Comic Diffusion

Introduction

Comic-Diffusion is a text-to-image model designed to generate comic-style images using various art styles. It allows creators to easily produce unique and consistent comic art by mixing different style tokens. This model was developed to facilitate comic project creation with flexibility and ease.

Architecture

Comic-Diffusion is built on the StableDiffusionPipeline and is part of the Diffusers library. It supports text-to-image generation by combining multiple art style tokens, allowing for creative experimentation with the order and combination of these tokens.

Training

The model has two versions:

• V2: Trained on six distinct art styles: charliebo, holliemengert, marioalberti, pepelarraz, andreasrocha, and jamesdaly. This version provides versatility by mixing any number of these tokens, influencing the generated results.
• V1: Trained solely on the James Daly art style, using the token "comicmay artstyle."

Artists involved in these styles are not affiliated with the model.

Guide: Running Locally

To run Comic-Diffusion locally, follow these steps:

1. Clone the Repository: Obtain the model files from the Hugging Face repository.
2. Install Dependencies: Ensure that Python and necessary libraries such as torch and diffusers are installed.
3. Load the Model: Use the StableDiffusionPipeline from the Diffusers library to load the model.
4. Generate Images: Input text prompts to produce images in the desired comic styles.

For optimal performance, consider using cloud GPUs from providers like AWS, Google Cloud, or Azure to handle the computational requirements effectively.

License

Comic-Diffusion is released under the CreativeML OpenRAIL-M license, allowing for creative and flexible use while adhering to the specified terms.