Gemini Vertex AI icon

Gemini Vertex AI

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

Actions5

Image → Exploration

AI-generated

Summary

Generate multiple variants of a reference image using different prompts in parallel. Provide the reference image via the Input Images parameter and a list of prompts via the Prompts List parameter; each prompt produces a separate output image.

Inputs

  • inputImages (required) — Reference image(s) to vary. Supply a binary field (e.g., 'data') containing the image and its MIME type.
  • promptsList (required) — Collection of text prompts; each prompt creates one variation of the reference image.
  • modelId — Model to use for generation. Defaults to 'gemini-2.5-flash-image'. Imagen 4 models expose additional parameters (aspect ratio, size, watermark, seed, etc.).
  • outputBinaryField — Name of the binary field where each generated image will be stored. Defaults to 'image'.
  • outputFilename — Base filename for outputs; a suffix (_1, _2, …) is added when multiple prompts are provided. Defaults to 'generated_image.png'.

Output shape

Each input item yields N output items, where N is the number of prompts. Each output item contains a JSON field with success metadata and a binary field (named by outputBinaryField) holding the generated image.

If continueOnFail is enabled, failures return an error JSON instead of throwing.

Examples

Example 1: Create three style variations of a product photo.

Set Input Images to the product photo binary field, add three prompts to Prompts List (e.g., 'vintage style', 'futuristic neon', 'pastel sketch'), leave Model at default, and keep Output Binary Field as 'image'.

Links

Discussion