# Nanobanana Prompts > NanoBanana（Google Gemini画像生成）向けの高品質プロンプトを生成。画像生成、編集、フェイススワップ、背景変更、キャラクター一貫性のプロンプト作成を支援。 - Skill: `majiayu000/nanobanana-prompts` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds add majiayu000/nanobanana-prompts` - Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/nanobanana-prompts/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: majiayu000 (https://skillmd.com/u/majiayu000) - Updated: 2026-09-09 - Page: https://skillmd.com/skills/majiayu000/nanobanana-prompts --- # NanoBanana Pro Prompt Engineer Google Gemini NanoBanana / NanoBanana Pro 向けの最適化されたプロンプトを生成するスキル。 **100-200語の簡潔で機能的なプロンプト**を生成します。 ## When to Use This Skill 以下の場合にこのスキルを使用： - 「画像プロンプトを作って」「NanoBananaプロンプト」 - 「顔を入れ替えて」「フェイススワップ」 - 「背景を変更して」「人物を別の場所に」 - 「同じキャラクターで別のシーン」 - 「衣装を統一して」「同じ服装で」 - 「スタイルを適用して」 ## NanoBanana Pro の特徴 - **32,768トークン** の入力に対応（従来モデルの64倍以上） - **最大14枚の参照画像** - Identity Locking対応 - **Markdown/JSON** 形式を理解 - **強力なプロンプト遵守** - 指示通りの画像を生成 - **画像編集** - 複数の同時編集が可能 - **テキストレンダリング** - 多言語テキストを画像内に正確に描画 - **4K高解像度出力** - プロフェッショナル品質 --- ## 4つの黄金ルール ### 1. 編集を優先する 80%完成した画像には新規生成ではなく、具体的な編集指示を与える ``` ❌ "Cool sunset photo" ✅ "Change the lighting to golden hour sunset, add warm orange tones (#FF8C00)" ``` ### 2. 自然言語を使用する キーワード羅列ではなく、完全な文で記述 ``` ❌ "car, neon, city, rain, night" ✅ "A futuristic sports car speeding through a rainy Tokyo street at night, neon signs reflecting on wet asphalt" ``` ### 3. 具体的で詳細に ``` ❌ "a woman" ✅ "A sophisticated elderly woman in her 60s wearing a vintage Chanel-style tweed suit, pearl earrings, silver hair in an elegant updo" ``` ### 4. 文脈・目的を提供する ``` ❌ "sandwich photo" ✅ "Professional food photography of a gourmet sandwich for a Brazilian high-end cookbook, appetizing presentation, natural lighting" ``` --- ## プロンプト構造テンプレート ### 基本構造 ``` [被写体の詳細説明], [構図・アングル], [照明・雰囲気], [スタイル/カメラ設定], [背景], [追加指示] ``` ### 高品質プロンプト例 ``` A professional product photograph of a sleek wireless headphone, shot with Canon EOS 90D DSLR camera, 85mm lens, f/2.8 aperture, soft studio lighting with rim light, pure white background, rule of thirds composition, Pulitzer Prize winning cover photo quality ``` --- ## プロンプトテクニック ### 1. 強調には ALL CAPS を使用 ``` The subject MUST be centered. NEVER include text in the image. ALL elements should follow the rule of thirds. ``` ### 2. Markdown リストで複数指示 ``` Make ALL of the following edits: - Add a red ribbon on the gift box - Change the background to sunset colors (#FF6B35, #F7931E) - Remove the watermark in the corner - Increase the contrast by 20% ``` ### 3. 詳細な色指定（HEXコード併用） ``` The dress should be royal purple (#9F2B68), with gold accents (#FFD700) on the embroidery ``` ### 4. カメラ・構図キーワード | カテゴリ | キーワード例 | |---------|-------------| | **カメラ** | Canon EOS 90D, Sony A7III, Hasselblad X2D | | **レンズ** | 85mm portrait lens, 24mm wide angle, 50mm f/1.4 | | **構図** | rule of thirds, golden ratio, centered composition | | **照明** | soft diffused lighting, rim light, golden hour | | **品質** | Pulitzer Prize winning, award-winning photography | ### 5. JSON形式での詳細指定（上級） ```json { "subject": { "type": "person", "gender": "female", "age": "mid-20s", "hair": "long black hair with subtle waves", "expression": "gentle smile, looking at camera" }, "clothing": { "top": "white silk blouse", "accessories": "pearl necklace" }, "setting": { "location": "modern office with large windows", "lighting": "natural daylight from left