Xross Road Releases Its Own Image Generation Model 'HANASEE-image-1.0'
extra mile Co., Ltd. announced 'HANASEE-image-1.0', an image generation model specialized for vertical-reading manga panels. It features character and art style consistency and will be available on the 'HANASEE' platform beta from April 30, 2026.
📋 Article Processing Timeline
- 📰 Published: April 24, 2026 at 19:10
- 🔍 Collected: April 24, 2026 at 10:31
- 🤖 AI Analyzed: April 25, 2026 at 03:39 (17h 8m after Collected)
Image of HANASEE-image-1.0
The IP content creation and distribution project 'Xross Road', developed by extra mile Co., Ltd. (Headquarters: Minato-ku, Tokyo; CEO: Sakae Tamate), will release 'HANASEE-image-1.0' (hereafter, this model), an image generation model specialized for panel image generation of vertical-reading manga, under the supervision and cooperation of professional manga artists. This model will be available in the closed beta version of the manga creation platform 'HANASEE', which will begin offering services on April 30, 2026.
■ What is the manga creation platform 'HANASEE'?
'HANASEE' is a manga creation platform where you can create ideas and stories as vertical-reading manga. Based on novels or ideas that form the prototype of a story, it is possible to produce serialized manga with consistent characters, worldviews, and art styles through its unique image generation model and creative workflow.
■ What is HANASEE-image-1.0?
'HANASEE-image-1.0' is an image generation model developed by Xross Road under the supervision and cooperation of professional manga artists, optimized for expressing continuous story information as manga panel images. Based on an open-source image generation model, it has undergone proprietary improvements and additional training specifically for manga panel generation.
Features of this model
Character Consistency
Image of multi-panel generation maintaining character consistency (Generated with HANASEE-image-1.0)
Character consistency is indispensable for expressing the continuity of a story in a manga work. In this model, proprietary optimization is performed to generate character information (hairstyle, facial features, clothing, etc.) while maintaining identity in panel images depicting various scenes and situations along the story. This realizes continuous visual generation on a per-panel or per-episode basis, which was difficult with conventional general-purpose models.
Manga Panel-Specific Art Styles and Consistency
Image ① of manga panel images reflecting art styles (Generated with HANASEE-image-1.0)
Image ② of manga panel images reflecting art styles (Generated with HANASEE-image-1.0)
Regarding 'art style', which is an important element supporting a creator's authorship in continuing serialization, this model achieves both an art style expression unprecedented in generative AI and consistency on a per-panel basis through its unique approach. At the time of launch, in addition to 'Refreshing', 'Fluffy', 'Bright', and 'Cute', we will offer 'Powerful' and 'Cool' as optimized art styles. Furthermore, we plan to provide a feature in the future that brings uniqueness to the work's style by mixing and using these art styles.
Expression of Compositions Suitable for Vertical-Reading Manga
Achieved by the collaboration between the creative workflow and HANASEE-image-1.0 for optimal composition according to panel size and angle of view
Composition is as important as character consistency and art style in the panel expression of vertical-reading manga. This model supports flexible composition generation corresponding to specified panel sizes and expression content. Furthermore, the creative workflow of 'HANASEE' judges the optimal panel size and angle of view according to the story development, and this model generates the composition based on those instructions, thereby realizing the expression of the rhythm of the story unique to vertical-reading manga on a panel-by-panel basis.
■ How to Use HANASEE-image-1.0
Image generation by this model will be available through the panel image generation feature in the closed beta version scheduled to be released on April 30, 2026.
About Participating in the Closed Beta Version
Guidance for this closed beta version will be provided by lottery among those who have registered via the usage application form by April 27, 2026. If you are interested, please register using the form below.
Closed Beta Usage Application Form:
https://docs.google.com/forms/d/e/1FAIpQLScsxaFy97klHtycfM1q8NGfJ9Qg06ph0vEHzRhyVeGhFa9DwA/viewform?usp=header
Regarding Corporate Utilization
Companies considering the utilization of this model are requested to contact the email address below, attaching details of the anticipated collaboration.
