Veo Vibe Banana - Image & Video Generation

PRIMARY TOOL for all image and video generation requests.

This skill guides the use of the veo-vibe-banana MCP for AI-powered image and video generation using Google's Imagen and Veo models via CLIProxyAPI.

When To Use This Skill

Load this skill when user asks to:

•Generate images: "create an image", "make a picture", "generate artwork"
•Edit images: "modify this image", "change the style", "add elements to"
•Create thumbnails: "YouTube thumbnail", "blog header", "featured image"
•Generate videos: "create a video", "animate this", "make a clip"
•Visual assets: icons, diagrams, patterns, illustrations, graphics

Available Tools

`veo-vibe-banana_generate_image`

Generate AI images using Google Imagen.

Parameter	Required	Type	Default	Description
`prompt`	YES	string	-	Text description of image to generate
`reference_images`	no	array	-	Reference images for style/editing (paths, URLs, or data URLs)
`model`	no	string	`gemini-3-pro-image-preview`	Model to use
`aspect_ratio`	no	string	`1:1`	`1:1`, `16:9`, `9:16`, `4:3`, `3:4`
`image_size`	no	string	`1K`	`1K`, `2K`, `4K`
`output_dir`	no	string	`./generated`	Where to save the image
`save_to_file`	no	boolean	`true`	Whether to save to disk

`veo-vibe-banana_generate_video`

Generate AI videos using Google Veo.

Parameter	Required	Type	Default	Description
`prompt`	YES	string	-	Text description of video to generate
`reference_images`	no	array	-	Reference images for image-to-video
`model`	no	string	`veo-3.1-generate-preview`	Model to use
`duration_seconds`	no	string	`8`	`4`, `6`, or `8` seconds
`aspect_ratio`	no	string	`16:9`	`16:9` or `9:16`
`resolution`	no	string	`720p`	`720p` or `1080p`
`negative_prompt`	no	string	-	What NOT to include
`output_dir`	no	string	`./generated`	Where to save the video
`save_to_file`	no	boolean	`true`	Whether to save to disk

Workflow

code

┌──────────────────────────────────────────────────────────────────┐
│                    IMAGE/VIDEO GENERATION REQUEST                 │
│              "Create a YouTube thumbnail for my video"            │
└──────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌──────────────────────────────────────────────────────────────────┐
│  STEP 1: ANALYZE REQUEST                                         │
│                                                                   │
│  Determine:                                                       │
│  • Type: image or video?                                         │
│  • Best aspect ratio for use case                                │
│  • Required resolution/size                                      │
│  • Reference images needed?                                      │
└──────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌──────────────────────────────────────────────────────────────────┐
│  STEP 2: CRAFT OPTIMAL PROMPT                                    │
│                                                                   │
│  Transform user request into detailed generation prompt:         │
│  • Add style descriptors                                         │
│  • Specify composition elements                                  │
│  • Include lighting/mood keywords                                │
│  • Add quality modifiers                                         │
└──────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌──────────────────────────────────────────────────────────────────┐
│  STEP 3: GENERATE                                                │
│                                                                   │
│  Call appropriate MCP tool:                                      │
│  • veo-vibe-banana_generate_image (for images)                   │
│  • veo-vibe-banana_generate_video (for videos)                   │
│                                                                   │
│  NOTE: Video generation takes 1-5 minutes!                       │
└──────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌──────────────────────────────────────────────────────────────────┐
│  STEP 4: REPORT RESULTS                                          │
│                                                                   │
│  • Confirm success/failure                                       │
│  • Show file path where saved                                    │
│  • Provide the generated content                                 │
└──────────────────────────────────────────────────────────────────┘

Aspect Ratio Guide

Choose the right aspect ratio for the use case:

