Fooocus: One-Click Stable Diffusion for Beginners

Fooocus is the antidote to complexity. While ComfyUI requires learning node-based workflows and AUTOMATIC1111 demands understanding dozens of settings, Fooocus delivers professional-quality image generation through a streamlined, Midjourney-inspired interface that “just works.”

This guide shows you how to deploy Fooocus, generate stunning images with minimal configuration, and understand when Fooocus is the right choice versus more advanced tools.

What You’ll Learn:

• Install Fooocus in under 5 minutes
• Generate high-quality images with simple prompts
• Use built-in styles and presets
• Understand image-to-image and inpainting
• Choose between Fooocus, ComfyUI, and AUTOMATIC1111

Time Investment: 15 minutes for setup | 30 minutes to master basics

Hardware Requirements: 6GB VRAM minimum | 8GB+ recommended

---

Understanding Fooocus

What is Fooocus?

Fooocus is a Stable Diffusion interface focused on simplicity and quality:

Design Philosophy:

• Minimal Configuration: Smart defaults for everything
• Quality First: Pre-tuned for professional results
• Automatic Optimization: Handles technical details automatically
• Fast Iteration: Quick prompt-to-image workflow

Key Features:

• Midjourney-style simplicity
• Built-in styles (Anime, Photorealistic, Cinematic, etc.)
• Automatic prompt enhancement
• Advanced sampling without complexity
• Inpainting and outpainting built-in

Fooocus vs Alternatives

---

Installation

Docker Installation (Easiest)

mkdir -p /opt/docker/fooocus
cd /opt/docker/fooocus

cat > docker-compose.yml << 'EOF'
version: '3.8'

services:
  fooocus:
    image: ghcr.io/lllyasviel/fooocus:latest
    container_name: fooocus
    restart: unless-stopped
    ports:
      - "7865:7865"
    volumes:
      - fooocus_models:/app/models
      - ./outputs:/app/outputs
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]

volumes:
  fooocus_models:
EOF

docker compose up -d

Access: http://your-server-ip:7865

Manual Installation

# Clone repository
git clone https://github.com/lllyasviel/Fooocus.git
cd Fooocus

# Install dependencies
pip install -r requirements.txt

# Launch
python entry_with_update.py --listen 0.0.0.0

First launch downloads models automatically (5-10 minutes).

---

Basic Usage

Generating Your First Image

Step 1: Open Fooocus web UI

Step 2: Enter a simple prompt:

a professional photograph of a sunset over mountains

Step 3: Click “Generate”

Step 4: Wait 10-30 seconds

That’s it! Fooocus handles:

• Model selection
• Sampling parameters
• Negative prompts
• Quality enhancements

Using Styles

Built-in Styles:

• Fooocus V2: Balanced default
• Fooocus Photograph: Photorealistic images
• Fooocus Anime: Anime/manga style
• Fooocus Cinematic: Movie-like quality
• SAI Photographic: Stability AI photo preset

Apply Style:

1. Select from “Style” dropdown
2. Generate as normal
3. Style automatically adjusts all parameters

Advanced Settings (Optional)

Click “Advanced” to reveal:

Performance:

• Speed: Faster (lower quality)
• Quality: Better (slower)
• Extreme Speed: SDXL Turbo (4 steps)

Aspect Ratios:

• Portrait, Landscape, Square
• Custom resolutions

Image Number:

• Generate 1-8 images per prompt
• Batch generation for variety

---

Image-to-Image

Transform existing images:

Step 1: Switch to “Image Prompt” tab

Step 2: Upload reference image

Step 3: Enter transformation prompt:

same scene but at night with neon lights

Step 4: Adjust “Stop At” slider:

• Low (0.3): Subtle changes
• Medium (0.5): Balanced
• High (0.7): Major transformation

Step 5: Generate

---

Inpainting & Outpainting

Inpainting (Edit Regions)

Step 1: Switch to “Inpaint or Outpaint” tab

Step 2: Upload image

Step 3: Method: “Inpaint”

