This project explores how the roles of the artist and AI can be reconsidered in the context of generative world-building.
While the structure of space and the position of perspective are designed by the artist, the dynamic changes within that space are delegated to AI, allowing the relationship between the two to be reconfigured.
The work examines where human decision-making ends and AI autonomy begins, serving as a prototype to explore this boundary.
Experiments exploring how to maintain visual consistency across multiple AI-generated shots.
Focus on camera continuity, background generation, and workflow design using ComfyUI and generative models.
These are fast prototyping experiments rather than polished outputs.
このプロジェクトは、生成AIによって世界観が構築される状況のなかで、作家とAIの役割のあり方を見直す試みです。
空間の構造や視点の位置は作家が設計し、その内部で生じる動きや変化をAIに委ねることで、両者の関係を組み替えていきます。
どこまでを人間が決め、どこからをAIに任せるのか。その境界を探るためのプロトタイプです。
Current directions
– Background generation with shot-to-shot consistency
– Layout guidance using simplified depth references
– Actor/background compositing workflows
– Comparing model behaviour across different setups
Example Input
Example Outputs
Code