@yeongjaeyou/claude-code-config 0.18.6 → 0.19.0

package/.claude/skills/midjourney-imagineapi/references/midjourney-parameters.md

@@ -0,0 +1,393 @@
+# Midjourney V7 Parameter Reference
+
+Complete parameter guide for Midjourney V7 (default model since June 2025).
+
+## Table of Contents
+
+- [Quick Reference Table](#quick-reference-table)
+- [Core Parameters](#core-parameters)
+- [V7-Specific Parameters](#v7-specific-parameters)
+- [Reference System](#reference-system)
+- [V7 vs V6.1 Compatibility](#v7-vs-v61-compatibility)
+- [Parameter Combinations](#parameter-combinations)
+- [Best Practices](#best-practices)
+- [Sources](#sources)
+
+---
+
+## Quick Reference Table
+
+| Parameter | Syntax | Range | Default | Notes |
+|-----------|--------|-------|---------|-------|
+| Version | `--v 7` | - | 7 | Current default |
+| Aspect Ratio | `--ar W:H` | any | 1:1 | No limit |
+| Stylize | `--s` | 0-1000 | 100 | Artistic influence |
+| Chaos | `--c` | 0-100 | 0 | Variation between images |
+| Quality | `--q` | 1, 2, 4 | 1 | Detail level |
+| Seed | `--seed` | 0-4294967295 | random | Reproducibility |
+| Weird | `--w` | 0-3000 | 0 | Unconventional results |
+| Experimental | `--exp` | 0-100 | 0 | Enhanced details (V7 only) |
+| Raw Mode | `--raw` | - | off | Less artistic interpretation |
+| Draft | `--draft` | - | off | 10x faster, half cost (V7 only) |
+| Omni Ref | `--oref` | URL | - | Character/object reference (V7) |
+| Omni Weight | `--ow` | 1-1000 | 100 | Reference strength |
+| Style Ref | `--sref` | URL/code | - | Style reference |
+| Style Weight | `--sw` | 0-1000 | 100 | Style strength |
+| Style Version | `--sv` | 1-6 | 6 | Sref model version |
+| Image Weight | `--iw` | 0-3 | 1 | Image prompt influence |
+| No | `--no` | text | - | Negative prompt |
+| Tile | `--tile` | - | off | Seamless patterns |
+
+---
+
+## Core Parameters
+
+### Aspect Ratio (--ar)
+
+Controls output image dimensions.
+
+```
+--ar 16:9    # Widescreen, cinematic
+--ar 9:16    # Vertical, mobile/portrait
+--ar 1:1     # Square (default)
+--ar 3:2     # Classic photo
+--ar 4:5     # Portrait, Instagram
+--ar 21:9    # Ultra-wide, panoramic
+```
+
+Common use cases:
+- Landscape/cinematic: `16:9`, `21:9`
+- Portrait/character: `9:16`, `4:5`, `2:3`
+- Product/social: `1:1`, `4:5`
+
+### Stylize (--s)
+
+Controls how strongly Midjourney applies its aesthetic training.
+
+| Value | Effect |
+|-------|--------|
+| 0-50 | Very literal, minimal artistic interpretation |
+| 50-100 | Balanced (default: 100) |
+| 100-300 | More artistic, stylized |
+| 300-600 | Highly stylized |
+| 600-1000 | Maximum artistic interpretation |
+
+```
+--s 50     # Closer to prompt, less stylization
+--s 250    # Moderately stylized
+--s 750    # Heavily stylized, artistic
+```
+
+### Chaos (--c)
+
+Controls variation between the 4 generated images.
+
+| Value | Effect |
+|-------|--------|
+| 0 | Very similar results (default) |
+| 1-25 | Subtle variation |
+| 25-50 | Moderate variation |
+| 50-100 | Highly varied results |
+
+```
+--c 0      # Consistent results
+--c 25     # Some variety
+--c 75     # Explore different interpretations
+```
+
+### Quality (--q)
+
+Controls detail level and generation time.
+
+| Value | Effect | GPU Cost |
+|-------|--------|----------|
+| 1 | Standard quality (default) | 1x |
+| 2 | Higher detail | 2x |
+| 4 | Maximum detail (V7 only) | 4x |
+
+```
+--q 1      # Standard, fast
+--q 2      # Enhanced detail
+--q 4      # Maximum fidelity (V7)
+```
+
+### Weird (--w)
+
+Introduces unconventional, unexpected elements.
