ComfyUI NSFW local setup
This guide installs an ahegao.red pack on your own Windows or Linux machine. One command sets up Docker, ComfyUI, and every model file the pack needs. The first start downloads 30 to 90 GB once. After that, generating runs at http://localhost:8188, on your GPU, with no uploads.
What you need
Check your computer against the pack you bought before you start.
| Your purchase | NVIDIA GPU memory | Free disk space |
|---|---|---|
| Image Pack | 16 GB+ | 30 GB |
| Video Pack | 32 GB+ | 60 GB |
| Image + Video Pack | 32 GB+ | 90 GB |
- A recent NVIDIA driver (check with nvidia-smi)
- Docker Desktop on Windows, or Docker with the NVIDIA Container Toolkit on Linux
- Your license key from ahegao.red
- A Hugging Face read token
- A CivitAI API key
Install on Windows
Open Windows PowerShell, not the old Command Prompt. Run:
irm https://ahegao.red/setup-local.ps1 | iex
If the installer asks for a restart, restart the computer and run the same command again.
Install on Linux
Open a Bash terminal and run:
bash <(wget -qO- https://ahegao.red/setup-local.sh)
Finish the setup
The installer checks your computer and asks for four items. Keep the terminal window open.
| The installer asks for | Where to get it |
|---|---|
| Your pack | Select Image, Video, or Image + Video when prompted |
| License key | Shown on ahegao.red after checkout. Keep it private |
| Hugging Face read token | huggingface.co, Settings, Access Tokens, type Read |
| CivitAI API key | civitai.com, Account settings, API Keys |
The first start downloads all model files. This happens once.
When the terminal reports "To see the GUI go to: http://0.0.0.0:8188", open http://localhost:8188 in your browser.
- Open the Workflows panel. Your purchased workflows appear there.
- Open one and replace the example prompt.
- Press Run.
Daily use
The workflow container starts automatically with Docker.
- Open http://localhost:8188 any time.
- The ahegao-workspace folder in your home directory stores all models and outputs. They survive updates and restarts.
- Stop the workflow with docker stop ahegao. Start it again with docker start ahegao. Both commands work in PowerShell and in Bash.
Updates
Updates ship weekly. Run your install command again (the Windows or the Linux one).
The update keeps all models and outputs.
Troubleshooting
| Symptom | Resolution |
|---|---|
| The command fails immediately | Paste it into the right window: PowerShell on Windows, Bash on Linux. Copy it as a single line |
| The command cannot be downloaded | Check your internet access. Company networks, VPNs, and ad-blockers can block ahegao.red |
| Docker is missing or shows no engine | Windows: start Docker Desktop and wait for "Engine running". Linux: run docker info and start the Docker service |
| No NVIDIA GPU found | Install the latest NVIDIA driver, restart, and run nvidia-smi. Continue once it shows your GPU |
| Low GPU memory warning | Setup continues. Try the workflow first. If renders run out of memory, use the RunPod setup instead |
| "The registry rejected the license key" | Copy the key again and paste it without extra spaces or lines |
| Docker cannot use the GPU (Windows) | Run wsl --update, restart, start Docker Desktop, and enable WSL integration under Settings, Resources |
| localhost:8188 does not open | Wait until the terminal prints the GUI line and keep Docker running, then reload |
| Not enough disk space | Free at least 30 GB (Image), 60 GB (Video), or 90 GB (Image + Video) and run setup again |
| Output quality looks distorted | The shipped workflows are pre-calibrated. Do not modify sampler, CFG, or LoRA settings |
Support
Write to support@ahegao.red if the steps above do not solve your problem. Include your license key, the pack you bought, and the full terminal output.
For quick questions, join the community Discord. The invite link is on ahegao.red.
Prefer a printable copy? The same setup ships as instructions-local.pdf on ahegao.red.
Adults only (18+). All sample output is synthetic, computer-generated fiction. Last verified October 2026.