting subjects"
  ]
}

Tool selection: veotools (cinematic video + audio support)
Quality estimate: 8/10

Engagement Optimization Insights

Based on Emily's 30 examples, high engagement correlates with:

For Images:

Relatable Aesthetics - Y2K nostalgia (4,717 likes), K-Beauty (3,229 likes)
Full-Body Framing - HEAD-TO-SHOES visible performs better than crops
Lo-Fi Authenticity - Grain, imperfections feel authentic vs over-polished
Specific Scenarios - "Mirror selfie in restroom" > "woman in room"
Cultural Trends - K-Beauty, gaming aesthetic, specific eras
Selective Color - B&W with color isolation (2,986 likes)

For Videos:

Single Camera Motion - One deliberate move (push-in, dolly, static)
Timeline Structure - Clear 3-beat structure (establish, action, resolve)
Realistic Physics - Natural movement (breath vapor, fabric wrinkles)
Specific Scenarios - "CCTV of donkey + baboon" > "animals walking"
Technical Detail - Exact cinematography specs increase perceived quality

See: reference/engagement-insights.md for full analysis

Validation Framework

After generating a prompt, validate quality:

1. Completeness Check (Required)

scene section present with description, mood
subject section with specific details
composition with aspect_ratio
lighting with source and color_temp
color_palette with hex codes
camera with focal_length_mm, aperture_f, iso
negative_prompt with 10+ items

2. Precision Check (Required)

All hex codes match #RRGGBB format
All camera specs are exact values (not ranges)
White balance in Kelvin (not "warm" or "cool")
Aspect ratio appropriate for platform
No vague terms ("nice", "good", "pretty")

3. Platform Check (Required)

Aspect ratio matches platform
Aesthetic appropriate (professional for LinkedIn, casual for Instagram)
Tool selection optimal (text → gpt-image-1, creative → nanobanana)

4. Engagement Check (Recommended)

Matches proven pattern from Emily's examples
Specific scenario (not generic)
Cultural/trend awareness
Storytelling elements present

Overall Score = (Completeness + Precision + Platform + Engagement) / 4

If score < 8/10, revise prompt. If score ≥ 9/10, save as reusable template.

Reference Documentation

This skill includes comprehensive reference materials using progressive disclosure:

Core References (Load as needed):

reference/prompt-patterns.md

Extracted patterns from all 30 examples
Reusable sections by category
Common structures and variations
When to load: Need to see pattern library

reference/platform-specs.md

Complete specs for Instagram, LinkedIn, Twitter, YouTube, TikTok
Aspect ratios, resolutions, design systems
Platform-specific best practices
When to load: Working with specific platform

reference/engagement-insights.md

What drives high engagement (analysis of likes/bookmarks)
Top 10 patterns by engagement
Engagement correlation analysis
When to load: Optimizing for virality

reference/validation-rules.md

Complete validation checklist
Quality scoring rubric
Common failures and fixes
When to load: Quality assurance needed

reference/mcp-tool-selection.md

Complete decision matrix
Cost, speed, quality comparison
Model-specific strengths
When to load: Choosing between tools

Emily's 30 Examples (Load specific category):

reference/emily-examples/social-selfie/ (5 examples, 1,478-4,717 likes)

Y2K aesthetic, K-Beauty, mirror selfies
High engagement, relatable content
When to load: Instagram casual content

reference/emily-examples/professional-editorial/ (6 examples, 294-1,472 likes)

Editorial photography, professional quality
Technical precision, clean aesthetics
When to load: LinkedIn, corporate content

reference/emily-examples/creative-artistic/ (5 examples, 13-1,143 likes)

Experimental, unique, cultural mashups
Creative freedom, artistic expression
When to load: Artistic projects, viral attempts

reference/emily-examples/cinematic-video/ (13 examples, 134-1,245 likes)

Video prompts with timeline beats
Cinematic techniques, motion specs
When to load: Video generation requests

Integration with Existing Skills

This skill automatically integrates with:

ai-image-generator skills:

Works alongside create-image skill
Provides structured prompts for edit-image skill
Enhances youtube-thumbnail-design skill
Optimizes linkedin-design skill

ai-video-agent workflows:

Provides structured video prompts
Applies cinematic patterns
Timeline beat structure

Workflow Integration: This skill is automatically invoked when agents detect visual generation requests. No manual invocation needed.

