Introduction

The "OrangeMixs" repository offers a collection of Merge models compatible with the StableDiffusionWebui:Automatic1111 platform and others. It primarily serves the Japanese community by providing easy access to frequently used models and a place for the creator to upload personal merge models.

Architecture

OrangeMixs includes various models such as VividOrangeMix, AbyssOrangeMix, EerieOrangeMix, and more. These models are designed to produce vivid, anime-style illustrations with different levels of realism and anatomical accuracy. The models are created using the U-Net Blocks Weight Merge method, which allows for detailed control over model content without affecting design.

Training

The models in OrangeMixs are trained through complex merging techniques involving multiple base models to enhance specific features such as realism in textures or anime-style illustrations. They incorporate elements from various pre-trained models like NovelAI and Gape, adjusted for specific artistic outputs.

Guide: Running Locally

To run the OrangeMixs models locally, follow these basic steps:

  1. Install Git: Ensure Git is installed on your system.
  2. Clone Repository: Execute git clone https://huggingface.co/WarriorMama777/OrangeMixs in your terminal.
  3. Sparse Checkout (Optional): For specific directories, use:
    • git sparse-checkout init --cone
    • git sparse-checkout add <directory_name>
  4. Checkout Main: Finalize with git checkout main.

Suggested Cloud GPUs

For intensive tasks, consider using cloud GPUs from providers like AWS, Google Cloud, or Paperspace.

License

The models are distributed under the CreativeML OpenRAIL-M license, which allows for open access and commercial use with conditions:

  1. Do not produce or share illegal/harmful content.
  2. The authors claim no rights over generated outputs.
  3. Redistribution and commercial use require including the same use restrictions and providing the license to users.

For the full license text, visit: CreativeML OpenRAIL License.

More Related APIs in Text To Image