# Image Generation I/O

### Intro

The Image Generation Endpoint enables you to generate unique images based on text descriptions (prompts). The generated images have a size of 1024x1024 pixels. You can further improve their resolution and quality using our Image Editing API (see the Resizing and Upscale sections).

### Input

Our image generation API accepts text prompts as inputs. The prompt must be between 3 and 1024 characters.

{% hint style="info" %}
Provide a detailed prompt to increase the likelihood of achieving the desired result.
{% endhint %}

```json
{
    "input": "A delicious ceviche cheesecake slice"
}
```

#### Limitations

Our system has a built-in filter that removes NSFW content. If you receive a plain black image as an output, it's possible that the content filter detected an NSFW result. If you believe this was a mistake, you can try adjusting the prompt.

### Output

Configuring output is optional. You may configure the output or omit it for default settings. If you omit the output, the API stores the result in a temporary bucket for 24 hours and provides you with a public URL to the image.

#### Connected storage

Alternatively, you can use a connected Cloud Storage. Once connected, you can refer to the storage directory as an output path. The output path must always end with a <mark style="color:purple;">`/`</mark> character.

{% hint style="info" %}
Currently, our API supports AWS S3 and Google Cloud Storage.
{% endhint %}

```json
{
    "output": "storage://storage-name/path/"
}
```