+
+| Value | Effect |
+|-------|--------|
+| 0 | Normal output (default) |
+| 1-100 | Subtle oddities |
+| 100-500 | Noticeable weirdness |
+| 500-1500 | Very unusual |
+| 1500-3000 | Extremely unconventional |
+
+```
+--w 50     # Slight quirks
+--w 500    # Noticeably strange
+--w 1500   # Very experimental
+```
+
+Note: Keep 5-30 for controlled results. Does not work well with mood boards.
+
+### No (--no)
+
+Negative prompt - excludes specified elements.
+
+```
+--no text, watermark           # Remove text/watermarks
+--no people, humans            # Exclude people
+--no blur, noise               # Avoid artifacts
+--no red                       # Exclude color
+```
+
+Multiple exclusions: comma-separated or multiple --no parameters.
+
+### Seed (--seed)
+
+Controls randomness for reproducibility.
+
+```
+--seed 12345    # Specific seed for consistent results
+```
+
+Use same seed + same prompt = similar output. Useful for:
+- Iterating on a concept
+- Creating variations of a successful image
+- A/B testing parameter changes
+
+---
+
+## V7-Specific Parameters
+
+### Draft Mode (--draft)
+
+10x faster generation at half GPU cost. V7 exclusive.
+
+```
+--draft    # Enable draft mode
+```
+
+Best for:
+- Rapid prototyping
+- Iterative concept development
+- Testing prompt variations
+
+Workflow:
+1. Generate with `--draft`
+2. Iterate 5-10 times
+3. Use "Enhance" on best result for full quality
+
+### Experimental (--exp)
+
+Enhances details and creativity. V7 exclusive.
+
+| Value | Effect |
+|-------|--------|
+| 0 | Off (default) |
+| 5 | Subtle enhancement |
+| 10 | Light detail boost |
+| 25 | Moderate enhancement |
+| 50 | Strong creative boost |
+| 100 | Maximum effect |
+
+```
+--exp 5     # Subtle detail enhancement
+--exp 25    # Noticeable improvement
+--exp 50    # Strong creative effect
+```
+
+Tips:
+- Start low (5-10) when combining with other parameters
+- High values (>25-50) can overwhelm --stylize and --p
+- Avoid for literal prompt interpretation or consistency needs
+
+### Raw Mode (--raw)
+
+Less artistic interpretation, more literal output.
+
+```
+--raw    # Enable raw mode
+```
+
+Note: In V7, use `--raw` instead of `--style raw`.
+
+---
+
+## Reference System
+
+### Style Reference (--sref)
+
+Transfer visual style from reference images.
+
+```
+prompt --sref <image_url>              # Single reference
+prompt --sref <url1> <url2>            # Multiple references
+prompt --sref random                   # Random style code
+prompt --sref 123456789                # Style code
+```
+
+### Style Weight (--sw)
+
+Controls style reference strength.
+
+```
+--sw 50      # Subtle style influence
+--sw 100     # Default
+--sw 500     # Strong style adherence
+--sw 1000    # Maximum style influence
+```
+
+In V7, --sw has more impact with sref codes than with images.
+
+### Style Version (--sv)
+
+Controls which sref model to use.
+
+| Value | Description |
+|-------|-------------|
+| --sv 6 | New V7 default (since June 2025) |
+| --sv 4 | Old V7 sref model (pre-June 2025) |
+| --sv 1-6 | Full range for uploaded images |
+
+```
+--sref random --sv 6    # New V7 style model
+--sref random --sv 4    # Legacy V7 style model
+```
+
+Old sref codes still work; use `--sv 4` for compatibility.
+
+### Omni Reference (--oref)
+
+Character/object reference system. Replaces --cref from V6.
+
+```
+prompt --oref <image_url>              # Reference image
+prompt --oref <url> --ow 200           # With weight
+```
+
+Features:
+- Works with characters, objects, vehicles, creatures
+- Better compatibility with external images
+- Higher consistency for subject preservation
+
+### Omni Weight (--ow)
+
+Controls omni reference strength.
