J W S T Deep Space diffusion

Introduction

JWST Deep Space Diffusion is a fine-tuned Stable Diffusion model based on version 1.5. It is trained on images captured by the James Webb Space Telescope and contributions from Judy Schmidt. The model is designed for generating text-to-image content using the token "JWST" to achieve a specific style.

Architecture

The model is based on the Stable Diffusion framework, leveraging the capabilities of text-to-image synthesis. It can be used with the Diffusers library for generating images. The model supports export to various formats like ONNX, MPS, and FLAX/JAX for optimization purposes.

Training

The model was trained using the DreamBooth technique and TheLastBen Colab notebook. It incorporates images from the JWST and is fine-tuned to produce high-quality space-themed imagery.

Guide: Running Locally

1. Clone Repository: Obtain the model files from the Hugging Face repository.
2. Install Dependencies: Ensure you have the Diffusers library installed, along with any other necessary packages.
3. Load Model: Use the provided CKPT file to load the model into your environment.
4. Generate Images: Input prompts using the "JWST" token to create images. Adjust settings such as steps, sampler, and CFG scale for desired outputs.
5. GPU Recommendation: For optimal performance, it is recommended to use cloud GPUs like those provided by AWS or Google Cloud.

License

The JWST Deep Space Diffusion model is distributed under the CreativeML OpenRAIL-M license. Key points include:

• Prohibition against generating illegal or harmful content.
• The authors do not claim rights over generated outputs, which must comply with the license terms.
• Redistribution and commercial use are allowed with the inclusion of license terms and restrictions for users. The full license can be accessed here.