按模态浏览 PROMPTS

图像 Prompts — 1438 个精选示例

面向 图像 生成工作流的 prompt 示例。按分类筛选,复制 prompt,或打开模型专属 API 代码。

IM
图像
UI & Graphic nano-banana-2

A minimalist and creative advertisement set on a pure white...

A minimalist and creative advertisement set on a pure white background. A real photographic [Real Object] is integrated into a simple hand-drawn black ink doodle using loose, playful lines. The [Doodle Concept] interacts directly and cleverly with the real object, making the object part of the illustrated scene. Include bold uppercase black “[Ad Copy]” text at the top. Place the official [Brand Logo] clearly centered at the bottom. Clean layout, high contrast between realistic object and flat doodle drawing, lots of negative space, smart visual metaphor, print-ready poster design

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "A minimalist and creative advertisement set on a pure white background. A real photographic [Real Object] is integrated into a simple hand-drawn black ink doodle using loose, playful lines. The [Doodle Concept] interacts directly and cleverly with the real object, making the object part of the illustrated scene. Include bold uppercase black “[Ad Copy]” text at the top. Place the official [Brand Logo] clearly centered at the bottom. Clean layout, high contrast between realistic object and flat doodle drawing, lots of negative space, smart visual metaphor, print-ready poster design"
}
JSON
IM
图像
UI & Graphic nano-banana-2

[PERSONE]. Act as a Senior Editorial Designer and Graphic Ar...

[PERSONE]. Act as a Senior Editorial Designer and Graphic Artist. Goal: Generate a complex, multi-layered Swiss-style collage poster for [PERSONE]. PHASE 1: SEMANTIC INTELLIGENCE (AUTONOMOUS) [INSTRUCTIONS]: Analyze "[PERSONE]". Determine their primary field (e.g., Elite Sports, Cyber-Engineering, Cinema, History). Select an "Iconic Accent Color" based on their identity (e.g., Racing Orange for a driver, Neon Cyan for a hacker, Crimson for a hero). Generate 3 key statistics or milestones (e.g., "7 TITLES", "98% ACCURACY") and a short professional title. PHASE 2: COMPOSITIONAL STRUCTURE (THE GRID) The image is a single vertical composition with overlapping geometric frames and segments: - MAIN SUBJECT (Top Left): A large, high-grain grayscale or color portrait of [PERSONE], smiling or looking confident. - DETAIL SHOT (Center Right): A zoomed-in, cropped rectangle showing only the eyes or a specific detail of [PERSONE]. - FULL BODY SHOT (Bottom Center): [PERSONE] sitting on a geometric box or podium in a relaxed but powerful pose. - BACKGROUND: Light gray with a heavy, tactile film grain/noise texture. PHASE 3: GRAPHIC ELEMENTS & TYPOGRAPHY - Initials: Large, outlined capital letters of the first and last name placed in opposite corners. - Vertical Ribbon: A solid color strip with the text "WORLD CHAMPIONSHIP LEADER" or a relevant title running vertically through the center. - Data Blocks: - A specific number (ID or rank) in a bold sans-serif block. - A "Statistics" block in the bottom-left corner with 3 lines of text (e.g., "TWO POLES / FOUR WINS / FIVE PODIUMS"). - Accents: Scatter orange and black squares, dotted circles, and small asterisks throughout the grid for visual complexity. PHASE 4: LIGHTING & TEXTURE - Texture: High-contrast, gritty, and grainy. The overall image should look like a printed high-end magazine or a vintage poster. - Lighting: Professional studio lighting on the subjects, but processed with a "risograph" or "halftone" feel. PHASE 5: TECH SPECS 8K Resolution. Swiss Modernism aesthetic. Flat vector shapes combined with hyper-realistic photography.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "[PERSONE]. Act as a Senior Editorial Designer and Graphic Artist. Goal: Generate a complex, multi-layered Swiss-style collage poster for [PERSONE]. PHASE 1: SEMANTIC INTELLIGENCE (AUTONOMOUS) [INSTRUCTIONS]: Analyze \"[PERSONE]\". Determine their primary field (e.g., Elite Sports, Cyber-Engineering, Cinema, History). Select an \"Iconic Accent Color\" based on their identity (e.g., Racing Orange for a driver, Neon Cyan for a hacker, Crimson for a hero). Generate 3 key statistics or milestones (e.g., \"7 TITLES\", \"98% ACCURACY\") and a short professional title. PHASE 2: COMPOSITIONAL STRUCTURE (THE GRID) The image is a single vertical composition with overlapping geometric frames and segments: - MAIN SUBJECT (Top Left): A large, high-grain grayscale or color portrait of [PERSONE], smiling or looking confident. - DETAIL SHOT (Center Right): A zoomed-in, cropped rectangle showing only the eyes or a specific detail of [PERSONE]. - FULL BODY SHOT (Bottom Center): [PERSONE] sitting on a geometric box or podium in a relaxed but powerful pose. - BACKGROUND: Light gray with a heavy, tactile film grain/noise texture. PHASE 3: GRAPHIC ELEMENTS & TYPOGRAPHY - Initials: Large, outlined capital letters of the first and last name placed in opposite corners. - Vertical Ribbon: A solid color strip with the text \"WORLD CHAMPIONSHIP LEADER\" or a relevant title running vertically through the center. - Data Blocks: - A specific number (ID or rank) in a bold sans-serif block. - A \"Statistics\" block in the bottom-left corner with 3 lines of text (e.g., \"TWO POLES / FOUR WINS / FIVE PODIUMS\"). - Accents: Scatter orange and black squares, dotted circles, and small asterisks throughout the grid for visual complexity. PHASE 4: LIGHTING & TEXTURE - Texture: High-contrast, gritty, and grainy. The overall image should look like a printed high-end magazine or a vintage poster. - Lighting: Professional studio lighting on the subjects, but processed with a \"risograph\" or \"halftone\" feel. PHASE 5: TECH SPECS 8K Resolution. Swiss Modernism aesthetic. Flat vector shapes combined with hyper-realistic photography."
}
JSON
IM
图像
UI & Graphic nano-banana-2