E-mail: info@hanasee.com
■ Future Developments
'HANASEE' Pre-registration
Currently, 'HANASEE' is accepting pre-registrations for the official release. Those who register via the URL below will be preferentially informed about service update information and the release date.
The IP content creation and distribution project 'Xross Road', developed by extra mile Co., Ltd. (Headquarters: Minato-ku, Tokyo; CEO: Sakae Tamate), will release 'HANASEE-image-1.0' (hereafter, this model), an image generation model specialized for panel image generation of vertical-reading manga, under the supervision and cooperation of professional manga artists. This model will be available in the closed beta version of the manga creation platform 'HANASEE', which will begin offering services on April 30, 2026.
■ What is the manga creation platform 'HANASEE'?
'HANASEE' is a manga creation platform where you can create ideas and stories as vertical-reading manga. Based on novels or ideas that form the prototype of a story, it is possible to produce serialized manga with consistent characters, worldviews, and art styles through its unique image generation model and creative workflow.
■ What is HANASEE-image-1.0?
'HANASEE-image-1.0' is an image generation model developed by Xross Road under the supervision and cooperation of professional manga artists, optimized for expressing continuous story information as manga panel images. Based on an open-source image generation model, it has undergone proprietary improvements and additional training specifically for manga panel generation.
Features of this model
Character Consistency
Image of multi-panel generation maintaining character consistency (Generated with HANASEE-image-1.0)
Character consistency is indispensable for expressing the continuity of a story in a manga work. In this model, proprietary optimization is performed to generate character information (hairstyle, facial features, clothing, etc.) while maintaining identity in panel images depicting various scenes and situations along the story. This realizes continuous visual generation on a per-panel or per-episode basis, which was difficult with conventional general-purpose models.
Manga Panel-Specific Art Styles and Consistency
Image ① of manga panel images reflecting art styles (Generated with HANASEE-image-1.0)
Image ② of manga panel images reflecting art styles (Generated with HANASEE-image-1.0)
Regarding 'art style', which is an important element supporting a creator's authorship in continuing serialization, this model achieves both an art style expression unprecedented in generative AI and consistency on a per-panel basis through its unique approach. At the time of launch, in addition to 'Refreshing', 'Fluffy', 'Bright', and 'Cute', we will offer 'Powerful' and 'Cool' as optimized art styles. Furthermore, we plan to provide a feature in the future that brings uniqueness to the work's style by mixing and using these art styles.
Expression of Compositions Suitable for Vertical-Reading Manga
Achieved by the collaboration between the creative workflow and HANASEE-image-1.0 for optimal composition according to panel size and angle of view
Composition is as important as character consistency and art style in the panel expression of vertical-reading manga. This model supports flexible composition generation corresponding to specified panel sizes and expression content. Furthermore, the creative workflow of 'HANASEE' judges the optimal panel size and angle of view according to the story development, and this model generates the composition based on those instructions, thereby realizing the expression of the rhythm of the story unique to vertical-reading manga on a panel-by-panel basis.
■ How to Use HANASEE-image-1.0
Image generation by this model will be available through the panel image generation feature in the closed beta version scheduled to be released on April 30, 2026.
About Participating in the Closed Beta Version
Guidance for this closed beta version will be provided by lottery among those who have registered via the usage application form by April 27, 2026. If you are interested, please register using the form below.
Closed Beta Usage Application Form:
https://docs.google.com/forms/d/e/1FAIpQLScsxaFy97klHtycfM1q8NGfJ9Qg06ph0vEHzRhyVeGhFa9DwA/viewform?usp=header
Regarding Corporate Utilization
Companies considering the utilization of this model are requested to contact the email address below, attaching details of the anticipated collaboration.
E-mail: info@hanasee.com
■ Future Developments
'HANASEE' Pre-registration
Currently, 'HANASEE' is accepting pre-registrations for the official release. Those who register via the URL below will be preferentially informed about service update information and the release date.