Philosophy: Maximize Quality Through Precision

Emily's methodology succeeds because:

Technical Precision - Exact specs eliminate ambiguity
Comprehensive Structure - 10+ sections ensure nothing is missed
Negative Prompts - Actively prevent common failures
Proven Patterns - Based on engagement validation
Platform Awareness - Optimized for specific contexts

The Goal: Every visual generated should be phenomenal, not just "good enough."

Quick Start Examples

Quick Example 1: Instagram Post

User: "Create Instagram post showing my new workspace"
→ Load social-selfie pattern (casual, relatable)
→ Generate structured prompt with full-body or environment focus
→ Tool: nanobanana
→ Result: Authentic, engaging workspace photo

Quick Example 2: LinkedIn Graphic

User: "Create LinkedIn graphic with productivity quote"
→ Load professional-editorial pattern (clean, trustworthy)
→ Generate structured prompt with typography focus
→ Tool: gpt-image-1 (text rendering)
→ Result: Professional, readable quote graphic

Quick Example 3: YouTube Thumbnail

User: "Create thumbnail for 'How I Built 10 AI Agents'"
→ Load creative-artistic OR professional (depending on style)
→ Generate structured prompt with bold composition, readable text
→ Tool: gpt-image-1 (text critical)
→ Result: High-CTR thumbnail with clear text

Quick Example 4: TikTok Video

User: "Create TikTok video showing product reveal"
→ Load cinematic-video pattern
→ Generate timeline beats (0-2.5s: tease, 2.5-5s: reveal, 5-8s: feature)
→ Tool: veotools or fal-video
→ Result: Engaging vertical video with clear story

Success Metrics

Prompts generated using this skill should achieve:

Quality score: 8-10/10 consistently
Engagement: Comparable to Emily's benchmarks
Technical: All precision standards met
Platform: Optimized for specific channel
Tool: Optimal MCP tool selected

Dependencies

No external dependencies required. This skill provides guidance for Claude to generate structured prompts.

Optional for future Python module implementation:

json (built-in)
dataclasses (built-in)
typing (built-in)

Version: 1.0 Created: 2025-10-29 Source: Emily's 30 engagement-validated prompts (@IamEmily2050) Methodology: Anthropic progressive disclosure pattern

Install Skill

SKILL.md

Visual Prompt Mastery

Overview

Core Methodology: Emily's JSON Structure

For Images (10+ sections required):

For Videos (adds timeline and motion):

Decision Tree: Pattern Selection

1. Social/Selfie Patterns (High Engagement: 1,478 - 4,717 likes)

2. Professional/Editorial Patterns (Engagement: 294 - 1,472 likes)

3. Creative/Artistic Patterns (Engagement: 13 - 1,143 likes)

4. Cinematic/Video Patterns (Engagement: 134 - 1,245 likes)

Instructions: Generating Optimal Prompts

Step 1: Analyze the Request

Step 2: Match to Pattern Category

Step 3: Load and Adapt Pattern

Step 4: Populate All Required Sections

Step 5: Apply Platform Optimizations

Step 6: Select Optimal MCP Tool

Step 7: Apply Technical Precision Standards

Step 8: Validate Prompt Quality

Pattern Templates (Factory Functions)

Template 1: Social Selfie (Instagram, Casual)

Template 2: Professional Editorial (LinkedIn, Corporate)

Template 3: Creative Artistic (Experimental, Viral)

Template 4: Cinematic Video (TikTok, Reels, YouTube)

Composition Patterns: Real Examples

Example 1: Simple Instagram Post Request

Example 2: Professional LinkedIn Graphic

Example 3: Cinematic Video Scene

Engagement Optimization Insights

Validation Framework

1. Completeness Check (Required)

2. Precision Check (Required)

3. Platform Check (Required)

4. Engagement Check (Recommended)

Reference Documentation

Core References (Load as needed):

Emily's 30 Examples (Load specific category):

Integration with Existing Skills

Philosophy: Maximize Quality Through Precision

Quick Start Examples

Quick Example 1: Instagram Post

Quick Example 2: LinkedIn Graphic

Quick Example 3: YouTube Thumbnail

Quick Example 4: TikTok Video

Success Metrics

Dependencies