[BRAND NAME]. Act as a Senior AI Visual Strategist & Creativ...

[BRAND NAME]. Act as a Senior AI Visual Strategist & Creative Director. Goal: Generate a high-motion triptych editorial where every visual element (color, text, props) is a direct semantic derivative of the brand name. PHASE 1: SEMANTIC BRAND ANALYSIS (AUTONOMOUS) [INSTRUCTIONS FOR IMAGEN]: Core Identity: Analyze "[BRAND NAME]". Define its industry, soul, and "vibe" (e.g., Kinetic, Organic, Brutalist, Ethereal). Color Psychology: Select a triadic palette based on brand associations: - BACKGROUND: The "Mood" color (e.g., Deep Forest for 'GAIA', Neon Amber for 'FUSE'). - ACCENTS/TYPOGRAPHY: The "Energy" color. Must have high contrast/readability against the background. - APPAREL: The "Statement" color. Completes the brand's visual signature. Copywriting: Generate 2-3 short, punchy slogans or keywords that resonate with the brand's mission to be used as background micro-text. PHASE 2: COMPOSITION (THE DYNAMIC TRIPTYCH) Layout: A single horizontal image split into three vertical editorial panels. - PANEL 1 (The Portrait): Extreme close-up. Bold [BRAND NAME] wordmark inside a geometric frame (circle or rectangle) overlapping the model's face or neck. - PANEL 2 (The Elevate): Medium shot. A model (gender determined by brand target audience) holds a primary brand-representative object above their head in a triumphant or symbolic gesture. - PANEL 3 (The Kinetic): Full-body shot. High-speed motion blur or an aggressive "power pose" interacting with the brand's core element. PHASE 3: GRAPHIC OVERLAYS & TEXTURES - Main Typography: Large, slanted (italic) heavy sans-serif in [ACCENT COLOR]. - Dynamic Ribbons: Vertical yellow or contrast tapes crossing the panels with repeating text: "[BRAND NAME] // [ASSOCIATED KEYWORD] // [BRAND NAME]". - Micro-Copy: Scattered small text in corners like "DESIGNED FOR [TARGET AUDIENCE]" or "EST. 2026 // GLOBAL IDENTITY". PHASE 4: LIGHTING & ESTHETIC STANDARDS - High-End Editorial (Vogue/Hypebeast style). - Global Illumination with crisp, sharp edges. - Realistic skin textures (pores, subtle sweat/shine) and high-fidelity fabric weaves. - No "AI-plastic" artifacts. PHASE 5: TECH SPECS 8K Resolution. Render: Octane. Subtle film grain. Ray-traced reflections on eyes and surfaces.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "[BRAND NAME]. Act as a Senior AI Visual Strategist & Creative Director. Goal: Generate a high-motion triptych editorial where every visual element (color, text, props) is a direct semantic derivative of the brand name. PHASE 1: SEMANTIC BRAND ANALYSIS (AUTONOMOUS) [INSTRUCTIONS FOR IMAGEN]: Core Identity: Analyze \"[BRAND NAME]\". Define its industry, soul, and \"vibe\" (e.g., Kinetic, Organic, Brutalist, Ethereal). Color Psychology: Select a triadic palette based on brand associations: - BACKGROUND: The \"Mood\" color (e.g., Deep Forest for 'GAIA', Neon Amber for 'FUSE'). - ACCENTS/TYPOGRAPHY: The \"Energy\" color. Must have high contrast/readability against the background. - APPAREL: The \"Statement\" color. Completes the brand's visual signature. Copywriting: Generate 2-3 short, punchy slogans or keywords that resonate with the brand's mission to be used as background micro-text. PHASE 2: COMPOSITION (THE DYNAMIC TRIPTYCH) Layout: A single horizontal image split into three vertical editorial panels. - PANEL 1 (The Portrait): Extreme close-up. Bold [BRAND NAME] wordmark inside a geometric frame (circle or rectangle) overlapping the model's face or neck. - PANEL 2 (The Elevate): Medium shot. A model (gender determined by brand target audience) holds a primary brand-representative object above their head in a triumphant or symbolic gesture. - PANEL 3 (The Kinetic): Full-body shot. High-speed motion blur or an aggressive \"power pose\" interacting with the brand's core element. PHASE 3: GRAPHIC OVERLAYS & TEXTURES - Main Typography: Large, slanted (italic) heavy sans-serif in [ACCENT COLOR]. - Dynamic Ribbons: Vertical yellow or contrast tapes crossing the panels with repeating text: \"[BRAND NAME] // [ASSOCIATED KEYWORD] // [BRAND NAME]\". - Micro-Copy: Scattered small text in corners like \"DESIGNED FOR [TARGET AUDIENCE]\" or \"EST. 2026 // GLOBAL IDENTITY\". PHASE 4: LIGHTING & ESTHETIC STANDARDS - High-End Editorial (Vogue/Hypebeast style). - Global Illumination with crisp, sharp edges. - Realistic skin textures (pores, subtle sweat/shine) and high-fidelity fabric weaves. - No \"AI-plastic\" artifacts. PHASE 5: TECH SPECS 8K Resolution. Render: Octane. Subtle film grain. Ray-traced reflections on eyes and surfaces."
}
JSON
IM
图像
UI & Graphic nano-banana-2