side", "time": "late afternoon, golden hour" }, "style": "professional corporate portrait, magazine quality" } ``` --- ## カテゴリ別プロンプトテンプレート ### ポートレート ``` Professional headshot portrait of [subject description], shot with 85mm lens at f/1.8, soft Rembrandt lighting, shallow depth of field with bokeh background, neutral expression, direct eye contact, high-end fashion magazine quality ``` ### 商品写真 ``` E-commerce product photography of [product], pure white background (#FFFFFF), soft diffused lighting from multiple angles, slight shadow underneath for depth, centered composition, ultra high resolution, commercial quality ``` ### 風景 ``` Breathtaking landscape photograph of [scene], shot during golden hour, dramatic clouds in the sky, foreground interest with [element], wide angle 16mm lens, National Geographic cover quality, vibrant colors with natural saturation ``` ### イラスト・アート ``` Digital illustration of [subject], [art style: anime/watercolor/oil painting/vector], [color palette description], detailed line work, professional concept art quality, trending on ArtStation ``` ### 広告・マーケティング ``` High-impact advertising visual for [product/service], bold composition with [主要要素], brand colors: [HEXコード], clean modern aesthetic, persuasive visual hierarchy, award-winning advertising campaign quality ``` --- ## 画像編集プロンプト（Pro Engineer機能） ### フェイススワップ（顔の入れ替え） ``` Using Image 1 as the face reference and Image 2 as the body/scene reference: - PRECISELY preserve the facial features from Image 1 (eyes, nose, mouth shape, skin tone) - Maintain the exact facial proportions and distinctive characteristics - Seamlessly blend the face onto the body in Image 2 - Match the lighting direction and color temperature of Image 2 - Preserve natural skin texture and shadows - ENSURE no distortion or uncanny valley effect Output: A natural-looking composite where the person from Image 1 appears authentically in the scene from Image 2 ``` ### 背景変更（人物を別の場所に） ``` Replace the background while preserving the subject with these specifications: - NEW BACKGROUND: [詳細な背景説明、場所、雰囲気] - Lighting adjustment: Match subject lighting to new environment (light source direction: [方向], color temperature: [warm/cool/neutral]) - Edge blending: Seamless natural transitions, no visible cutout lines - Shadow preservation: Cast appropriate shadows based on new light source - Reflection handling: Add/adjust reflections if scene requires (water, glass, etc.) - Atmospheric integration: Apply consistent haze, color grading to unify scene CRITICAL: Subject MUST appear naturally present in the new location ``` ### 衣装の統一（複数画像で同じ服装） ``` Apply consistent outfit across all provided images: - REFERENCE OUTFIT from Image [N]: [詳細な服装説明] - Maintain EXACT: color (#HEX), fabric texture, fit, style details - Adapt naturally to each pose and lighting condition - Preserve wrinkles and folds appropriate to body position - Keep accessories consistent: [アクセサリー詳細] - Match color saturation and brightness across all outputs ALL images MUST show the identical outfit with natural variations only for pose/lighting ``` ### キャラクター一貫性（Identity Locking） ``` Generate new scene maintaining character identity from reference images: IDENTITY LOCK - Preserve from reference Image(s) [1-14]: - Facial structure: exact proportions, distinctive features - Skin tone and texture - Hair: color, style, length, texture - Body type and proportions - Age appearance - Any unique characteristics: [特徴] NEW SCENE: - Setting: [新しい場所・状況] - Pose: [ポーズ説明] - Expression: [表情] - Lighting: [照明設定] - Camera angle: [アングル] The character MUST be instantly recognizable as the same person from references ``` ### スタイル適用（アートスタイル変換） ``` Transform the image to [スタイル名] style while preserving: - Core composition and subject placement - Recognizable features and identity - Overall mood and atmosphere Style specifications: - Art style: [具体的なスタイル: anime/oil painting/watercolor/pencil sketch/etc.] - Color palette: [色調指定] - Line work: [線の特徴: bold/delicate/sketchy/clean] - Texture: [テクスチャ: smooth/rough/painterly/digital] - Reference artist/work: [参考アーティスト or 作品] MAINTAIN subject recognizability while fully embracing the new artistic style ``` ### 複数編集の例 ``` Edit this image with ALL of the following changes: - Replace the sky with a dramatic sunset (#FF4500 to #800080 gradient) - Add lens flare from the sun position - Enhance the subject's contrast - Remove all distracting elements in the background - Add a subtle vignette effect ``` --- ## 100-200語プロンプト最適化 ### 最適なプロンプト構造（推奨） ``` [主題 30-40語] + [技術仕様 20-30語] + [スタイル/雰囲気 20-30語] + [制約/必須条件 30-50語] ``` ### 例：ポートレート（156語） ``` Create a professional corporate headshot of a confident Asian businesswoman in her early 40s. She has shoulder-length black hair with subtle highlights, wearing a tailored navy blue blazer over a cream silk blouse. Her expression is warm yet authoritative, with a slight smile and direct eye contact. Technical: Shot with Canon EOS R5, 85mm f/1.4 lens, studio lighting setup with key light at 45 degrees, fill light opposite, and hair light from above. Shallow depth of field with creamy bokeh background in neutral gray gradient. Style: Modern executive portrait suitable for LinkedIn and company website. Clean, polished, Fortune 500 CEO aesthetic. MUST include: Catch lights in eyes, natural skin texture, professional color grading. NEVER include: Harsh shadows, overly retouched skin, distracting background elements. Output in 4K resolution, suitable for print and digital. ``` --- ## 品質向上のコツ 1. **具体的に記述** - 「きれいな花」→「鮮やかな赤いバラ、朝露の水滴がついた花びら」 2. **プロ機材を指定** - カメラ、レンズ、照明を具体的に 3. **品質キーワード追加** - "award-winning", "professional", "magazine quality" 4. **否定形も活用** - "NO text", "NEVER include watermarks" 5. **複数指示はリスト形式で** --- ## 使用フロー ``` ユーザー: 「猫の画像のプロンプト作って」 → スキルが以下を提供: 1. 基本プロンプト（すぐ使える版） 2. 詳細プロンプト（カスタマイズ可能版） 3. JSON形式（高精度版） ユーザーは目的に応じて選択し、gemini-image-generator で画像生成 ``` ## 連携スキル - **gemini-image-generator** - 生成したプロンプトで画像を生成 - **copywriting-helper** - 広告用画像のコンセプト立案 --- ## インタラクティブ使用例 ``` ユーザー: 「この写真の顔を別の人に入れ替えて」 → スキルが質問: 1. 顔の参照画像はどれですか？ 2. 体/シーンの画像はどれですか？ 3. 特に保持したい特徴は？ → 最適化されたフェイススワッププロンプトを生成（100-200語） → gemini-image-generator で実行 ``` ``` ユーザー: 「同じキャラクターで別のシーンを作って」 → スキルが質問: 1. 参照画像は何枚ありますか？（最大14枚） 2. 新しいシーンの設定は？ 3. ポーズや表情の希望は？ → Identity Locking プロンプトを生成 → キャラクターの一貫性を保った新シーンを生成 ``` --- ## References - [awesome-nanobanana-pro (GitHub)](https://github.com/ZeroLu/awesome-nanobanana-pro) - [Max Woolf's NanoBanana Prompt Guide](https://minimaxir.com/2025/11/nano-banana-prompts/) - [Nano Banana Prompt Library](https://nanobananaprompt.org/) - [DEV Community - NanoBanana Pro Prompting Guide](https://dev.to/googleai/nano-banana-pro-prompting-guide-strategies-1h9n) - [AIPRM - Nano Banana Pro Prompt Engineer](https://app.aiprm.com/gpts/g-RBjaM21Jr/nano-banana-prompt-engineer-text-to-image)