Image Generation

Generate pixel art from text prompts using AI models. Supports style references, negative prompts, seed control, and multiple output sizes.

Text-to-Pixel-Art

The core generation endpoint accepts a text description and returns a generation job. The job runs asynchronously — poll the status endpoint to retrieve the result.

POST /generations/image
{
  "prompt": "16-bit red mage with cloak",
  "size": "128x128",
  "seed": 42
}

Style Reference

Guide the output aesthetic by providing a reference image. The AI extracts the style, color palette, and visual tone from your reference and applies it to your prompt.

POST /generations/style
{
  "prompt": "wooden treasure chest",
  "style_image": "base64_encoded_image",
  "strength": 0.6
}

Negative Prompts

Exclude unwanted elements by passing a negative_prompt. This helps avoid common artifacts or undesired features in your generations.

Seed Control

Set a specific seed value to produce reproducible outputs. This is useful for A/B testing prompts or generating consistent variations.

Image Generation

Text-to-Pixel-Art
Generate pixel art from text prompts with model selection, style control, negative prompts, and seed support.
Style Reference
Provide an image as style reference to guide the output aesthetic.
UI Elements
Generate game UI elements — buttons, panels, icons, and HUD components.