Skip to content
かみもと · 本地AI视频工作流·· 22 days agoAI score56

MOSS-Image generates manga single pages and spreads, sharing prompt techniques for maintaining character consistency

【MOSS-Image】漫画1ページと見開きを描いてみた!一貫性を保つプロンプトの工夫あり

AI briefing

The author used the local image generation model MOSS-Image to generate black-and-white manga single pages and spreads, and shared three prompt techniques for maintaining character consistency: fixing the character description text, declaring the characters overall, and embedding the same description in each panel. As a result, the two characters remained distinct to the end, and the large spread panel was also successful; the text was mostly accurate, but complex Chinese characters were occasionally miswritten, and unspecified sound effects were added.

Open original note.com