Fooocus

Easy Stable Diffusion image generation with Fooocus on Clore.ai

Generate images with Fooocus - the easiest way to use Stable Diffusion.

All examples can be run on GPU servers rented through CLORE.AI Marketplace.

Renting on CLORE.AI

1. Visit CLORE.AI Marketplace

2. Filter by GPU type, VRAM, and price

3. Choose On-Demand (fixed rate) or Spot (bid price)

4. Configure your order:

   • Select Docker image

   • Set ports (TCP for SSH, HTTP for web UIs)

   • Add environment variables if needed

   • Enter startup command

5. Select payment: CLORE, BTC, or USDT/USDC

6. Create order and wait for deployment

Access Your Server

• Find connection details in My Orders

• Web interfaces: Use the HTTP port URL

• SSH: ssh -p <port> root@<proxy-address>

What is Fooocus?

Fooocus is a streamlined Stable Diffusion interface that:

• Requires zero configuration

• Uses SDXL by default

• Has built-in styles and presets

• Handles all optimizations automatically

Requirements

Quality	Min VRAM	Recommended
Basic	4GB	RTX 3060
Standard	8GB	RTX 3070
High Quality	12GB+	RTX 3090/4090

Quick Deploy

Docker Image:

pytorch/pytorch:2.11.0-cuda12.8-cudnn9-devel

Ports:

22/tcp
7865/http

Command:

apt-get update && apt-get install -y git libgl1 libglib2.0-0 && \
cd /workspace && \
git clone https://github.com/lllyasviel/Fooocus.git && \
cd Fooocus && \
pip install -r requirements_versions.txt && \
python launch.py --listen 0.0.0.0 --port 7865

Accessing Your Service

After deployment, find your http_pub URL in My Orders:

1. Go to My Orders page

2. Click on your order

3. Find the http_pub URL (e.g., abc123.clorecloud.net)

Use https://YOUR_HTTP_PUB_URL instead of localhost in examples below.

First Launch

On first run, Fooocus automatically downloads:

• SDXL base model (~6.5GB)

• SDXL refiner (~6GB)

• Required embeddings

This takes 10-15 minutes on first launch.

Using Fooocus

Basic Generation

1. Open http://<proxy>:<port>

2. Enter your prompt

3. Click "Generate"

That's it! No settings needed.

Styles

Fooocus includes 200+ built-in styles:

Popular Styles:

• Fooocus Enhance - Better details

• Fooocus Sharp - Crisp edges

• Cinematic - Movie look

• Anime - Japanese animation

• Photographic - Realistic photos

Quality Presets

Preset	Speed	Quality
Speed	Fast	Good
Quality	Medium	Great
Extreme Speed	Fastest	Basic

Advanced Features

Enable Advanced Mode

Click "Advanced" checkbox to access:

• Negative prompts

• Aspect ratios

• Image count

• Random seed control

Image-to-Image

1. Enable "Input Image" tab

2. Upload source image

3. Choose mode:

   • Upscale - Enhance resolution

   • Vary - Create variations

   • Inpaint - Edit parts

Inpainting

1. Upload image
2. Click "Inpaint or Outpaint"
3. Draw mask on areas to change
4. Describe what to generate
5. Click Generate

Outpainting

Extend images beyond borders:

1. Upload image

2. Select "Inpaint or Outpaint"

3. Check direction boxes (Left, Right, Top, Bottom)

4. Generate to extend canvas

Using LoRAs

Download LoRAs

cd /workspace/Fooocus/models/loras
wget https://civitai.com/api/download/models/<model_id> -O my_lora.safetensors

Apply LoRA

1. Go to "Model" tab

2. Select LoRA in dropdown

3. Adjust weight (0.5-1.0)

Custom Models

Add Custom Checkpoints

cd /workspace/Fooocus/models/checkpoints

# Download model
wget https://huggingface.co/model/file.safetensors

Refresh UI or restart to see new models.

