Skip to content
ComfyUI tutorial ComfyUI Advanced Tutorial4. Stable Diffusion 3.5

Stable Diffusion 3.5 Workflow Tutorial in ComfyUI

Stable Diffusion 3.5 is the latest AI image generation model, offering multiple powerful model variants. With ComfyUI, users can easily perform local inference and experience the capabilities of these models.

This tutorial organizes the following resources, mainly about how to use Stable Diffusion 3.5 in ComfyUI:

  • Stable Diffusion 3.5 FP16 version ComfyUI related workflow
  • Stable Diffusion 3.5 FP8 version ComfyUI related workflow (low VRAM solution)
  • Stable Diffusion 3.5 online resources and API

Introduction to Stable Diffusion 3.5 Models

Stable Diffusion 3.5 is the latest generation AI image generation model released by Stability AI. It has achieved significant breakthroughs in image quality and prompt adherence, marking a new era in AI drawing technology. The series offers multiple model variants to meet different user needs:

Stable Diffusion 3.5 Diverse Outputs

Currently, Stable Diffusion 3.5 officially provides the following versions:

Model NameParametersFeatures
Stable Diffusion 3.5 Large8 billionThe most powerful base model, ideal for professional use at 1 MP resolution.
Stable Diffusion 3.5 Large TurboDistilledGenerates high-quality images in just 4 steps, significantly faster.
Stable Diffusion 3.5 Medium2.5 billionDesigned for consumer hardware, supports 0.25 to 2 MP resolution, easy to customize.

Stable Diffusion 3.5 Community License

The community license allows free use for research, non-commercial purposes, and organizations or individuals with annual revenue not exceeding $1 million. For more details, please refer to the Community License Agreement.

Stable Diffusion 3.5 Original FP16 Version ComfyUI Workflow

This section will introduce the installation of the official version models and the download of workflow files. The workflow files and examples are from the ComfyUI Blog.

Download Stable Diffusion 3.5 Model Files

  1. Download Stable Diffusion 3.5 model files
Resource NameDownload LinkInstallation Folder
Stable Diffusion 3.5 LargeDownload Linkmodels/checkpoint
Stable Diffusion 3.5 Large TurboDownload Linkmodels/checkpoint

Please refer to the instructions and installation file paths to complete the download and installation of the corresponding models. Choose one according to your needs.

  1. Corresponding clip model files
Resource NameDownload LinkInstallation Folder
clip_g.safetensorsDownload Linkmodels/clip
clip_l.safetensorsDownload Linkmodels/clip
t5xxl_fp16.safetensorsDownload Linkmodels/clip

If you have used version 3.0 or other related models, you might already have these files. If this is your first time using them, please download and install them.

Download Stable Diffusion 3.5 Workflow Files

Below are the related examples provided by ComfyUI. You can download the corresponding workflow file:

Drag this file into the ComfyUI interface and run the generation.

Stable Diffusion 3.5 FP8 Low VRAM Solution

If you encounter memory issues during generation, try the following low VRAM options:

Download Stable Diffusion 3.5 FP8 Model Files

  1. Download Stable Diffusion 3.5 FP8 model files
Resource NameDownload LinkInstallation Folder
Stable Diffusion 3.5 Large FP8Download Linkmodels/checkpoint

Please refer to the instructions and installation file paths to complete the download and installation of the corresponding models.

  1. Corresponding clip model files
Resource NameDownload LinkInstallation Folder
clip_g.safetensorsDownload Linkmodels/clip
clip_l.safetensorsDownload Linkmodels/clip
t5xxl_fp8_e4m3fn.safetensorsDownload Linkmodels/clip
(experimental)t5xxl_fp8_e4m3fn_scaled.safetensorsDownload Linkmodels/clip

If your memory exceeds 32GB, it is still recommended to use the t5xxl_fp16 model.

Download Stable Diffusion 3.5 FP8 Workflow Files

Below are the related examples provided by ComfyUI. You can download the corresponding workflow file:

Drag this file into the ComfyUI interface and run the generation.

Online Execution of Stable Diffusion 3.5

You can also run Stable Diffusion 3.5 online via Hugging Face Spaces:

In addition to downloading model weights from Hugging Face for self-hosting, users can also access Stable Diffusion 3.5 through the following platforms:

With ComfyUI, you can fully leverage the powerful features of Stable Diffusion 3.5, easily creating and managing image generation workflows. Whether for professional use or personal creation, Stable Diffusion 3.5 offers you exceptional image quality and flexibility.

For more information, visit the ComfyUI Blog.

Other Resources

  1. Stable Diffusion 3.5 LoRA Models https://huggingface.co/Shakker-Labs/SD3.5-LoRA-Linear-Red-Light https://huggingface.co/Shakker-Labs/SD3.5-LoRA-Futuristic-Bzonze-Colored https://huggingface.co/Shakker-Labs/SD3.5-LoRA-Chinese-Line-Art

  2. ControlNet Models

  • To be updated