> ## Documentation Index
> Fetch the complete documentation index at: https://learn.workflow.dog/llms.txt
> Use this file to discover all available pages before exploring further.

# Generate Image

> Create an image with a Google Gemini image model.

The Google Gemini **Generate Image** action returns one generated image file.
It supports a broad set of aspect ratios and model-dependent output
resolution.

## Inputs

| Input                   | Type                         | Required  | Description                                                                       |
| ----------------------- | ---------------------------- | --------- | --------------------------------------------------------------------------------- |
| **Third-party account** | Google Generative AI account | Yes       | The API key used for generation.                                                  |
| **Prompt**              | String                       | Yes       | A description of the image to create.                                             |
| **Model**               | Enum                         | Yes       | Gemini 2.5 Flash Image or Gemini 3 Pro Image. Defaults to Gemini 2.5 Flash Image. |
| **Aspect Ratio**        | Enum                         | Yes       | `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `4:5`, `5:4`, `9:16`, `16:9`, or `21:9`.       |
| **Image Size**          | Enum                         | Sometimes | `1K`, `2K`, or `4K`; shown only for Gemini 3 Pro Image. Defaults to `2K`.         |

## Output

| Output    | Type | Description                                                |
| --------- | ---- | ---------------------------------------------------------- |
| **Image** | File | The generated image with the MIME type returned by Gemini. |

<Note>
  WorkflowDog makes up to three generation attempts when Gemini returns no image
  data. The action fails if all attempts produce no image.
</Note>

<Tip>
  Pick the aspect ratio for the destination first—such as `9:16` for a story or
  `16:9` for a banner—then describe composition with that frame in mind.
</Tip>