This is a simple one-click workflow for generating SillyTavern expressions — now updated to Version 2. Here’s what you’ll need:

Required Tools:

• ComfyUI Manager (for missing nodes): https://github.com/Comfy-Org/ComfyUI-Manager

• YOLOv8 Face Detection: https://github.com/akanametov/yolo-face

• SAM Model (download sam_vit_b_01ec64.pth): https://huggingface.co/YouLiXiya/YL-SAM/blob/main/sam_vit_b_01ec64.pth

File Directory Setup:

• SAM model → ComfyUI_windows_portable\ComfyUI\models\sams\sam_vit_b_01ec64.pth

• YOLOv8 model → ComfyUI_windows_portable\ComfyUI\models\ultralytics\bbox\yolov8m-face.pt

Don’t worry — it’s super easy. Just follow these steps:

1. Enter the character’s name.

2. Load the image.

3. Set the seed, sampler, steps, and CFG scale (for best results, match the seed used in your original image).

4. Add a LoRA if needed (or bypass it if not).

5. Hit "Queue".

The output image will have a transparent background by default.
Want a background? Just bypass the BG Remove group (orange group).

Expression Groups:

• Neutral Expression (green group): This is your character’s default look in SillyTavern. Choose something that fits their personality — cheerful, serious, emotionless — you know what they’re like.

• Custom Expression (purple group): Use your creativity here. You’re a big boy, figure it out 😉

Pro Tips:

• Use a neutral/expressionless image as your base for better results.

• Models trained on Danbooru tags (like noobai or Illustrious-based models) give the best outputs.

Have fun and happy experimenting! 🎨✨