ComfyUI-SD3-nodes

ComfyUI-SD3-nodes
★ 8

ComfyUI 兼容SD3 支持节点封装工作流简化
为 Stable Diffusion 3 Medium 提供支持的 ComfyUI 节点集合，封装内置节点并简化使用流程，方便构建与共享 SD3 工作流。
💡 快速搭建并运行支持 SD3 的 ComfyUI 工作流。
🍴 4 Forks💻 Python🔄 2024-06-14
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/b6135d9bd930
SD3 Default Workflow
📄 README

ComfyUI-SD3-nodes

Nodes that support Stable Diffusion 3 Medium and are a little bit easier to understand. They are a wrapper of ComfyUI’s built-in nodes.

Download the JSON format workflow

Requirements

  • Upgrade ComfyUI to the latest version. You can either use ComfyUI-Manager to update, or run git pull in the ComfyUI folder.

    • Node List:

    1. SD3 Load Checkpoint

    Load the SD3 models.

  • ckpt_name: Choose the SD3 model. If you don’t have the model, please go to Huggingface, provide personal information and download it.

  • shift: A hyperparameter. According to the SD3 paper, it works best at 3.0 and 6.0.

    • 2. SD3 Load CLIPs

    Load the three CLIPs SD3 borrows:

    CLIP-G,

    CLIP-L,

    T5 XXL

  • clip-g: Choose the CLIP-G model.

  • clip-l: Choose the CLIP-L model.

  • t5xxl: Choose the T5 XXL model.

    • 3. SD3 Empty Latent

  • resolution: Choose a resolution from the recommended presets. It follows this guideline: “Resolution should be around 1 megapixel and width/height must be multiples of 64”.

  • batch_size: The number of images that will be generated in one batch.