Skip to main content
With the Decart API, you can generate high-quality images from text descriptions. Simply describe what you have in mind, and let the model bring it to life.
If you want more creative input from the model, you can also keep your description broad and let Lucy fill in the details.
Our models support a wide range of styles—from portraits and selfies to cartoons, abstract art, and more.

​
Quick start

Here’s a simple example using Lucy-Pro, our highest quality model. 
curl -X POST https://api.decart.ai/v1/generate/lucy-pro-t2i \
  -H "X-API-KEY: $DECART_API_KEY" \
  -F "prompt=A serene mountain landscape at sunset" \
  -F "resolution=720p" \
  --output generated-image.jpg

​
Parameters

  • prompt (required) — Text description of the image to generate.
  • resolution (optional) — Output size: 480p or 720p (default: 720p).
For complete API documentation including response formats and error codes, see the API Reference.

​
Available Models and Options

Decart provides different configurations to balance speed, quality, and cost depending on your use case. We currently offer two model tiers:
  • Lucy-Pro — our flagship model for generating diverse, high-quality images.
  • Lucy-Dev (Coming Soon) — a faster, cheaper model ideal for quick prototyping, simple or small images.
Supported resolutions:
  • 720p (720×1280) — detailed, higher quality, ideal for large displays.
  • 480p (480×832) — faster, cheaper, well-suited for prototyping or smaller outputs.

​
Prompt Engineering

Prompt engineering is about giving AI models the right kind of detail so they reliably produce what you have in mind. Because image generation is unpredictable, the way you phrase instructions directly shapes how cinematic, realistic, or stylized the result will be. With Lucy, prompts that borrow from visual design and photography—camera angles, lighting, atmosphere, composition, and style—tend to produce far richer results than simple object descriptions. Strong prompts are specific (around 80–100 words), contextualized (time, place, mood), and structured to describe both the subject and how it is framed. Our testing shows that Lucy handles some directions well—like close-ups, wide shots, low or high angles—while struggling with extreme or abrupt changes. Subtler framing cues and clearly described lighting (soft light, backlight, volumetric, etc.) are more reliable. The key is to avoid vague or overly complex instructions and instead lean on precise, cinematic language. Bottom line: write prompts as if you’re composing a photograph or directing a scene. Set the subject, define the mood, and specify perspective and lighting deliberately. By iterating and refining, you’ll discover which cues Lucy respects and how to guide it toward consistently compelling images.

​
Endpoint and Full API Reference

​
Next steps

Now that you know the basics of image generation, you might want to check out one of these resources next.

Generate an image in the Platform

Try out this and other models in the interactive platform.

Full API reference

Check out all the options for image generation in the API reference.

Image Editing

Learn how to edit your images LucyEdit-Pro.
⌘I