• 목록

Generative AI is increasingly used to create game character images, backstories, and gameplay ideas. However, effective game character design requires coherence across identity, narrative, visual semiotics, gameplay mechanics, and intended player perception. This study examines cross-dimensional consistency mismatches in AI-assisted game character design and translates a diagnostic taxonomy into a UI concept. Based on ten pilot cases, expert annotation, overlap comparison, and adjudication, we developed a revised codebook consisting of five design dimensions, ten mismatch types, severity criteria, and justification rules. We then proposed a high-fidelity UI concept and conducted a formative walkthrough with three experts. The findings suggest that the proposed concept can externalize hidden conflicts and support early-stage revision, while highlighting the need for designer control to avoid over-standardization in creative decision-making.