+
+| Value | Effect |
+|-------|--------|
+| 1-100 | Allow style changes, preserve key features |
+| 100-200 | Balanced (default: 100) |
+| 200-400 | Strong fidelity for facial features/details |
+| 400+ | Only with high stylize values |
+
+```
+--ow 50      # Flexible, allows style variation
+--ow 200     # Strong character preservation
+--ow 300     # Maximum fidelity (use carefully)
+```
+
+Tips:
+- Keep under 400 unless using high stylize values
+- High --stylize or --exp requires higher --ow for character preservation
+- Example: `--stylize 1000 --ow 400 --exp 100`
+
+Limitations:
+- Not compatible with: Fast Mode, Draft Mode, --q 4
+- Uses 2x GPU time compared to regular V7 images
+- Not compatible with: Vary Region, Pan, Zoom Out
+
+### Image Weight (--iw)
+
+Controls influence of image prompts.
+
+```
+--iw 0.5    # Less image influence
+--iw 1      # Default balance
+--iw 2      # Strong image influence
+--iw 3      # Maximum image influence
+```
+
+---
+
+## V7 vs V6.1 Compatibility
+
+### Supported in V7
+- All core parameters (--ar, --s, --c, --q, --w, --no)
+- Reference system (--sref, --oref)
+- New features (--draft, --exp)
+
+### Not Supported in V7
+- Multi-Prompting (`::` weight syntax)
+- Stop Parameter (--stop)
+- Character Reference (--cref) - replaced by --oref
+
+### V6.1 Fallback Features
+These use V6.1 rendering even when V7 is selected:
+- Upscalers (Subtle & Creative)
+- Pan
+- Zoom Out
+- Inpainting (Editor)
+
+---
+
+## Parameter Combinations
+
+### Portrait Photography
+```
+cinematic portrait --ar 4:5 --s 150 --q 2
+```
+
+### Landscape/Cinematic
+```
+epic landscape --ar 16:9 --s 200 --q 2
+```
+
+### Character Consistency
+```
+character description --oref <url> --ow 200 --s 100
+```
+
+### Style Transfer
+```
+subject description --sref <style_url> --sw 300 --s 150
+```
+
+### Experimental Art
+```
+abstract concept --exp 50 --w 100 --s 500
+```
+
+### Rapid Iteration
+```
+concept idea --draft --c 50
+```
+
+### Production Quality
+```
+detailed scene --q 4 --s 200 --exp 10
+```
+
+---
+
+## Best Practices
+
+1. Start with defaults, adjust incrementally
+2. Use --draft for exploration, enhance winners
+3. Combine --oref with --sref for character + style consistency
+4. Keep --ow under 400 unless using high --stylize
+5. Use --seed for reproducible iterations
+6. V7 treats entire prompt equally - word order matters less
+7. Short, high-signal phrases work well in V7
+
+---
+
+## Sources
+
+- [Midjourney Parameter List](https://docs.midjourney.com/hc/en-us/articles/32859204029709-Parameter-List)
+- [Midjourney Version Documentation](https://docs.midjourney.com/hc/en-us/articles/32199405667853-Version)
+- [Omni Reference Guide](https://docs.midjourney.com/hc/en-us/articles/36285124473997-Omni-Reference)
+- [Style Reference Documentation](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference)
+- [Draft & Conversational Modes](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes)

package/.claude/skills/midjourney-imagineapi/references/midjourney-style-guide.md

@@ -0,0 +1,296 @@
+# Midjourney V7 Style Guide
+
+Comprehensive guidelines for creating optimized Midjourney V7 prompts.
+
+## Table of Contents
+
+- [V7 Prompt Characteristics](#v7-prompt-characteristics)
+- [5-Layer Prompt Structure](#5-layer-prompt-structure)
+- [Descriptive Vocabulary](#descriptive-vocabulary)
+- [Genre-Specific Patterns](#genre-specific-patterns)
+- [Reference System Usage](#reference-system-usage)
+- [Quality Checklist](#quality-checklist)
+- [Common Mistakes to Avoid](#common-mistakes-to-avoid)
+- [V7-Optimized Tips](#v7-optimized-tips)
+- [Sources](#sources)
+
+---
+
+## V7 Prompt Characteristics
+
+### Key Differences from Previous Versions
+
+1. **Equal Attention**: V7 views the entire prompt with equal weight - word order matters less
+2. **Shorter Prompts Work**: Short, high-signal phrases are effective
+3. **No Multi-Prompting**: The `::` weight syntax is not supported in V7
+4. **Natural Language**: V7's LLM-based interpretation understands creative intent better
+5. **Default Personalization**: V7 tailors outputs to individual aesthetic preferences
+
+### Optimal Prompt Length
+
+- **20-50 words**: Optimal for most prompts
+- **50-75 words**: Complex scenes with multiple elements
+- **Under 20 words**: May lack sufficient detail
+- **Over 75 words**: Diminishing returns
+
+---
+
+## 5-Layer Prompt Structure
+
+### Layer 1: Subject & Composition (Required)
+
+The primary subject and its positioning.
