LICENSE

Copyright (c) 2024 ELAN MITSUA Project, Mitsua Likes Attribution-NonCommercial License dated December 16, 2024

The Mitsua Likes Attribution-NonCommercial License (hereinafter referred to as “the Terms”) sets forth the conditions 
for the use of the image generation AI model developed and released by the ELAN MITSUA Project / Abstract Engine Co., Ltd. 
(hereinafter referred to as “MITSUA Project”), known as “Mitsua Likes,” as well as derivative models (hereinafter “Derivative Models”), 
and any data generated using Mitsua Likes or any data generated using Derivative Models 
(hereinafter collectively referred to as “Generated Data”). Please read the Terms in full before using Mitsua Likes. 
Users of Mitsua Likes shall be deemed to have agreed to the Terms upon using Mitsua Likes. 
In the event that a separate written agreement is concluded, such agreement shall take precedence over the Terms.

License for Use of Generated Data and Attribution of Rights
- Use of the Generated Data is restricted to non-commercial purposes only. 
  However, commercial use shall be permissible within the scope defined in the Terms if:
    - The user is an individual or a team of up to 10 members using the Generated Data for their own creative activities, or
    - The user is a corporate entity that has obtained prior written consent from the MITSUA Project.
- All rights related to the Mitsua Likes model belong exclusively to the MITSUA Project or the respective rights holders 
  who have granted usage rights to the MITSUA Project. Except as explicitly set forth in the Terms, no such rights are transferred or assigned to the user.
- If a user generates content containing the name of a “Public Character” (a character independently designated 
  by the MITSUA Project and operated by the MITSUA Project or a third party who has granted the MITSUA Project the right to use the character), 
  or if the Generated Data is determined to be similar to such a character by a “Similarity Determination Model,” 
  the user shall use the Generated Data in accordance with the Terms and the terms of use applicable to the character in question.
- When publishing or sharing the Generated Data with third parties, users must display credits for Mitsua Likes. 
  The manner of displaying credits can be in any reasonable format, as long as it does not mislead others into believing that 
  the generation was conducted by the MITSUA Project, or any rights holder who has permitted the MITSUA Project to use their content. 
  If the user significantly modifies the Generated Data, credits must still be displayed with a reasonable supplementary explanation. 
  In the Terms, “credits for Mitsua Likes” means displaying one of the following statements. 
  In the case of Generated Data produced by a Derivative Model, it must be clearly indicated that a Derivative Model was used:
    - In the case of Mitsua Likes:
        - "画像生成:Mitsua Likes" (in Japanese)
        - "Generated by Mitsua Likes" (in English)
    - In the case of a Derivative Model (XXX = name of the Derivative Model):
        - "画像生成:XXX (Mitsua Likes派生)"
        - "Generated by XXX derived from Mitsua Likes"
- If the name or any related designation/information of the official character “Elan Mitsua” for which the MITSUA Project holds rights is added to credits, 
  explanatory text, or the like for the Generated Data, the content of the Generated Data must not violate the prohibitions 
  outlined in the “Elan Mitsua Fan Fiction Guidelines.”
- Pursuant to the Terms, the MITSUA Project grants users a perpetual, worldwide, non-exclusive, sublicensable, royalty-free, 
  and irrevocable license to reproduce, create derivatives of, publicly display, distribute, share, sublicense, 
  and otherwise distribute the Generated Data created by the user using Mitsua Likes.
- Except as provided in the Terms, the MITSUA Project does not claim any rights to the Generated Data
- The MITSUA Project may request a user to cease using the Mitsua Likes model or Derivative Models, and to cease using the Generated Data, 
  if such user’s use is found to be in violation of the Terms or if the Generated Data is produced as a result of violation of the Terms. 
  The user shall comply with such requests immediately.
- If a user permits a third party to use the Generated Data, the user must establish provisions that ensure compliance with the Terms.

