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

# Image Generation

> AI image generation with style presets, quality tiers, and variation support. Generate and customize images through conversation.

# Image generation

The DeFAI chat supports AI-powered image generation via DALL-E and Stable Diffusion models through OpenRouter. Generate images through natural language prompts, then customize with variations, style changes, and quality upgrades.

## How to use

Select any agent or type directly:

* `"Generate a cyberpunk city at night"`
* `"Create an image of a futuristic trading terminal"`
* `"Draw a logo concept for a crypto project"`

## Parameter request

Before generation, a **ParamRequestPanel** may appear with:

| Option           | Choices                                                                                   |
| ---------------- | ----------------------------------------------------------------------------------------- |
| **Quality tier** | Flash (cheapest, fastest), Standard, Premium (highest quality) — each shows price in USDC |
| **Aspect ratio** | 1:1, 16:9, 9:16, 4:3, 3:4                                                                 |
| **Style preset** | Auto, Vibrant, Muted, Cinematic, B\&W, Anime, Neon, Vintage                               |
| **Background**   | Auto, Transparent, White, Blur                                                            |
| **AI model**     | Model selector for choosing the generation model                                          |

## Result panel: ImageGenResultPanel

| Element                      | Description                                                        |
| ---------------------------- | ------------------------------------------------------------------ |
| **Header**                   | "SuperImageGen" icon, cost badge (USDC amount), info toggle        |
| **Image display**            | Square aspect ratio with loading skeleton and error state fallback |
| **Download button**          | Save the generated image                                           |
| **Info panel** (collapsible) | Model used, image size, detected style                             |

## Action cards

After generation, three action cards appear for customization:

<CardGroup cols={3}>
  <Card title="Variation" icon="clone">
    Choose variation intensity: Similar (small changes), Different (moderate changes), Extreme (major reimagining). Each triggers a new generation with payment.
  </Card>

  <Card title="Style" icon="palette">
    Pick a new style from the same 8 presets (Vibrant, Muted, Cinematic, B\&W, Anime, Neon, Vintage, Auto). Applies the style to the existing image concept.
  </Card>

  <Card title="Quality" icon="sparkles">
    Upgrade or change the quality tier (Flash/Standard/Premium). Higher tiers produce more detailed, higher-resolution images.
  </Card>
</CardGroup>

## Image modification

A **ImageModificationPanel** also supports:

* Pre-built modification presets (Vibrant, Neon, Vintage, Anime, Cinematic, Enhance, B\&W)
* Custom prompt textarea for specific modifications
* AI model selector
* Payment + execution flow

Each variation/modification is a new generation that requires its own USDC payment via the X402 payment flow.

<Info>
  Image generation costs vary by quality tier. Flash is the cheapest (\~\$0.08), Premium is the most expensive but produces the highest quality. Costs are shown before you confirm.
</Info>