banana
Generate and edit images using AI with text prompts and reference images, supporting multiple aspect ratios, resolutions, and style transfers.
Instructions
Generate images using Nano Banana Pro (Gemini 3 Pro Image).
CAPABILITIES:
Text-to-image generation with high quality output
Image editing and transformation with reference images
Multiple aspect ratios and resolutions (1K/2K/4K)
Style transfer and multi-image fusion
Optional search grounding for factual content
RESPONSE FORMAT:
Returns XML with file paths to generated images
Images are saved to disk (no base64 in response)
Includes text descriptions and optional thinking process
BEST PRACTICES:
Be descriptive: describe scenes, not just keywords
Use negative constraints in prompt: "no text", "no watermark"
For editing: provide reference image and specify what to keep
For style transfer: provide style reference image
Supports: reference images with roles (edit_base, style_ref, etc.).
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| prompt | Yes | Image generation prompt. Structure: <goal>what you want to generate (can be a statement)</goal> <context>detailed background info - the more the better</context> <hope>desired visual outcome, can be abstract</hope>. Example: <goal>Generate 6 weather icons for a mobile app</goal> <context>Target users are young professionals, app has a friendly casual vibe, needs to match existing UI with rounded corners</context> <hope>pastel colors, consistent 3px stroke, 64x64 base size</hope> | |
| images | No | Reference images for editing or style transfer. Roles: edit_base (image to edit), subject_ref (person/character), style_ref (style reference), layout_ref (layout), background_ref, object_ref. | |
| aspect_ratio | No | Output image aspect ratio. Default: 1:1 (square). | 1:1 |
| resolution | No | Output resolution. 1K (1024px), 2K (2048px), 4K (4096px). Default: 4K. | 4K |
| use_search | No | Enable search grounding for factual content. Adds text to response. | |
| include_thoughts | No | Include model's thinking process in response. | |
| temperature | No | Controls randomness (0.0-2.0). Higher = more creative. Default: 1.0. | |
| top_p | No | Nucleus sampling threshold (0.0-1.0). Default: 0.95. | |
| top_k | No | Top-k sampling (1-100). Default: 40. | |
| num_images | No | Number of images to generate (1-4). Default: 1. | |
| save_path | Yes | Base directory for saving images. Files saved to {save_path}/{task_note}/. | |
| task_note | Yes | Subdirectory name for saving images (English recommended, e.g., 'hero-banner', 'product-shot'). Also shown in GUI. |