Gemini Vertex AI icon

Gemini Vertex AI

Generate or edit images using Gemini 2.5 Flash Image (Nano Banana) via Vertex AI

Actions5

Image → Fusion

AI-generated

Summary

Fuses multiple input images using a text prompt via a Gemini or Imagen 4 model.

Inputs

  • projectId (required) — Google Cloud Project ID
  • location (required) — Google Cloud region (e.g., us-central1, global)
  • modelId (required) — Gemini or Imagen model to use (or 'custom' to specify a custom model ID)
  • customModelId — Custom model ID when modelId is set to 'custom'
  • prompt (required) — Text prompt describing how to fuse the input images
  • inputImages (required) — One or more images to fuse; each entry specifies the binary field name and MIME type (PNG, JPEG, WEBP, GIF)
  • outputBinaryField — Name of the binary field to store the generated image (default: 'image')
  • outputFilename — Filename for the generated image (default: 'generated_image.png')
  • aspectRatio — Aspect ratio for Imagen 4 models (1:1, 3:4, 4:3, 9:16, 16:9)
  • imageSize — Resolution for Imagen 4 Standard/Ultra models (1K, 2K; Fast does not support 2K)
  • negativePrompt — Things to exclude from the image (Imagen 4)
  • enhancePrompt — Whether to auto‑rewrite the prompt for better results (Imagen 4)
  • addWatermark — Add an invisible SynthID watermark; must be false to use Seed (Imagen 4)
  • seed — Integer for reproducible results (Imagen 4 only when watermark is off)

Output shape

binary

Output image is placed in the binary field named by outputBinaryField (default: image).

Examples

Example 1: Fuse a product photo with a background image using a prompt like 'Place the product on the background with realistic lighting'.

Provide Project ID, Region, select a Gemini model (e.g., gemini-2.5-flash-image), set Prompt, add two Input Images (product and background), keep default outputBinaryField and outputFilename.

Links

Discussion