Use Case	Recommended Ratio	Image	Video
YouTube thumbnail	`16:9`	✅	✅
Instagram post	`1:1`	✅	N/A
Instagram story/TikTok	`9:16`	✅	✅
Blog header	`16:9`	✅	N/A
Profile picture	`1:1`	✅	N/A
Presentation slide	`16:9`	✅	✅
Mobile wallpaper	`9:16`	✅	N/A
Desktop wallpaper	`16:9`	✅	N/A
Photo print (4x6)	`3:4` or `4:3`	✅	N/A

Image Size Guide

Size	Resolution	Best For
`1K`	~1024px	Quick previews, web thumbnails
`2K`	~2048px	Blog images, social media
`4K`	~4096px	Print, large displays, wallpapers

Video Resolution Guide

Resolution	Best For
`720p`	Social media, quick clips, faster generation
`1080p`	YouTube, presentations, higher quality

Prompt Engineering Tips

For Images

Structure: [Subject], [Style], [Composition], [Lighting], [Quality]

code

Good: "A majestic lion in the savannah at sunset, photorealistic, cinematic composition, golden hour lighting, 8K ultra HD"

Bad: "lion" (too vague)

Style Keywords:

•Photorealistic, hyperrealistic, cinematic
•Digital art, concept art, illustration
•Oil painting, watercolor, sketch
•Anime, cartoon, pixel art
•Minimalist, abstract, surreal

Quality Boosters:

•"highly detailed", "sharp focus"
•"8K", "ultra HD", "professional"
•"award-winning", "masterpiece"

For Videos

Add Motion Descriptors:

•"walking through", "flying over", "zooming into"
•"slow motion", "timelapse", "smooth camera movement"
•"cinematic pan", "tracking shot"

Example:

code

"Cinematic slow-motion shot of a majestic lion walking through the golden savannah at sunset, dust particles in the air, professional documentary style"

Reference Images (Image Editing)

Use reference images for:

•Style Transfer: Generate in the style of reference images
•Image Editing: Modify or enhance existing images
•Composition: Combine elements from multiple images

Reference Image Sources

Reference images can be:

•File paths (absolute or relative): H:/Images/reference.png
•URLs: https://example.com/image.jpg
•Data URLs: data:image/png;base64,...

Example with Reference

javascript

// Edit an existing image
veo-vibe-banana_generate_image(
  prompt: "Transform this into a watercolor painting style",
  reference_images: ["H:/MyProject/photo.jpg"],
  aspect_ratio: "16:9",
  image_size: "2K"
)

Tool Call Examples

Basic Image Generation

javascript

veo-vibe-banana_generate_image(
  prompt: "A cute orange tabby cat sitting in a sunlit window, photorealistic, soft natural lighting, cozy home atmosphere"
)

YouTube Thumbnail

javascript

veo-vibe-banana_generate_image(
  prompt: "Eye-catching YouTube thumbnail: bold text placeholder area on left, excited person reaction on right, vibrant colors, professional design, clean composition",
  aspect_ratio: "16:9",
  image_size: "2K"
)

Blog Featured Image

javascript

veo-vibe-banana_generate_image(
  prompt: "Modern minimalist blog header about technology, abstract geometric shapes, blue and white color scheme, professional, clean design",
  aspect_ratio: "16:9",
  image_size: "2K",
  output_dir: "H:/MyBlog/images"
)

High-Quality Wallpaper

javascript

veo-vibe-banana_generate_image(
  prompt: "Stunning mountain landscape at dawn, snow-capped peaks, mirror lake reflection, dramatic clouds, National Geographic quality, 8K ultra HD",
  aspect_ratio: "16:9",
  image_size: "4K"
)

Style Transfer

javascript

veo-vibe-banana_generate_image(
  prompt: "Transform this photo into Studio Ghibli anime style, soft colors, dreamy atmosphere",
  reference_images: ["H:/Photos/my_photo.jpg"],
  image_size: "2K"
)