Conditions for Distribution and Modification of Models
- Of the models constituting Mitsua Likes, no modifications are allowed to any part except for the Text Encoder.
- The Text Encoder is subject to the CC BY-SA 4.0 license, which takes precedence. 
  Provided that appropriate credit is given, any changes are indicated, and the adapted work is distributed under the same license, 
  the Text Encoder may be used, modified, and redistributed for both commercial and non-commercial purposes.
- All other models (U-Net, VAE, the Similarity Determination Model, and the similarity determination parameters) may not be modified. 
  However, if the user obtains prior written agreement from the MITSUA Project, the user may create a Derivative Model under conditions stipulated in writing.
- The use conditions for Derivative Models and the Generated Data produced by Derivative Models are subject to the Terms and shall be used under the same conditions.
- Hosting and redistributing the Mitsua Likes model or Derivative Models is permitted solely for non-commercial purposes 
  in completely non-profit, free-of-charge services or applications. 
  In such cases, the credits for Mitsua Likes must be displayed prominently on the relevant website or application.
- When redistributing the Mitsua Likes model or Derivative Models, users must clearly indicate any modifications made, 
  and include in the terms of use of any service that incorporates Mitsua Likes or a Derivative Model a provision requiring compliance with the Terms.
- If a user obtains prior written agreement from the MITSUA Project and prepares additional training data to create a Derivative Model 
  through fine-tuning or similar processes, such training data must be limited to data over which the user themselves holds complete rights, 
  data for which explicit permission has been granted through contractual agreement with the rights holder, public domain data, 
  or data distributed under an open license that allows modification and redistribution, and can be appropriately used.
- If a user obtains prior written agreement from the MITSUA Project and prepares additional training data to create a Derivative Model, 
  the user must conduct sufficient examinations to ensure that no data infringing third-party rights is included. 
  Furthermore, if the user publishes or shares the Derivative Model or any Generated Data produced from it, 
  the user must disclose details of the additional training data in a reasonable manner.

Prohibitions
Users of the Mitsua Likes model, Derivative Models, and any Generated Data produced by Mitsua Likes or by Derivative Models shall not engage in the following acts:
1. Hosting, redistributing, selling, or commercially using the Mitsua Likes model or a Derivative Model as part of a paid or for-profit service/application.
2. Using data that includes MITSUA Project’s or a third party’s copyrighted works or personal information as input data 
  or training data for the Mitsua Likes model or Derivative Models without the explicit permission of the rights holder.
3. When sharing Generated Data produced by a Derivative Model, failing to disclose details of the additional training data used.
4. Discriminating against, defaming, or insulting the MITSUA Project or a third party, thereby damaging their honor or credibility.
5. Infringing or potentially infringing the intellectual property rights or privacy of the MITSUA Project or a third party.
6. Disseminating information or content that unjustly harms the interests of the MITSUA Project or a third party.
7. Disseminating false information or content.
8. Modifying the VAE or the Similarity Determination Model of Mitsua Likes or a Derivative Model.
9. If Generated Data is produced by including the name of a Public Character, or is determined by the Similarity Determination Model to be 
  that of a Public Character, engaging in any act that violates the terms of use for that character.
10. When publishing or sharing Generated Data, failing to appropriately display credits for Mitsua Likes.
11. Displaying the Name, etc. of Elan Mitsua in the credits for Generated Data that violates the prohibitions of the Elan Mitsua Fan Fiction Guidelines.
12. Misrepresenting Generated Data as not being AI-generated.
13. Distributing or selling Generated Data as materials for profit-making purposes.
14. Registering or using Generated Data as the primary constituent element of trademarks, logos, symbol marks, image characters, avatars, or profile images.
15. Performing acts that are illegal, contravene public order and morals, or could be considered antisocial.
16. Creating a dataset composed primarily of Generated Data for the purpose of machine learning.
17. Using a dataset composed primarily of Generated Data for machine learning.
18. Using Generated Data as input for generative AI models, services, or applications developed or operated by the user or a third party.
19. Any act that violates the Terms.

No Warranty
- The Mitsua Likes model is provided “as is,” and the MITSUA Project makes no warranties whatsoever 
  regarding the safety, reliability, completeness, effectiveness, accuracy, or any other aspect of the Mitsua Likes model.
- The MITSUA Project does not guarantee, explicitly or implicitly, 
  that the Mitsua Likes model complies with applicable laws, does not infringe rights, can be used continuously, or is free from defects or other issues.

Liability
- The MITSUA Project shall not be liable for any direct or indirect losses arising from the use of the Mitsua Likes model, 
  the Generated Data of Mitsua Likes, or any Derivative Models or Generated Data produced by Derivative Models.
- In the event of a dispute with a third party arising from the use of the Mitsua Likes model, the Generated Data of Mitsua Likes, 
  or any Derivative Models or Generated Data produced by Derivative Models, the user shall resolve such dispute at their own responsibility and expense.
- If a user’s violation of the Terms causes damage to the MITSUA Project, the user shall compensate for such damages.

Language
- the Terms are written in Japanese, and the Japanese version shall be the authentic version. 
  Any English translation is provided solely for reference, and in the event of any inconsistencies or discrepancies 
  between the Japanese and English versions, the Japanese version shall prevail.

ELAN MITSUA Project / Abstract Engine