JSAI2024

Presentation information

Poster Session

Poster session » Poster session

[3Xin2] Poster session 1

Thu. May 30, 2024 11:00 AM - 12:40 PM Room X (Event hall 1)

[3Xin2-100] An Image Generation User Interface for Supporting Manga Character Design

〇Shun Kitagawa1, Taro Hatakeyama1, Komei Hiruta1, Atsushi Hashimoto2,1, Satoshi Kurihara1 (1.Keio University, 2. OMRON SINIC X Corporation)

Keywords:Image Generation, User Interface, Manga

We propose an image generation system, aiming to support creators achieving manga character designs beyond their individual creative capacity. Typically, creators design manga characters based on story plots. Such plots provide various conditions for the target character design. Therefore, in our context of manga character generation, the system is requested to have both controllability and unexpectedness in generated results. Most existing user interfaces (UIs) for image generation require creators to input conditions by text. Such UI generates results only based on a creator's idea and random seeds, which hardly yield unexpected elements in generated characters. We propose a UI with tag-based input to address this issue. Users can control the results by selecting tags, while the system adds random tags to lead to unexpected results. We have conducted user tests on the system with professional creators. The results show that they designed a character beyond their individual creative capacity.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password