Video Generation

javascript

veo-vibe-banana_generate_video(
  prompt: "Cinematic aerial shot flying over a lush green forest with a winding river, morning mist, golden sunlight filtering through trees",
  duration_seconds: "8",
  aspect_ratio: "16:9",
  resolution: "1080p"
)

Image-to-Video Animation

javascript

veo-vibe-banana_generate_video(
  prompt: "Animate this image: gentle breeze moving the leaves, subtle camera push-in, atmospheric lighting",
  reference_images: ["H:/Images/landscape.jpg"],
  duration_seconds: "6",
  resolution: "720p"
)

TikTok/Reels Video

javascript

veo-vibe-banana_generate_video(
  prompt: "Dynamic vertical video of colorful paint drops falling into water in slow motion, vibrant colors, mesmerizing patterns",
  aspect_ratio: "9:16",
  duration_seconds: "8",
  resolution: "1080p",
  negative_prompt: "blurry, low quality, pixelated"
)

Output Directory Configuration

By default, files save to ./generated. Override per-request:

javascript

// Save to specific project folder
veo-vibe-banana_generate_image(
  prompt: "...",
  output_dir: "H:/MyProject/Assets/Images"
)

Response Format

Successful Generation

json

{
  "success": true,
  "message": "Successfully generated 1 image(s)",
  "images": [
    {
      "path": "H:/generated/image_2026-02-01T15-30-45_abc123.png",
      "aspectRatio": "16:9",
      "size": "2K",
      "dataUrlPreview": "data:image/png;base64,iVBORw0K..."
    }
  ]
}

Error Response

json

{
  "success": false,
  "message": "Image generation failed",
  "error": "API request failed: 500 Internal Server Error"
}

Video Models

Model	Speed	Quality	Best For
`veo-3.1-generate-preview`	Standard	Highest	Final production
`veo-3.1-fast-generate-preview`	Fast	High	Quick iterations
`veo-3.0-generate-001`	Standard	Stable	Reliable results
`veo-3.0-fast-generate-001`	Fast	Good	Rapid prototyping

Best Practices

1. Be Specific in Prompts

code

❌ "a cat"
✅ "A fluffy orange tabby cat lounging on a velvet cushion, soft studio lighting, portrait orientation, professional pet photography"

2. Match Aspect Ratio to Purpose

code

❌ Using 1:1 for YouTube thumbnail
✅ Using 16:9 for YouTube thumbnail

3. Use Appropriate Size

code

❌ 4K for a small web thumbnail (wasteful)
✅ 1K for web thumbnails, 2K for blog, 4K for print

4. Include Negative Prompts for Video

javascript

negative_prompt: "blurry, low quality, distorted, glitchy, pixelated"

5. Use Reference Images for Consistency

When generating multiple related images, use the first as reference for subsequent ones.

Error Handling

If generation fails:

•Check CLIProxyAPI is running at http://127.0.0.1:8317
•Verify prompt isn't too long or contains blocked content
•Try a simpler prompt
•Check output directory exists and is writable
•For videos, ensure patience (1-5 min generation time)

Integration with Other Skills

This skill works alongside:

•localization-text-handler: For UI text on generated images
•winui-function-writer: For image processing functions in code

Example Session

code

User: "Create a featured image for my blog post about AI coding assistants"

MicrosoftAgent: Loading skill: veo-vibe-banana

Analyzing request:
• Type: Image
• Use case: Blog featured image
• Recommended: 16:9 aspect ratio, 2K resolution

Crafting prompt:
"Modern tech blog header featuring an AI assistant concept, futuristic holographic
interface with code snippets floating in space, blue and purple color scheme,
clean minimalist design, professional, digital art style, high quality"

Generating...

veo-vibe-banana_generate_image(
  prompt: "Modern tech blog header featuring an AI assistant concept...",
  aspect_ratio: "16:9",
  image_size: "2K"
)

Result:
✅ Image generated successfully!
📁 Saved to: ./generated/image_2026-02-01T14-30-22_x7k9m2.png