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

# Generation settings

> Aspect ratio, batch size, reference weights, and the settings specific to each model.

Open the gear icon beside the prompt box.

<Frame caption="The generation settings popover">
  <img src="https://mintcdn.com/wji/1nwL1jaWRdmAhYLq/images/image-generation/settings-01.png?fit=max&auto=format&n=1nwL1jaWRdmAhYLq&q=85&s=7bb42e6807db3650e7a9ebf2bcbebe4f" alt="Generation settings popover showing aspect ratio thumbnails, batch size slider, and weight sliders" width="1600" height="900" data-path="images/image-generation/settings-01.png" />
</Frame>

## Settings you will see most

| Setting              | What it does                                                      |
| -------------------- | ----------------------------------------------------------------- |
| **Aspect ratio**     | A grid of ratio thumbnails. Only ratios the model supports appear |
| **Batch size**       | How many results per run. The range comes from the model          |
| **Reference weight** | How strongly references influence the result                      |
| **Omni weight**      | On models that support it                                         |
| **Style weight**     | On models that support it                                         |
| **Stylize**          | Typically 0–1000, on Midjourney-family models                     |
| **Style prompt**     | Opens the style picker                                            |
| **Duration**         | Video only                                                        |

**Auto match** is a valid aspect ratio choice on some models — it matches the input image rather
than forcing a shape.

## Seed

When a model exposes a seed, it appears as a tag on the main form. `-1` means unspecified, so the
model picks its own.

The seed is what makes a result repeatable. If you get an image that is nearly right, note its seed
from the history card — clicking the seed tag copies it — and reuse it while you adjust the prompt.

## Model-specific settings

Everything else in the popover is generated from the model definition and changes as you switch
models. Each control has a tooltip explaining what it does.

Common ones:

* **Resolution** and **output format**
* **Chaos** — variety, 0 to 100
* **Quality**
* **Image weight** and **style reference weight** on Midjourney-family models
* **Raw mode** and **HD**

## Batch size and storyboards

Choosing a storyboard layout forces batch size to 1, because the grid *is* the single output. See
[Storyboard mode](/image-generation/storyboard-mode).