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

# Image Mode

> Generate images inside chat with Nano Banana 2, Gemini 2.5 Flash, and friends.

Flip Foundry into **Image Mode** and your next prompt produces images instead of text. Same chat, same history — different engine.

Toggle it from **Settings → Image Mode**, or hit `/` and pick **Image** from the mode switcher.

<Note>
  **Screenshot coming.** Image Mode toggle + first generation.
</Note>

## Models

Foundry ships with three image models, all Gemini-family under the hood. Pick from the model dropdown in Image Mode.

| Model | Best for | Notes |
| - | - | - |
| **Nano Banana 2** (`gemini-3.1-flash-image-preview`) | The flagship. Strongest generations, most controls. | Supports extended aspect ratios (up to 8:1 and 1:8) and resolutions from 0.5K to 4K. |
| **Gemini 2.5 Flash Preview** (`gemini-2.5-flash-image-preview`) | Faster iterations on standard aspect ratios. | Standard aspect ratios and quality presets only. |
| **Gemini 2.5 Flash** (`gemini-2.5-flash-image`) | Stable fallback. | Same controls as the Preview variant. |

## Controllable options

| Control | Range | Available on |
| - | - | - |
| **Quantity** | 1–4 images per generation | All models |
| **Aspect ratio** | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3 | All models |
| **Extended aspect ratio** | 4:1, 1:4, 8:1, 1:8 | Nano Banana 2 only |
| **Quality** | Low / Standard / HD (1K / 2K / 4K) | All models |
| **Resolution** | 0.5K · 1K · 2K · 4K | Nano Banana 2 only |

The composer hides controls a model doesn't support — switch to Nano Banana 2 to unlock the wider aspect ratios.

## Tips for prompts

* **Lead with the subject**, then style, then composition. *"A bonsai pine in a ceramic pot, watercolor, centered on cream paper."*
* **Cite a style or artist** when relevant — Gemini handles named styles well.
* **Iterate inline.** Don't restart — reply *"now make the pot blue and shift the composition slightly left"* and Foundry edits from the prior generation.
* **Use 4 generations** for divergent ideas, 1 for refinement passes.

## Limits

* No negative-prompt UI today — phrase exclusions in natural language inside the prompt (*"…no people, no text overlays"*).
* No reference-image upload control in the current Image Mode controls. Use Agent Mode + an image-aware Skill if you need image-in/image-out.
* Generated images live in the chat; right-click → Save to download. We don't surface a gallery view yet.

## Related

<CardGroup cols={2}>
  <Card title="Video Mode" icon="video" href="/foundry/modes/video">
    Move from stills to motion.
  </Card>

  <Card title="3D Website Mode" icon="cube" href="/foundry/modes/3d-websites">
    Brief in, immersive site out.
  </Card>

  <Card title="Connectors: Fal AI" icon="plug" href="/foundry/connectors/pre-installed">
    Heavier image/video lifting via Fal.
  </Card>

  <Card title="Tutorial: First image" icon="play" href="/foundry/tutorials/first-image">
    Hands-on walkthrough.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.