+
+```
+[Subject] [action/pose/position], [composition]
+```
+
+Examples:
+- "woman standing in wheat field, centered composition"
+- "vintage sports car on coastal highway, rule of thirds"
+- "modern skyscraper reaching into clouds, low angle perspective"
+
+### Layer 2: Style & Aesthetic (Required)
+
+Artistic direction and visual treatment.
+
+```
+[style/genre], [mood], [color treatment]
+```
+
+Examples:
+- "cinematic photography, moody and atmospheric, rich earth tones"
+- "editorial fashion photography, elegant and refined, high contrast"
+- "documentary portrait, intimate and raw, desaturated colors"
+
+### Layer 3: Lighting & Environment (Recommended)
+
+Lighting conditions and environmental context.
+
+```
+[lighting type], [time of day], [atmosphere]
+```
+
+Examples:
+- "golden hour sunlight, late afternoon, light fog"
+- "studio softbox lighting, controlled environment, clean backdrop"
+- "dramatic rim lighting, blue hour, scattered clouds"
+
+### Layer 4: Technical Details (Recommended)
+
+Photography specifications and quality markers.
+
+```
+[camera/lens], [depth of field], [quality terms]
+```
+
+Examples:
+- "shot on Canon R5 85mm f/1.4, shallow depth of field, professional photography"
+- "medium format Hasselblad, tack sharp focus, 8k resolution"
+- "35mm film grain, Portra 400 colors, highly detailed"
+
+### Layer 5: Artistic References (Optional)
+
+Stylistic guidance when appropriate.
+
+```
+[artist/photographer], [art movement], [cultural reference]
+```
+
+Examples:
+- "in the style of Annie Leibovitz, contemporary portrait"
+- "inspired by Blade Runner 2049, neo-noir cinematography"
+- "reminiscent of Gregory Crewdson, staged photography"
+
+---
+
+## Descriptive Vocabulary
+
+### Lighting Terms
+- Golden hour, blue hour, twilight, dusk, dawn
+- Rim lighting, backlighting, side lighting, three-point lighting
+- Soft diffused light, hard directional light, volumetric lighting
+- Studio lighting, natural light, candlelight, neon glow
+- God rays, lens flare, atmospheric haze, light shafts
+
+### Composition Terms
+- Rule of thirds, centered composition, symmetrical framing
+- Low angle, high angle, bird's eye view, worm's eye view
+- Dutch angle, over-the-shoulder, close-up, wide shot
+- Negative space, leading lines, framing within frame
+- Depth layering, foreground interest, background separation
+
+### Mood & Atmosphere
+- Cinematic, dramatic, ethereal, dreamlike, surreal
+- Moody, atmospheric, intimate, epic, grandiose
+- Melancholic, nostalgic, futuristic, timeless
+- Serene, tranquil, chaotic, dynamic, peaceful
+- Mysterious, ominous, hopeful, triumphant
+
+### Color & Tone
+- Rich earth tones, muted pastels, vibrant colors, desaturated
+- High contrast, low contrast, tonal depth, color grading
+- Warm tones, cool tones, complementary colors, analogous palette
+- Monochromatic, duotone, technicolor, sepia toned
+- Film emulation: Fuji Velvia, Kodak Portra, Ilford HP5
+
+### Quality Enhancers
+- Highly detailed, ultra realistic, photorealistic, 8k resolution
+- Professional photography, award-winning, museum quality
+- Tack sharp, crisp focus, pin sharp, crystal clear
+- Fine art photography, editorial quality, commercial photography
+- Masterful composition, expert color grading
+
+### Texture & Material
+- Soft fabric, rough concrete, polished metal, weathered wood
+- Smooth glass, organic textures, intricate details
+- Translucent, reflective surface, matte finish, glossy sheen
+- Fine grain, subtle texture, rich tactile quality
+
+### Camera & Technical
+- Camera brands: Canon R5, Sony A7R, Hasselblad, Phase One, RED, ARRI
+- Focal lengths: 24mm, 35mm, 50mm, 85mm, 200mm
+- Apertures: f/1.4, f/2.8, f/4, f/5.6, f/8
+- Film stocks: Kodak Portra 400, Fuji Velvia, Ilford HP5
+- Formats: Medium format, large format, 35mm film, digital cinema
+
+---
+
+## Genre-Specific Patterns
+
