Skip to content

ファッション業界に特化した生成AI活用支援ツール「Maison AI」 写真風やイラスト風など好みの画像が作れる機能をはじめ、新機能追加でさらに使いやすくアップデート! ーChatGPTモデルも最新のGPT-4 Turboを搭載ー

OpenFashion Co., Ltd. (Headquarters: Minato-ku, Tokyo, hereinafter referred to as OpenFashion), which develops services and products that combine fashion with the latest technology including AI, has developed Maison, a generative AI utilization support tool specialized for the fashion industry. We have announced a feature update for AI (Maison AI).
Maison AI is currently in a trial period and can be used for free until the end of January 2024. Please see the release text for details.

OpenFashion Co., Ltd. is working on developing services and products that combine fashion with the latest technology, including AI.

The service ``Maison AI'', whose beta version was launched in August of this year, is an AI tool for text and image generation specialized in the fashion industry. In October, "Image Generate", which generates images from text, was added as a new function, and a free trial is being offered until the end of January 2024, and many people are already using it.

This time, in order to let more people know about the appeal of generation AI and how it can be used to improve business efficiency by using Maison AI, we have updated the functions of image generation and ChatGPT installed in Maison AI. See below for details.

  • New feature ①: Image style function

We have added an "image style function" that allows you to generate images in your preferred style by simply specifying a style such as "anime," "photo," "pixel art," "origami," or "3D model style" when generating images.

[Installed style example]

・Normal : Standard image style

・Photographic : A style characterized by photo-like reality

・Pixel art : Retro style created using pixels

・Analog film : A style with a texture that looks like it was shot with a film camera.

Equipped with a total of 18 style functions including normal.

While using the image generation function, many users have expressed their desire to create what they envisioned more easily, such as ``I wanted to create a more realistic photo-like image,'' or ``I wanted to generate an image that looked like an illustration.'' It was submitted. With the image style function, by simply specifying the style when generating an image, it is possible to more accurately generate expressions and image atmospheres that were previously difficult for users to prompt (instructions input by the user). It's now possible.


An example of specifying a photo, origami, pixel art, 3D model, and image style from above and generating it with the prompt "pink cat"

  • New feature ②: Image size specification function

With the addition of the image style feature, you can now also select the image size.

Until now, the only size that could be used to generate images was a square of 1024px x 1024px, but you can now select from a total of 9 sizes to generate images, such as 640px x 1536px vertically or 1536px x 640px horizontally depending on the purpose of use. receive.


Output at 1024px x 1024px (top left), 640px x 1536px (top right), and 1536px x 640px (bottom) and reduce without changing the ratio.

Since adding the image generation function in October of this year, many users have been using this function. In the future, we plan to update the functions related to image generation, such as the ability to generate a model's face image and replace part of the image. Please look forward to it.

  • Model update: ChatGPT model is now the latest version

With the model update of ChatGPT, you can now enjoy faster and more accurate conversations with AI.
・Newly installed model

ChatGPT3.5: ChatGPT3.5 gpt-3.5-turbo-1106

ChatGPT4.0:ChatGPT4.0 Turbo gpt-4-1106-preview

Both are the latest models announced in early November 2023 (6th and 7th local and Japan time).

This has made it possible to read files with larger amounts of data than before, and it has also become possible to input and output more characters in a single chat.

Furthermore, in the previous model, the AI ​​retained information up to 2021, but in the latest model, the AI ​​has updated the information to April 2023, so it can also answer questions that could not be answered before. Some of the questions can be answered correctly.

The free trial of Maison AI is scheduled to run until the end of January next year.

Please take advantage of this free opportunity. For more information, please see the service website below.

  • About free trial

・Trial offer period: End of January 2024

・Target functions: Workspace creation, member invitation, AI chat, AI agent, template, image generation, history

・Target AI model

[Text AI model] ChatGPT3.5 / ChatGPT3.5 - 16K /ChatGPT4.0

[Image generation model] Stable Diffusion XL 1.0

・Maximum usage: We plan to set a certain upper limit for each workspace, such as the number of participants.

・MaisonAI service site

・MaisonAI guide page (notion)

You can see more detailed instructions on how to use MaisonAI.

  • “Maison AI” function introduction

◆Image generation function

This is a new feature released in October of this year. With the "image generation function", it is now possible to generate images from sentences (text) on Maison AI.

The new function "Image generation function" can be used in the following situations.

・Image image of planning materials

・Fashion design images

・Material images for websites, blogs, e-mail newsletters, web advertisements, etc.

◆AI agent function

By setting a specific job type to the AI, you can get more accurate answers even for questions that require prerequisite knowledge. Maison AI, which specializes in the fashion industry, has AI agents for not only human resources and legal affairs, but also fashion designers and pattern makers.

*Prompts for each AI agent are posted on the sample site.

Other functions - AI chat: You can freely chat with AI just like ChatGPT. Simply enter the characteristics of a product using words such as "dress" or "pastel color," and it will generate a product description.

・Template: By setting a specific task, you can use it quickly without having to input it every time.

  • “Maison AI” planned additional functions

◆In painting function

We are also planning to add a function that allows you to generate a model's face image and replace part of the image. With this planned feature, you can easily create various images of the same outfit, such as changing the model's face or showing items such as belts and sunglasses.

*Functional contents are subject to change.

We are also currently holding a contest that utilizes Maison AI. See below for details.

  • Currently holding “Maison AI Prompt Contest”

Currently, OpenFashion is holding the "Maison AI Prompt Contest" to devise text-generating AI prompts that will help companies improve their business efficiency.

We also offer supplementary prizes for prompts that can be expected to improve business operations. Please feel free to submit prompts that you use every day or new ideas. For more information, please see the OpenFashion event page below.

・Maison AI Prompt Contest Overview

*Due to the extension of the event period, we are accepting applications until December 31st.

At OpenFashion, we will continue to promote the use of generative AI in the fashion industry to improve efficiency and create an environment where humans can focus their time where they should.