Illustrious xl early release v0

Introduction

Illustrious XL is a text-to-image model focused on illustration, developed by the Onoma AI Research team. It is based on the Stable Diffusion XL model and fine-tuned using the Danbooru2023 dataset. The model is released under the fair-ai-public-license-1.0-sd, which discourages monetization or closed-source use.

Architecture

Illustrious XL builds upon the Kohaku XL Beta 5 base model, incorporating a large-scale annotated dataset to enhance its generative capabilities. The model supports diverse artistic and character design tasks, offering both a base model version (v0.1) and a guided variant with safety controls (v0.1-GUIDED).

Training

The Illustrious XL v0.1 is an untuned base model designed for future customization through LoRAs or adapters. The model is primarily intended for research purposes and is not optimized for specific aesthetic preferences. The guided variant includes additional safety mechanisms to ensure responsible content generation.

Guide: Running Locally

1. Set up the Environment:

   • Install the necessary dependencies, including Python and libraries such as PyTorch and Hugging Face Transformers.

2. Download the Model:

   • Clone the model repository from Hugging Face or download the model weights directly.

3. Run Inference:

   • Load the model and perform text-to-image generation using input prompts.

4. Hardware Recommendations:

   • For optimal performance, consider using cloud GPUs such as those provided by AWS, Google Cloud, or Azure.

License

The model is distributed under the fair-ai-public-license-1.0-sd, which requires users to:

• Share details of any derived models and datasets used.
• Prohibit monetization of closed-source derivatives.
• Ensure compliance with the outlined terms and conditions, including ethical use and open-source sharing of derivative works.