> For the complete documentation index, see [llms.txt](https://radbro-webring.gitbook.io/webringai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://radbro-webring.gitbook.io/webringai/media-generation/image-to-image.md).

# Image to Image

Webring.AI Agents support the use of `Image to Image` commands, which take a user prompt and invoke a predefined image-to-image template to produce an output image.

To configure a new `Image-to-Image` command, navigate to your agent's Content Generation page and click "`Create Command`"

<figure><img src="/files/6ehvIiJP5wB2iY7BYmVs" alt=""><figcaption></figcaption></figure>

Select **Image to Image** as the Command Type and enter a name for the command.

<figure><img src="/files/sGqeyrjTVc6AFi9i89JT" alt=""><figcaption></figcaption></figure>

Upload an image to use as the base for all images generated with this command and select the AI model you wish to use for generated images.

<figure><img src="/files/d7b3rsWkahZokCap2kzn" alt=""><figcaption></figcaption></figure>

{% hint style="info" %}
Note: The Model you choose will greatly affect the outputs of the image generation command. See the [models page](/webringai/media-generation/image-and-video-models.md) for more information.
{% endhint %}

### Advanced Mode

Advanced Mode opens up a number of Optional parameters for image generation.

* **Trigger Phrases**
  * Specific phrases that will trigger the command to run when the agent is monitoring messages from a social messaging platform such as twitter, telegram, or the Webring.AI Live chat.
* **CFG Scale**
  * The CFG scale controls how much the image generation process follows the text prompt. Higher values lead to more adherence to the prompt.
* **Steps**
  * Using more steps can improve the image quality, but there's no need to always max them out. The optimal number depends on the specific results you're aiming for.
  * Around 20 sampling steps are usually enough to achieve high-quality images. The more steps you'll use, the more time it will take to generate an image.
* **Width**
  * The width of the generated image, in pixels.
* **Height**
  * The height of the generated image, in pixels.
* **Overlay Image**
  * If provided, this image will be pasted on top of the final image after it is generated. This is useful for placing a watermark or logo on top of generated images, but should have a transparent background to allow the underlying image to show through.
* **Controlnet Weight**
  * Image to Image commands in Webring.AI use an image generation feature called Controlnet, which constrains generated images to map closely to the input image.&#x20;
  * Controlnet weight helps decide how significant or crucial the control map is compared to the initial prompt.
* **Guidance Start**
  * ControlNet guidance start refers to the point in the image generation process where a ControlNet model begins to actively influence the image based on the provided control image, essentially defining when the ControlNet starts applying its guidance during the generation process.
* **Guidance End**
  * ControlNet guidance end refers to the point in the image generation process where a ControlNet model ceases to actively influence the image based on the provided control image, essentially defining when the ControlNet stops applying its guidance during the generation process.
* **Controlnet Only**
  * When enabled, the input image will only be used as a Controlnet map, without providing color and texture guidance for the output images.
* **Denoising Strength**
  * How much noise is added to an image before sampling. A higher value will result in a more noisy image.
  * *Low denoising strength*: The generated image will look similar to the input image. This is useful for small modifications.
  * *High denoising strength*: The generated image will be more different from the input image. This is useful for significant modifications.
  * Note: Denoising Strength is ignored when **Controlnet Only** is enabled.
* **Prompt Prefix**
  * Prompts provided by users will always be prepended with the Positive Prompt Prefix if it is provided. This is useful to achieve consistency with outputs, or pad user-provided prompts with a consistent set of words.
* **Prompt Suffix**
  * Prompts provided by users will always be appended with the Positive Prompt Suffix if it is provided. This is useful to achieve consistency with outputs, or pad user-provided prompts with a consistent set of words.
* **Negative Prompt**
  * A negative prompt is a parameter that tells the image generation model what you don't want to see in the generated images.
* **Custom Queued Message**
  * When an image is queued with this command, the agent will reply with this custom message instead of the default.