# SDXL Model ## 1 Call SDXL Model Click the **GenAI** button, then click the **SDXL** button, and then send prompts to generate the desired image. You can also send an image and use the caption as prompts to generate a new image. {% hint style="info" %} You can use the **SDXL Model** directly by sending the */gen\_sdxl* command. {% endhint %}

Call SDXL Model

## 2 Settings Click the **Settings** button, then click the **SDXL Settings** button to enter the setting page of the SDXL Model. You can set parameters such as **LoRA** and **LoRA Weight** for the SDXL Model to get a more appropriate output. {% hint style="info" %} You can enter the **Settings** page by sending the */settings* command. {% endhint %}

Parameter Settings for the SDXL Model

### 2.1 LoRA LoRA (Low-Rank Adaptation of Large Language Models) is a popular and lightweight training technique that significantly reduces the number of trainable parameters. Click the **LoRA** button and the bot will display the LoRA models that can be selected. You can also set a valid URL that can download the **SDXL 1.0** LoRA model by sending message. {% hint style="info" %} You can find LoRA models at . Note that the LoRA version must be **SDXL 1.0**. You can send multiple lora model urls, please use ^\*^ to connect lora model urls. {% endhint %}

Set Up LoRA Model

### 2.2 LoRA Weight LoRA weight represents the influence weight of the LoRA model. After the LoRA setting is completed, the corresponding LoRA weight needs to be set. {% hint style="info" %} When setting multiple LoRAs, the number of LoRA weights and the number of LoRAs must be equal. {% endhint %}

Set Up LoRA Weight

### 2.3 Negative Prompt Negative prompts tell GenAI model what you do not want in the output. Click the **Negative Prompt** button and send a message to customize negative prompts.

Custom Negative Prompts

### 2.4 CFG Scale The CFG scale is a number from 0 to 10, indicating the influence of prompts on the generated images.

Set Up CFG Scale

### 2.5 Steps The steps is a number from 0 to 50, indicating the number of denoising steps.

Set Up Steps

### 2.6 Seed The seed is an integer greater than 0. With other parameters being the same, the same seed produces the same output image. The default value for seed is 0, which means a random value is used.

Set Up Seed

## 3 Examples ### 3.1 Pre-provided LoRA Model We provide multiple LoRA models for users to choose from. Here we take China\_Splashed\_Ink as an example to generate images.

Set Up Pre-provided LoRA

We send prompt to generate an ink drawing style image.

Generate an Ink-style Image

### 3.2 Image-to-Image Let’s generate a boss battle style image using the pre-provided XL\_Boss\_Battle LoRA. We set the LoRA Weight to 0.9.

Set Up Boss_Battle LoRA

We use prompts to generate an image of fighting a T-Rex.

Generate an Image of Fighting a T-Rex

We use the generated image of fighting a T-Rex as the initial image to generate an image of fighting an alien.

Image-to-Image example

--- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://genai-doc.phoenix.global/sdxl-model.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.