SD Zoom and Smart CLI

This project contains two main scripts: sd_zoom.py for generating zoomed sequences of images using Stable Diffusion, and smart_cli.py for an interactive prompt generation and image creation experience.

Requirements

• Python 3.8+
• Stable Diffusion WebUI with API access enabled
• FFmpeg (for video compilation)
• Ollama (for smart_cli.py)

Installation

1. Clone this repository
2. Install the required Python packages:

   pip install webuiapi ollama
   

3. Ensure FFmpeg is installed on your system
4. Copy config_example.py to config.py and modify it according to your setup:

   cp config_example.py config.py
   

   Edit config.py to set your Stable Diffusion server details and other preferences.

Configuration

The config.py file contains various settings that you can customize:

• SD_SERVERS: List of available Stable Diffusion servers
• DEFAULT_SERVER: Index of the default server in the SD_SERVERS list
• Default values for negative prompts, steps, CFG scale, zoom factor, etc.
• HR Upscaler settings
• Ollama model for prompt generation

Modify these settings according to your preferences and setup.

SD Zoom Usage

python sd_zoom.py --prompt "Your prompt here" [options]

Options

• --host: Stable Diffusion WebUI API host (default: from config)
• --port: Stable Diffusion WebUI API port (default: from config)
• --prompt: Initial prompt for image generation (required)
• --negative_prompt: Negative prompt (default: from config)
• --seed: Seed for image generation (-1 for random, default: -1)
• --steps: Number of steps for image generation (default: from config)
• --cfg_scale: CFG scale for image generation (default: from config)
• --width: Width of generated images (default: 960)
• --height: Height of generated images (default: 640)
• --zoom_factor: Zoom factor between iterations (default: from config)
• --count: Number of images to generate in the sequence (default: 3)
• --denoising_strength: Denoising strength for img2img (default: from config)
• --output_prefix: Prefix for output filenames (default: "zoomed_image")
• --framerate: Framerate for the output video (default: from config)
• --sleep_between_frames: Sleep duration between frame generations (default: 0)
• --hr_upscaler: Use HR upscaler (flag, default: False)
• --hr_scale: HR upscaler scale factor (default: from config)

Smart CLI Usage

The Smart CLI provides an interactive experience for generating prompts and creating zoomed image sequences.

python smart_cli.py

Features

• AI-assisted prompt generation and evolution using Ollama
• Interactive user preferences for video length, resolution, and more
• Automatic prefix generation for output files
• Choice between configured Stable Diffusion servers
• Logging of the entire process

Process

1. The script displays example prompts for inspiration
2. You can describe your desired image or choose a random example
3. The AI generates an initial prompt based on your description
4. You can provide feedback to evolve the prompt iteratively
5. Once satisfied, you can set preferences for the video output
6. The script generates the zoomed image sequence and compiles a video

Output

Both scripts create a new folder for each run, containing:

1. A sequence of PNG images, each representing a step in the zoom process
2. An MP4 video file combining all the images into a smooth zoom animation
3. A log file (for smart_cli.py) detailing the entire process

Notes

• Adjust the parameters in config.py to control the default behavior of the scripts
• Experiment with different prompts and settings to achieve various effects
• The smart_cli.py script provides a more user-friendly and AI-assisted approach to creating zoom sequences

Example Prompts

Here are some example prompts to inspire your creations:

1. Mystical Forest:

   ((hyper-detailed)), lush forest, ancient trees, misty atmosphere, sunbeams filtering through leaves, {moss-covered rocks|gnarled roots}, hidden fairy houses, glowing mushrooms, ((depth of field)), ethereal lighting, 8k resolution
   

2. Futuristic Cityscape:

   ((ultra-realistic)), sprawling cyberpunk metropolis, neon-lit skyscrapers, flying vehicles, holographic billboards, {rain-slicked streets|bustling hover-trains}, intricate architectural details, moody atmospheric lighting, reflective surfaces, 8k resolution
   

3. Underwater Coral Reef:

   ((photorealistic)), vibrant coral reef, diverse tropical fish, {sea turtles|manta rays}, shimmering water caustics, schools of colorful fish, intricate coral formations, sunlight filtering through water, underwater flora, crystal clear water, deep ocean background, 8k resolution
   

4. Vast Desert Landscape:

   ((highly detailed)), expansive desert, towering sand dunes, {ancient ruins|mysterious oasis}, dramatic shadows, heat haze, distant mountains, weathered rock formations, lone nomad figure, golden hour lighting, wispy clouds, 8k resolution
   

5. Fractal Vine Mosaic:

   ((intricate details)), abstract fractal vines, {emerald|ruby|sapphire} mosaic tiles, spiraling patterns, ((infinite recursion)), gleaming metallic accents, {sacred geometry|mandelbrot set}, iridescent color shifts, flowing organic shapes, 8k resolution
   

Feel free to use these prompts as starting points and modify them to suit your creative vision!

Troubleshooting

• If you encounter connection errors, ensure your Stable Diffusion WebUI is running and the API is enabled
• For Ollama-related issues, check that Ollama is installed and running correctly on your system
• If the video compilation fails, make sure FFmpeg is properly installed and accessible from the command line

Contributing

Contributions to improve the scripts or add new features are welcome! Please submit a pull request or open an issue to discuss proposed changes.

License

This project is licensed under the MIT License - see the LICENSE file for details.