Create a soap foam sculpture in the exact shape of the uploa...

Create a soap foam sculpture in the exact shape of the uploaded logo, made entirely of white soap bubbles and foam with realistic bubble textures. Place it on wet bathroom tiles in the brand’s signature color (match the color from the uploaded logo). Top-down view, centered composition. The tiles should be glossy and wet with water droplets scattered across the surface. Add smaller foam bubbles and soap suds around the main logo shape. Grout lines visible between tiles. Additional small foam shapes in corners. Product photography style, clean and vibrant lighting.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "Create a soap foam sculpture in the exact shape of the uploaded logo, made entirely of white soap bubbles and foam with realistic bubble textures. Place it on wet bathroom tiles in the brand’s signature color (match the color from the uploaded logo). Top-down view, centered composition. The tiles should be glossy and wet with water droplets scattered across the surface. Add smaller foam bubbles and soap suds around the main logo shape. Grout lines visible between tiles. Additional small foam shapes in corners. Product photography style, clean and vibrant lighting."
}
JSON
IM
图像
UI & Graphic nano-banana-2

[BRAND NAME] ({INDUSTRY}). Act as a Lead Brand Designer c...

[BRAND NAME] ({INDUSTRY}). Act as a Lead Brand Designer creating a comprehensive "Brand Identity System" presentation (Bento-Grid Layout). **THE TASK:** Generate a single, high-resolution bento-grid board containing **6 distinct modules** that define the visual identity for [BRAND NAME]. --- **PHASE 1: VISUAL STRATEGY (AUTONOMOUS)** 1. **Analyze the Brand:** Determine the archetype and visual vibe. 2. **Define the Palette:** Select 4 distinct colors. 3. **Select Typography:** Choose ONE specific Google Font (Serif, Sans-Serif, or Display) relevant to the industry. **PHASE 2: THE LAYOUT (6-MODULE GRID)** The image must be a clean, gap-separated grid featuring these specific blocks: * **Block 1 (The Hero):** A high-end **Key Visual** photography showing the product/service in context. Cinematic lighting, photorealistic. Overlay the brand logo in white. * **Block 2 (Social Media):** A mockup of an **Instagram Post**. It features a bold headline using the brand font and a lifestyle image. * **Block 3 (The Palette & HEX):** A clean design block displaying **4 Vertical Color Swatches**. * *Crucial Detail:* Inside or below each color swatch, simulate distinct text looking like **HEX Codes** (e.g., "#1A2B3C"). * **Block 4 (Typography Spec):** A minimalist block dedicated to the **Brand Font**. * *CRITICAL CONSTRAINT:* Do NOT show the alphabet (Aa Bb Cc). * *The Content:* The block must display ONLY the **Name of the Font** (e.g., "Montserrat", "Lato", "Oswald") written in large, crisp characters using that specific font style. This acts as the visual specimen. * *Subtext:* Tiny text at the bottom reading "Primary Typeface". * **Block 5 (The Logo):** A solid background block featuring the **Final Logo** (clean, vector