Back to Generator

Local Installation

Set up your own uncensored image generation system with these steps:

System Requirements

  • Windows/Linux/Mac (Mac with M1/M2 recommended for best performance)
  • 16GB+ RAM (32GB+ recommended for 8K+ generations)
  • NVIDIA GPU with 8GB+ VRAM (24GB+ recommended for 16K generations)
  • Python 3.10+ installed
  • 50GB+ disk space for models and dependencies

Quick Installation

Step 1: Clone Repository

git clone https://github.com/uncensoredcanvas/ai-core.git
cd ai-core

Step 2: Install Dependencies

pip install -r requirements.txt
# For CUDA acceleration:
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Step 3: Download Models

python download_models.py --xl --refiner --loras --controlnet

(Approx. 35GB download - includes SDXL 1.0 base + refiner, 15 popular LoRAs, and ControlNet models)

Step 4: Launch Interface

# For web interface:
python webui.py --listen --no-hashing

# For optimized CLI:
python generate.py --prompt "your prompt" --negative "negative prompt" --steps 50 --width 4096 --height 4096

Advanced Configuration

Custom Models

Place .safetensors or .ckpt files in:
/models/Stable-diffusion/

Place LoRA files in:
/models/Lora/

Place ControlNet files in:
/models/ControlNet/

Performance Optimization

Enable xFormers

Reduce VRAM usage by 20-30%

Use TensorRT

2-3x faster generation on NVIDIA GPUs

FP16 Precision

Faster with minimal quality loss