+### Portrait Photography
+```
+[Subject description], [framing], cinematic portrait photography,
+[lighting], [mood], shot on [camera] [lens] [aperture],
+[background treatment], [color grading], [quality terms]
+```
+
+### Landscape/Environment
+```
+[Scene description], [time of day], [weather/atmosphere],
+[composition], [lighting quality], shot on [camera] [lens],
+[color treatment], [artistic reference]
+```
+
+### Product Photography
+```
+[Product] [positioning], editorial product photography,
+[lighting setup], [background], shot on [camera] [lens],
+[material emphasis], [quality terms]
+```
+
+### Architectural
+```
+[Building/structure], [perspective], [time/lighting],
+[environment], [mood], shot on [camera] [lens],
+architectural photography, [artistic treatment]
+```
+
+### Fashion Editorial
+```
+[Subject] [outfit/styling], [pose/action], [setting],
+[lighting], high fashion editorial aesthetic, shot on [camera] [lens],
+[artistic reference], [quality terms]
+```
+
+---
+
+## Reference System Usage
+
+### Style Reference (--sref)
+
+For consistent visual style across generations:
+
+```
+prompt text --sref <style_image_url> --sw 200
+```
+
+Use cases:
+- Maintaining brand aesthetic
+- Series with consistent look
+- Matching existing visual style
+
+### Omni Reference (--oref)
+
+For character/subject consistency:
+
+```
+prompt text --oref <character_image_url> --ow 200
+```
+
+Use cases:
+- Character consistency across scenes
+- Product variations
+- Subject in different contexts
+
+### Combined References
+
+For both style and subject consistency:
+
+```
+prompt text --oref <character_url> --ow 150 --sref <style_url> --sw 200
+```
+
+---
+
+## Quality Checklist
+
+Before finalizing a prompt, verify:
+
+- [ ] Main subject clearly defined and positioned
+- [ ] Lighting described with direction and quality
+- [ ] Mood/atmosphere conveyed through descriptive terms
+- [ ] Technical photography details add authenticity
+- [ ] Color treatment or grading specified
+- [ ] Composition guidance included
+- [ ] At least 2-3 quality enhancers present
+- [ ] Appropriate parameters appended
+- [ ] Prompt reads naturally and coherently
+- [ ] No redundant or contradictory terms
+
+---
+
+## Common Mistakes to Avoid
+
+### Too Generic
+- Bad: "beautiful landscape photo"
+- Good: "misty mountain valley at golden hour, dramatic god rays, cinematic landscape photography --ar 16:9 --s 200"
+
+### Contradictory Elements
+- Bad: "bright sunlight, moody dark atmosphere"
+- Good: "overcast diffused light, moody and atmospheric"
+
+### Overstuffed with Keywords
+- Bad: Listing every possible quality term
+- Good: Select 2-3 most relevant quality enhancers
+
+### Missing Technical Anchor
+- Bad: "nice photo of a person"
+- Good: "shot on Canon R5 85mm f/1.4, professional portrait photography"
+
+### Vague Style Reference
+- Bad: "artistic and creative"
+- Good: "inspired by Annie Leibovitz, contemporary editorial portrait"
+
+### Using Deprecated V6 Features
+- Bad: "mountain::2 sky::1" (multi-prompting not supported)
+- Good: "dramatic mountain with expansive sky" (natural language)
+
+---
+
+## V7-Optimized Tips
+
+1. **Trust Natural Language**: V7 understands creative intent - describe what you want naturally
+2. **Front-load Key Elements**: Put the most important subject/concept first
+3. **Use Draft Mode**: Iterate quickly with `--draft`, then enhance winners
+4. **Leverage Personalization**: V7 adapts to your aesthetic preferences over time
+5. **Combine References**: Use both --oref and --sref for maximum consistency
+6. **Start with Default Stylize**: `--s 100` is well-balanced; adjust from there
+7. **Use --exp Sparingly**: Start at 5-10, increase if needed
+
+---
+
+## Sources
+
+- [V7 Alpha Release Notes](https://updates.midjourney.com/v7-alpha/)
+- [V7 Default Model Update](https://updates.midjourney.com/v7-is-now-the-default-model/)