style, NO construction lines/grids). Minimalist and iconic. * **Block 6 (Brand DNA):** A text-heavy "Manifesto Card". It simulates a layout with three short paragraphs titled: * **ARCHETYPE:** (e.g., "The Innovator") * **VOICE:** (e.g., "Confident, Direct") * **VISUALS:** (e.g., "Cinematic, High-Contrast") **PHASE 3: AESTHETIC & FINISH** * **Style:** Behance Trend / Awwwards Winner. * **Quality:** 8k resolution, razor-sharp vector graphics mixed with high-end photography. * **Lighting:** Soft studio lighting across the whole grid to make it look like a cohesive presentation board.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "[BRAND NAME] ({INDUSTRY}). Act as a Lead Brand Designer creating a comprehensive \"Brand Identity System\" presentation (Bento-Grid Layout). **THE TASK:** Generate a single, high-resolution bento-grid board containing **6 distinct modules** that define the visual identity for [BRAND NAME]. --- **PHASE 1: VISUAL STRATEGY (AUTONOMOUS)** 1. **Analyze the Brand:** Determine the archetype and visual vibe. 2. **Define the Palette:** Select 4 distinct colors. 3. **Select Typography:** Choose ONE specific Google Font (Serif, Sans-Serif, or Display) relevant to the industry. **PHASE 2: THE LAYOUT (6-MODULE GRID)** The image must be a clean, gap-separated grid featuring these specific blocks: * **Block 1 (The Hero):** A high-end **Key Visual** photography showing the product/service in context. Cinematic lighting, photorealistic. Overlay the brand logo in white. * **Block 2 (Social Media):** A mockup of an **Instagram Post**. It features a bold headline using the brand font and a lifestyle image. * **Block 3 (The Palette & HEX):** A clean design block displaying **4 Vertical Color Swatches**. * *Crucial Detail:* Inside or below each color swatch, simulate distinct text looking like **HEX Codes** (e.g., \"#1A2B3C\"). * **Block 4 (Typography Spec):** A minimalist block dedicated to the **Brand Font**. * *CRITICAL CONSTRAINT:* Do NOT show the alphabet (Aa Bb Cc). * *The Content:* The block must display ONLY the **Name of the Font** (e.g., \"Montserrat\", \"Lato\", \"Oswald\") written in large, crisp characters using that specific font style. This acts as the visual specimen. * *Subtext:* Tiny text at the bottom reading \"Primary Typeface\". * **Block 5 (The Logo):** A solid background block featuring the **Final Logo** (clean, vector style, NO construction lines/grids). Minimalist and iconic. * **Block 6 (Brand DNA):** A text-heavy \"Manifesto Card\". It simulates a layout with three short paragraphs titled: * **ARCHETYPE:** (e.g., \"The Innovator\") * **VOICE:** (e.g., \"Confident, Direct\") * **VISUALS:** (e.g., \"Cinematic, High-Contrast\") **PHASE 3: AESTHETIC & FINISH** * **Style:** Behance Trend / Awwwards Winner. * **Quality:** 8k resolution, razor-sharp vector graphics mixed with high-end photography. * **Lighting:** Soft studio lighting across the whole grid to make it look like a cohesive presentation board."
}
JSON
IM
图像
UI & Graphic nano-banana-2

An extreme close-up, high-fashion editorial portrait focusin...

