English Version:
Qwen creates prompts for SDXL.
Download the model at this link.
Install the required packages using the following command:
pip install -r requirements.txtUse the following command-line arguments to operate this script:
-c,--checkpoint-path: Specifies the checkpoint name or path, defaulting tohahahafofo/Qwen-1_8B-Stable-Diffusion-Prompt.-x,--sdxl-path: Defines the SDXL Checkpoint name or path. Options are:Lykon/dreamshaper-xl-turbo, recommendednum_inference_stepsare between 6 and 16 for enhanced image quality.stabilityai/sdxl-turbo, suggestednum_inference_stepsare between 1 and 4 for quicker processing.
--share: Generates a publicly shareable link for the interface. Default is set toFalse.--inbrowser: Opens the interface in a new tab of the default browser automatically. Default isFalse.--server-port: The port number for the demo server. Default is8000.--server-name: The name for the demo server. Default is0.0.0.0.--file-server-port: Port number for the file server. Default is8001.--qwen-only: Only use Qwen to generate prompts. Default isFalse.
Run the web demo with the following command:
python web_demo.pyChinese:
Qwen 用于为 SDXL 创建Prompt。
在此链接下载模型。
使用以下命令安装所需包:
pip install -r requirements.txt使用以下命令行参数操作此脚本:
-c,--checkpoint-path:指定检查点名称或路径,默认为hahahafofo/Qwen-1_8B-Stable-Diffusion-Prompt。-x,--sdxl-path:定义 SDXL 检查点名称或路径。选项包括:Lykon/dreamshaper-xl-turbo,建议num_inference_steps在 6 到 16 之间,以获得更好的图像质量。stabilityai/sdxl-turbo,建议num_inference_steps在 1 到 4 之间,以实现更快的处理速度。
--share:生成接口的公开共享链接。默认设置为False。--inbrowser:自动在默认浏览器的新标签页中打开接口。默认为False。--server-port:演示服务器的端口号。默认为8000。--server-name:演示服务器的名称。默认为0.0.0.0。--file-server-port:文件服务器的端口号。默认为8001。--qwen-only: 仅使用 Qwen 生成 Prompt。默认为False。
使用以下命令运行:
python web_demo.py