Recommended Models

Model	Style	Size
Juggernaut XL	Photorealistic	6.5GB
DreamShaper XL	Artistic	6.5GB
RealVisXL	Realistic	6.5GB
Animagine XL	Anime	6.5GB

Face Swap

Built-in face swap feature:

1. Enable "Image Prompt" tab

2. Upload face image

3. Set "FaceSwap" as type

4. Generate with face prompt

Upscaling

Built-in Upscaler

1. Upload image to "Upscale or Vary"

2. Select "Upscale (2x)"

3. Generate

Vary Options

• Vary (Subtle) - Small changes

• Vary (Strong) - Significant changes

Describe Image

Reverse prompt engineering:

1. Go to "Describe" tab

2. Upload image

3. Get prompt suggestions

Performance Optimization

For 8GB VRAM

python launch.py --listen 0.0.0.0 --always-offload-from-vram

For 6GB VRAM

python launch.py --listen 0.0.0.0 --always-low-vram

For 4GB VRAM

python launch.py --listen 0.0.0.0 --always-cpu

Batch Processing

Generate Multiple Images

In Advanced mode:

• Set "Image Number" to desired count

• All images generate with different seeds

Prompt Queue

Use wildcards for variations:

a {red|blue|green} car on the street

Generates 3 images with different colors.

API Access

Enable API

python launch.py --listen 0.0.0.0 --port 7865

API Endpoint

import requests

response = requests.post(
    "http://localhost:7865/v1/generation/text-to-image",
    json={
        "prompt": "a beautiful sunset over mountains",
        "negative_prompt": "",
        "style_selections": ["Fooocus Enhance"],
        "performance_selection": "Quality",
        "aspect_ratios_selection": "1024×1024",
        "image_number": 1,
        "image_seed": -1,
    }
)

# Get image
image_data = response.json()

Presets

Create Custom Preset

Edit presets/default.json:

{
    "default_model": "juggernautXL.safetensors",
    "default_refiner": "None",
    "default_loras": [
        ["detail_lora.safetensors", 0.5]
    ],
    "default_styles": ["Fooocus Enhance", "Cinematic"]
}

Launch with Preset

python launch.py --preset anime

Comparison: Fooocus vs Others

Feature	Fooocus	A1111	ComfyUI
Setup	None	Medium	Complex
Learning Curve	Easy	Medium	Hard
Flexibility	Low	High	Highest
Best For	Beginners	Power users	Developers

Troubleshooting

Out of Memory

# Use low VRAM mode
python launch.py --always-low-vram

# Or disable refiner

# In UI: Model tab > Refiner > None

Slow Generation

• Enable "Extreme Speed" preset

• Disable refiner

• Use smaller resolution

Model Not Loading

# Check model exists
ls -la /workspace/Fooocus/models/checkpoints/

# Check file size (should be ~6GB for SDXL)
du -h /workspace/Fooocus/models/checkpoints/*.safetensors

Black Images

• Reduce CFG scale

• Change seed

• Try different prompt

Download Results

# Images are in outputs folder
scp -P <port> -r root@<proxy>:/workspace/Fooocus/outputs/ ./

Cost Estimate

Typical CLORE.AI marketplace rates (as of 2024):

GPU	Hourly Rate	Daily Rate	4-Hour Session
RTX 3060	~$0.03	~$0.70	~$0.12
RTX 3090	~$0.06	~$1.50	~$0.25
RTX 4090	~$0.10	~$2.30	~$0.40
A100 40GB	~$0.17	~$4.00	~$0.70
A100 80GB	~$0.25	~$6.00	~$1.00

Prices vary by provider and demand. Check CLORE.AI Marketplace for current rates.

Save money:

• Use the Spot market for interruptible work — about a third of servers price spot below on-demand (median ~13% off), the rest match it

• Pay with CLORE tokens

• Compare prices across different providers

Next Steps

• Stable Diffusion WebUI - More control

• ComfyUI Workflows - Node-based

• FLUX Generation - Newer model