stable-diffusion text-to-image stable-diffusion-diffusers diffusers

Mitsua Diffusion One Model Card

Mitsua Diffusion One is a latent text-to-image diffusion model, which is a successor of Mitsua Diffusion CC0.

This model is trained from scratch using only public domain/CC0 or copyright images with permission for use, with using a fixed pretrained text encoder (OpenCLIP ViT-H/14, MIT License).

This will be used as a base model for AI VTuber Elan Mitsua🖌️’s activity.

❗❗ Currently, the model is still of low quality and lacks diversity ❗❗

Further training will be done fully opt-in basis.

If you are interested in, please click here to submit an opt-in application.

We are active on a Discord server for opt-in contributors only. Communication is currently in Japanese.

❗❗ To train this model, images from opt-in contributors have not yet been used ❗❗

Header

You can check here to all prompts to generate these images.

License

This model is open access and available to all, with a Mitsua Open RAIL-M license further specifying rights and usage. The Mitsua Open RAIL-M License specifies:

  1. You can't use the model to deliberately produce nor share illegal or harmful outputs or content
  2. The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license
  3. You can't use the model to infringe any rights of other by feeding image sources or model weights to the model (e.g. using another person's copyrighted image for fine-tuning without permission, using another person's copyrighted image as a source for image2image without permission).
  4. You can't misrepresent that a generated image as not AI-generated.
  5. You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the Mitsua Open RAIL-M to all your users (please read the license entirely and carefully) Please read the full license here

Training Data Sources

All data was obtained ethically and in compliance with the site's terms and conditions. No copyright images are used in the training of this model without the permission. No AI generated images are in the dataset.

Approx 11M images in total with data augmentation.

  1. Their work is released under a CC0 license, but if you are considering using this model to create a work inspired by their NFT and sell it as NFT, please consider paying them a royalty to help the CC0 NFT community grow.

Training Notes

Cosine similarity (as a proof of full-scratch training)

Developed by