PXD Image Generator is a Streamlit-based application that allows you to generate AI images from text descriptions using the Stable Diffusion v1.5 model. This app is designed to give users full control through options like positive/negative prompts, number of inference steps, guidance scale (CFG), and random or manual seed selection.
- Interactive Interface: Enter positive and negative prompts to guide image results according to your preferences.
- Advanced Settings:
- Number of Inference Steps: Control image detail by adjusting the number of diffusion steps.
- Guidance Scale (CFG Scale): Set how closely the model follows the given prompt.
- Seed: Choose between a random seed for variation or a manual seed for reproducibility.
- Automatic GPU/CPU Support: Automatically detects and utilizes GPU if available, or falls back to CPU with appropriate settings.
- Model Caching: Uses
@st.cache_resourceto load the model once and improve efficiency.
Positive Prompt:
snow-capped mountain range at night reflecting a vibrant aurora borealis, long exposure, ethereal lighting, sense of wonder and tranquility
Negative Prompt:
low quality, blurry, ugly, distorted, bad anatomy, deformed, text, watermark, extra fingers, malformed hands
-
Clone the Repository:
git clone https://github.com/Pixel4bit/pxd-image-generator.git cd pxd-image-generator -
Create and Activate a Virtual Environment:
python -m venv venv source venv/bin/activate # For Unix or MacOS venv\Scripts\activate # For Windows
-
Install Dependencies:
pip install -r requirements.txt
-
Run the Streamlit App:
streamlit run streamlit_app.py
- Python: Version 3.8 or later
- Core Dependencies:
streamlitdiffuserstransformersacceleratescipytorchPillowinvisible_watermark
- Hardware:
- GPU with CUDA support (optional but recommended for optimal performance)
pxd-image-generator/
├── streamlit_app.py # Main Streamlit app script
├── requirements.txt # Python dependencies list
└── README.md # Project documentation
This project is licensed under the Apache 2.0 License.
Contributions are welcome! Feel free to open an issue or pull request for improvements, new features, or discussions.
Developed by Pixel4bit (pianxd). For questions or suggestions, feel free to reach out via GitHub.