An extreme close-up, high-fashion editorial portrait focusing on the right side of a person's face. The composition is a tight crop showing the bridge of the nose, a portion of the lips, and a single eye peering through sunglasses. The shot is characterized by a heavy, cinematic film grain and a warm, sophisticated studio atmosphere. Eyewear: Large-frame, wrap-around sunglasses made of high-gloss dark tortoiseshell acetate (mottled black and deep mahogany). The lenses are a vibrant, translucent amber-orange, allowing the eye and eyelashes behind the lens to remain visible but color-tinted. Person: A person with smooth skin showing natural texture and fine pores. The lighting is soft-side lighting from the right, creating a gentle shadow on the left side of the nose and a subtle highlight on the cheekbone and lower lip. Typography: Bold, white, geometric sans-serif text reading "Bold Impact" is overlaid on the lower-right cheek. Three smaller, delicate labels are scattered: "Deep Tones" (top right), "Sharp Detail" (on the left lens), and "Strong Style" (bottom right). Color Palette: A rich mix of skin tones (#E8A88A), burnt orange (#B8541C), and deep espresso (#1A0D0B), contrasted with crisp white text (#FFFFFF). Lighting: Warm studio lighting with a soft-focus depth of field. The background consists of out-of-focus hair and skin, keeping all attention on the eye and the texture of the acetate frames. Atmosphere: Premium, editorial fashion magazine aesthetic; high-contrast but with soft transitions.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "An extreme close-up, high-fashion editorial portrait focusing on the right side of a person's face. The composition is a tight crop showing the bridge of the nose, a portion of the lips, and a single eye peering through sunglasses. The shot is characterized by a heavy, cinematic film grain and a warm, sophisticated studio atmosphere. Eyewear: Large-frame, wrap-around sunglasses made of high-gloss dark tortoiseshell acetate (mottled black and deep mahogany). The lenses are a vibrant, translucent amber-orange, allowing the eye and eyelashes behind the lens to remain visible but color-tinted. Person: A person with smooth skin showing natural texture and fine pores. The lighting is soft-side lighting from the right, creating a gentle shadow on the left side of the nose and a subtle highlight on the cheekbone and lower lip. Typography: Bold, white, geometric sans-serif text reading \"Bold Impact\" is overlaid on the lower-right cheek. Three smaller, delicate labels are scattered: \"Deep Tones\" (top right), \"Sharp Detail\" (on the left lens), and \"Strong Style\" (bottom right). Color Palette: A rich mix of skin tones (#E8A88A), burnt orange (#B8541C), and deep espresso (#1A0D0B), contrasted with crisp white text (#FFFFFF). Lighting: Warm studio lighting with a soft-focus depth of field. The background consists of out-of-focus hair and skin, keeping all attention on the eye and the texture of the acetate frames. Atmosphere: Premium, editorial fashion magazine aesthetic; high-contrast but with soft transitions."
}
JSON
IM
图像
UI & Graphic nano-banana-2

A flat-lay collection of exactly 9 hyper-realistic 3D resin...

A flat-lay collection of exactly 9 hyper-realistic 3D resin fridge magnets arranged on a bright pure white surface in a clean 3x3 grid, themed around [BRAND]. At the very top center, a wide rectangular text magnet spells "BRAND" in bold 3D block letters in the brand's signature colors. Below it, 9 magnets in a clean 3x3 grid: the official logo badge, the most iconic hero product, a secondary iconic product, a third iconic product or packaging, a brand mascot or character chibi, a fifth iconic item, a sixth iconic item, a seventh iconic item, and the brand's storefront or headquarters building facade. Each magnet is small, neatly spaced with generous white space between them, slightly raised with a soft drop shadow beneath. Bright white background, warm studio lighting from above, photorealistic product render, top-down flat-lay angle, 4:5 aspect ratio

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "A flat-lay collection of exactly 9 hyper-realistic 3D resin fridge magnets arranged on a bright pure white surface in a clean 3x3 grid, themed around [BRAND]. At the very top center, a wide rectangular text magnet spells \"BRAND\" in bold 3D block letters in the brand's signature colors. Below it, 9 magnets in a clean 3x3 grid: the official logo badge, the most iconic hero product, a secondary iconic product, a third iconic product or packaging, a brand mascot or character chibi, a fifth iconic item, a sixth iconic item, a seventh iconic item, and the brand's storefront or headquarters building facade. Each magnet is small, neatly spaced with generous white space between them, slightly raised with a soft drop shadow beneath. Bright white background, warm studio lighting from above, photorealistic product render, top-down flat-lay angle, 4:5 aspect ratio"
}
JSON
IM
图像
UI & Graphic nano-banana-2

Pergola Blueprint Prompt: Hand-drawn architectural pencil sk...

