This worflow was inspired by：

https://www.reddit.com/r/StableDiffusion/comments/17cpa3w/i_noticed_some_coherent_expression_workflows_got/

Hello! My bro. It's the first time I've posted a workflow on openart.  Thanks for you attension!

CheckPoint used:

[IP DESIGN | 3D可爱化模型]

Download it:  https://civitai.com/models/89804/ip-design-or-3d

These node widgets can be adjusted:

1. FaceID

-  [Group] First Diffusion -> EasyLoader (A1111)

   lora_model_strength: It is recommended that the value is not too large. I think between 0~0.5 is appropriate

   a1111_prompt_style: Can generate images similar to a1111

   (To use this, You need to Download ComfyUI_smZNodes)

-  [Group] ipadapter FaceId -> Apply Apply FaceID

   weight

   weight_v2

2. Canny

-  [Group] Controlnet -> EasyControlet (Advanced)

   stregth: I think between 0.35~0.5 is appropriate, it depends largely on the image

   scale_soft_weights: It maybe to be less than 1 or the transition won't be soft enough.

   (To use this, You need to Download ComfyUI-Advanced-ControlNet)

3. Expression Mask

-  [Group] Facial segmentation -> ImageBlur

   blur_radius: Adjustment according to the picture effect

-  [Group] Facial segmentation -> Image Shadows and Highlights

   highlight_threshold: I think between 0~5 is appropriate

-  [Group] Mask paste -> Mask Dilate Region

   iterations: Adjustment according to the picture effect

-  [Group] Mask paste -> ImageBlur

   blur_radius: Adjustment according to the picture effect