Venice AI Image Generator

Venice is a command-line tool for generating AI images using the Venice.ai API. It provides extensive customization options and manages image generation queues automatically.

Author

Original work by Hunter Rose (@HunterR0se)

Installation

1. Create a .venice directory in your home folder:

mkdir ~/.venice

2. Two configuration files are needed in the ~/.venice directory:
   • prompt.json (configuration settings)
   • elements.json (customization elements)

The application will create template versions of these files if they don't exist.

Configuration

API Key

Go to Visit Venice.AI and select "API" to generate an API key. Only available to paying members.

You'll need to add your Venice.ai API key to prompt.json.

If prompt.json does not exist, you will be prompted to provide your API key in the terminal, then a new file will be generated with the pre-sets below:

{
    "model": "fluently-xl",
    "prompt_name": "Hooded Hacker",
    "name_as_subdir": true,
    "prompt": "a modern hacker wearing a hoodie",
    "negative_prompt": "blur, distort, distorted, blurry, censored, censor, pixelated",
    "output_dir": "/home/YOURHOME/Pictures/venice",
    "api_key": "YOUR_API_KEY",
    "num_images": 42,
    "min_config": 7.5,
    "max_config": 15.0,
    "height": 1280,
    "width": 1280,
    "steps": 35,
    "style": true,
    "enable_face": true,
    "enable_type": true,
    "enable_clothing": true,
    "enable_poses": true,
}

Key Settings

• Model: Available models include:

  • fluently-xl (default, fastest)
  • flux-dev (highest quality)
  • flux-dev-uncensored
  • pony-realism (most uncensored)
  • lustify-sdxl (too sexy 4 u ..or maybe gross ...probably)
  • stable-diffusion-3.5 (most creative - supposedly)

• NumImages: How many images to generate

• Width/Height: Image dimensions (default 1280x1280)

• Steps: Generation steps (5-50, default 35)

• OutputDir: Where generated images are saved

Feature Toggles

Enable/disable specific enhancement categories:

{
    "EnableFace": true,
    "EnableType": false,
    "EnableHair": false,
    "EnableEyes": false,
    "EnableClothing": true,
    "EnableBackground": false,
    "EnablePoses": false,
    "EnableAccessories": false,
    "EnableDirty": false
}

Output

• Generated images are saved to the specified OutputDir (default: ~/Pictures/venice).
  • If name-as-subdir is true, a new subfolder is created for each run in the main output folder.
• Filenames include the image iteration, seed, cfg scale.
• Promt enhancements are recorded for each image in a PromptLog.txt file. Each entry records filename, image style, and added elements.
• Progress display shows:
  • Completion percentage
  • Current status
  • Active prompt
  • Model & configuration
  • Feature toggle states
  • Error status

Rate Limiting

The application automatically handles rate limiting:

• 2 second delay between generations
• Automatic retries on errors
• Graceful handling of API limits

Error Handling

• Failed generations are tracked and displayed
• Error messages appear in the progress display and get recorded to the PromptLog.txt file
• Auto-retry for common errors (rate limits, server issues)
  • Also attempts to mitigates "bad request" spamming in the event if errors caused by something in a prompt.

Usage

1. Ensure your API key is set in ~/.venice/prompt.json
2. Run the application:

./venice

3. Monitor progress in the terminal display
4. Use Ctrl+C to gracefully stop generation

Customization

The elements.json file contains categorized prompt elements for:

• Styles
• Face features
• Character types
• Hair styles
• Eye colors
• Clothing
• Poses
• Accessories
• Backgrounds

Edit these categories to customize the available elements for generation.

Tips

• Keep prompts under 1250 characters
• Monitor the error display for issues
• Use Ctrl+C for clean shutdown
• Check output directory for generated images
• Adjust cfg_scale (7.5-15.0) to control generation stability

Requirements

• Venice.ai API key
• Write access to ~/.venice directory
• Sufficient disk space for output