News
Please click here for latest announcements.
Technology Overview
Image generation using AI, which has become a hot topic in recent years, generally takes input commands called “prompt”. It is a one-to-one process between human and AI, or one could say that the human is giving orders to the AI. Mitsui-chan is based on the concept of "art created with AI and everyone" and will gradually create artworks by taking direction from viewers' comments on YouTube. AI Vtuber Mitsua-chan aims to create an unprecedented participatory art experience where people create art together through VTuber!
An ethical issue is pointed out as to recent AI image generation that copyrighted works have been incorporated into training data sets without permission, even though there are no legal problems in collection of data in such ways. Mitsua-chan doesn’t use the existing trained Stable Diffusion models as they are, but is trained from scratch with public domain or properly licensed images to challenge to improve the ethical issues for artists by giving back to the providers of the images on which the model is trained!
How Training Works
Mitsua-chan's AI is based on the "latent diffusion model" mechanism, and there are three stages in the training process.
-
STEP 1
In the first phase, the model is trained from scratch only with over 11 million images which are public domain/CC0 or licensed with appropriate permission for Mitsua-chan to study the basic concepts of art in an ethical manner. This trained model is published as Mitsua Diffusion One. Please note that this model focuses on solving ethical problems and does not have sophisticated artistic expression yet.
-
STEP 2
In the second step, images associated with the Twitter hashtag "#みつあ勉強用" text tweets will be collected from the Twitter accounts of the opt-in target accounts and added to the training data set. To apply for opt-in, follow Mitsua-chan's Twitter account and select "opt-in" from the application form here. After that, Mitsuachan's account will follow the target account, and our staff will confirm your identity via DM. The account whose identity has been verified will be eligible to opt-in. However, accounts whose content is deemed inappropriate by the management, such as content offensive to public order and morals, unauthorized reproduction, or posting images that infringe on the copyrights of others, will not be eligible for opt-in. We will not collect images from accounts other than opt-in eligible accounts. These opt-in accounts will also be eligible for returns. You may also change to opt-out afterwards. Learning after the second stage will be done sequentially from scratch, and no residual effects of excluded data will remain in the learning model.
-
STEP 3
Finally, in the third phase, images from those who offer their images to be included in the data set, specifically by tweeting with the hashtag "#Mitsua_ for_ study", are incorporated into the training data set. Those images will be used multiple times in the final stage of training and will have a significant impact on Mitsua Elan's learning of a drawing style and forming of an artistic concept. Those who provide images to Elan Mitsua's data set by tweeting with the hashtag will be rewarded according to the number of images they contribute to Elan Mitsua's painting training. There are several rules to follow on submission of images, so please refer to the Training Image Guidelines for details.
Elan Mitsua’s learning process
How to participate in Elan Mitsua’s learning
How YouTube streaming works
Elan Mitsua's painting is performed interactively through YouTube streaming.
While discussing the theme of the work or deciding the composition style in real time together with the audience through YouTube comments, the picture will be painted little by little.
Completed works will be released under the CC0 1.0 Universal (CC0 1.0) Public Domain Dedication license. However, if the work contains any derivative elements, the license of the published work will be subject to the respective official derivative guidelines. Details of the streaming and the first streaming schedule will be announced on our SNS account.