Step 4: Draw mask over area to change

Step 5: Enter what to add:

a golden retriever sitting

Step 6: Generate

Outpainting (Extend Images)

Step 1: Method: “Outpaint”

Step 2: Choose direction (Left, Right, Top, Bottom)

Step 3: Fooocus automatically extends image

---

Model Management

Default Models

Fooocus downloads optimal models automatically:

• Base: juggernautXL_version6
• Refiner: sd_xl_refiner_1.0
• LoRAs: Quality enhancers

Custom Models

Add Your Own:

# Download to models directory
cd /opt/Fooocus/models/checkpoints/
wget https://civitai.com/api/download/models/XXX -O custom-model.safetensors

Select in UI:

• Advanced → Model → Select from dropdown

Presets

Save your favorite configurations:

Create Preset:

1. Configure all settings
2. Advanced → Preset → “Save as New”
3. Name it (e.g., “My Anime Style”)
4. Load anytime from Preset dropdown

---

Performance Optimization

GPU Optimization

Built-in Optimizations:

• Automatic mixed precision
• xformers memory attention
• Model offloading for low VRAM

Performance Modes:

Speed Mode:

Advanced → Performance → Speed

• 8-12 steps
• 20-30 seconds per image

Quality Mode:

Advanced → Performance → Quality

• 30-40 steps
• 45-60 seconds per image

Extreme Speed:

Advanced → Performance → Extreme Speed

• 4 steps (Turbo)
• 10-15 seconds per image

VRAM Requirements

---

Practical Examples

Portrait Photography

Prompt: professional headshot of a business executive, 
studio lighting, neutral background, sharp focus

Style: Fooocus Photograph
Aspect Ratio: Portrait
Performance: Quality

Landscape Art

Prompt: epic mountain landscape at golden hour, 
dramatic clouds, vibrant colors, highly detailed

Style: Fooocus Cinematic
Aspect Ratio: Landscape
Performance: Quality

Anime Character

Prompt: anime girl with blue hair, school uniform, 
cherry blossoms, detailed eyes, soft lighting

Style: Fooocus Anime
Aspect Ratio: Portrait
Performance: Speed

Product Photography

Prompt: professional product photo of smartwatch, 
white background, studio lighting, commercial quality

Style: SAI Photographic
Aspect Ratio: Square
Performance: Quality

---

When to Use Fooocus

Choose Fooocus When:

• You’re new to AI image generation
• You want quick, quality results without configuration
• You don’t need advanced control (ControlNet, complex workflows)
• You prefer simplicity over flexibility
• You’re generating images for personal use, not commercial production

Choose ComfyUI When:

• You need pixel-perfect control (ControlNet)
• You want reusable, shareable workflows
• You’re doing professional/commercial work
• You enjoy technical complexity
• You need custom pipelines (AnimateDiff, advanced upscaling)

Choose AUTOMATIC1111 When:

• You want a middle ground (more control than Fooocus, simpler than ComfyUI)
• You use many extensions
• You’re familiar with traditional SD interfaces
• You need features not in Fooocus (training, etc.)

---

Troubleshooting

Issue: Slow Generation

Solution:

1. Switch to Speed mode
2. Reduce resolution
3. Use Extreme Speed for testing
4. Check GPU is detected: nvidia-smi

Issue: Out of Memory

Solution:

# Restart with low VRAM mode
python entry_with_update.py --lowvram

Or reduce resolution to 512x512.

Issue: Poor Quality

Solution:

1. Switch to Quality mode
2. Add details to prompt
3. Try different style presets
4. Use negative prompt (Advanced)

---

Next Steps

Learning Path

Day 1: Generate 50+ images with different prompts Day 2: Experiment with all style presets Day 3: Try image-to-image transformations Day 4: Practice inpainting and outpainting Week 2+: Decide if you need more control (consider ComfyUI)

Related Guides

• ComfyUI: Advanced Workflows - For when you outgrow Fooocus
• AI GPU Selection Guide 2025

Guide Version: 1.0 Last Updated: December 2024 Difficulty: Beginner Time to Complete: 15 minutes