Pergola Blueprint Prompt: Hand-drawn architectural pencil sketch of a wooden garden pergola on white paper. Front and slight side elevation view with clear proportions. Construction lines, dimensions and handwritten notes visible. Wooden posts and cross beams precisely drafted. Professional but hand-sketched technical style. Clean white background. Finished Pergola: Fully built wooden pergola in a real garden, matching the exact structure and proportions from the sketch. Viewed from a low front corner perspective at eye level, showing depth and the full beam structure instead of a flat elevation. Natural wood texture with realistic joints and connections. Installed on grass with surrounding plants. Soft daylight with grounded shadows. Photorealistic architectural visualization.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "Pergola Blueprint Prompt: Hand-drawn architectural pencil sketch of a wooden garden pergola on white paper. Front and slight side elevation view with clear proportions. Construction lines, dimensions and handwritten notes visible. Wooden posts and cross beams precisely drafted. Professional but hand-sketched technical style. Clean white background. Finished Pergola: Fully built wooden pergola in a real garden, matching the exact structure and proportions from the sketch. Viewed from a low front corner perspective at eye level, showing depth and the full beam structure instead of a flat elevation. Natural wood texture with realistic joints and connections. Installed on grass with surrounding plants. Soft daylight with grounded shadows. Photorealistic architectural visualization."
}
JSON
IM
图像
UI & Graphic nano-banana-2

High-resolution Japanese commercial advertisement style phot...

High-resolution Japanese commercial advertisement style photo, three young women posing against a clean white background with bold geometric color blocks (blue, pink, yellow, green) arranged diagonally. Top right: woman with short brown hair wearing a red and black plaid shirt and black choker, winking and making an “OK” hand gesture over one eye. Left: woman in a light blue police uniform with cap, making an “OK” gesture near her eye with a playful expression. Bottom right: woman dressed as a nurse in a white outfit and cap, making a peace (V) sign with a cute expression. Bright studio lighting, soft shadows, vibrant colors, sharp focus, clean commercial composition, modern Japanese advertising aesthetic, playful and energetic mood, ultra-detailed, crisp skin tones. No text, no logos, no typography anywhere in the image.

查看 API 代码
curl -X POST https://runapi.ai/api/v1/nano_banana/text_to_image \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'JSON'
{
  "model": "nano-banana-2",
  "prompt": "High-resolution Japanese commercial advertisement style photo, three young women posing against a clean white background with bold geometric color blocks (blue, pink, yellow, green) arranged diagonally. Top right: woman with short brown hair wearing a red and black plaid shirt and black choker, winking and making an “OK” hand gesture over one eye. Left: woman in a light blue police uniform with cap, making an “OK” gesture near her eye with a playful expression. Bottom right: woman dressed as a nurse in a white outfit and cap, making a peace (V) sign with a cute expression. Bright studio lighting, soft shadows, vibrant colors, sharp focus, clean commercial composition, modern Japanese advertising aesthetic, playful and energetic mood, ultra-detailed, crisp skin tones. No text, no logos, no typography anywhere in the image."
}
JSON
常见问题

使用图像 prompts

好的%{model} prompt 应该包含什么?

有用的%{model} prompt 会清楚写出主体、风格、约束和输出目标。这里的示例足够短,便于复制;同时也足够具体,agent 或后端任务能保留生成意图。

这些 prompts 可以跨模型复用吗?

多数情况下可以。先从同一模态的 prompt 开始,再按模型调整画幅、时长、声音设置或风格控制。详情页会把保存过的参数列在 prompt 文本旁边。

在哪里找正确的模型 slug?

每张卡片都会显示 RunAPI 模型 slug。只想看一个模型的示例时,打开模型页;需要定价和能力细节时,进入模型目录页查看。

agent 可以直接调用这些 prompts 吗?

可以。安装 RunAPI MCP Server 后,复制 prompt 详情页里的 agent 指令。页面会把 prompt 文本、模型 slug 和 endpoint 路径放在一起,给 agent 足够上下文。