Comfy Stack - AI Model Deployment

🔌 Step 1: SSH Access

First connect to your server:

ssh -i ~/.ssh/YOUR_KEY_FILE ubuntu@YOUR_SERVER_IP

If you get a host key warning, run:

ssh-keygen -R YOUR_SERVER_IP

This happens when the server IP changes or is reused.

🚀 Step 2: Clone & Deploy

# Clone the repository
rm -rf comfy-stack  # Remove existing directory if needed
git clone https://github.com/baihne/comfy-stack.git
cd comfy-stack

# Choose your deployment option:

For 100GB VPS (Recommended)

./deployment/deploy_hunyuan_2mv_optimized.sh TAILSCALE_AUTH_KEY standard

Access: http://100.x.x.x:7861

For Larger VPS (Hybrid Setup)

./deployment/deploy_hunyuan_hybrid.sh TAILSCALE_AUTH_KEY standard

Access: http://100.x.x.x:7862

ComfyUI + Wan2.2

./deployment/deploy_comfyui_wan.sh TAILSCALE_AUTH_KEY

Access: http://100.x.x.x:8188

ComfyUI + Flux 2 (dev, repack)

HF_TOKEN=<your_hf_token_if_needed> ./deployment/deploy_comfyui_flux2_dev.sh TAILSCALE_AUTH_KEY

Access: http://100.x.x.x:8188

ComfyUI + Flux 2 (full upstream)

HF_TOKEN=<your_hf_token_if_needed> ./deployment/deploy_comfyui_flux2_full.sh TAILSCALE_AUTH_KEY

Access: http://100.x.x.x:8188

One-step launcher (choose stack interactively)

./deploy.sh

Select Flux 2 (repack or full), Wan2.2, or Hunyuan variants; you can also input Tailscale/HF tokens when prompted.

📱 Step 3: Access Services

Service	Port	Access URL
Hunyuan3D-2mv	7861	http://100.x.x.x:7861
Hunyuan3D Hybrid	7862	http://100.x.x.x:7862
Hunyuan3D-2.1	7860	http://100.x.x.x:7860
ComfyUI	8188	http://100.x.x.x:8188

🎯 Model Variants

• standard - Best quality
• turbo - Fast generation
• fast - Rapid prototyping

📚 Additional Documentation

See docs/ folder for detailed guides and troubleshooting.