This project is a script for generating images from textual prompts using the Stable Diffusion model. It allows for detailed customization through a YAML configuration file and provides a flexible command-line interface for easy use. This is preset to work on M series of macbooks.
The script generate_image.py utilizes the Stable Diffusion model from Hugging Face's Diffusers library to generate images based on textual descriptions. It offers various parameters like the number of steps, guidance scale, and image dimensions to fine-tune the generation process. Users can input their desired settings through a YAML configuration file, making it easy to experiment with different prompts and settings.
- Generate images from text prompts
- Customizable settings through a YAML file
- Adjustable image dimensions, step count, and guidance scale
- Command-line interface for easy interaction
Current model configures is https://civitai.com/models/25694?modelVersionId=143906
Ensure you have Python 3.7+ installed on your system. Then, follow these steps to set up the project.
-
Clone the Repository:
git clone https://aahmed-se/your-username/generate_image.git cd generate_image -
Install Dependencies:
pip install -r requirements.txt
-
Usage
Prepare Your Configuration: Edit the
config.yamlfile to specify your image generation parameters and text prompts.Run the Script:
python generate_image.py
Using a custom config file
python generate_image.py --config /path/to/your_config.yaml
-
Customization
You can modify the config.yaml to change the generation parameters. Available settings:
prompt: The textual description of the image you want to generate.negative_prompt: Descriptions of what you want to avoid in the image.steps: Number of inference steps.scale: Guidance scale for the generation.seed: Random seed for reproducibility.height: Height of the generated image.width: Width of the generated image.sampler: The type of sampling algorithm to use.