Delta-K: הגברת יצירת מופעים רב-מופעים באמצעות הרחבת תשומת הלב צוס-עבר

תקציר מקורי באנגליתarXiv:2603.10210v2 Announce Type: replace-cross Abstract: While Diffusion Models excel in text-to-image synthesis, they frequently suffer from catastrophic concept omission when generating complex multi-instance scenes. Existing training-free methods attempt to resolve this by rescaling attention maps, which merely exacerbates unstructured noise without establishing coherent semantic representations. To address this, we propose Delta-K, a backbone-agnostic, plug-and-play inference framework that resolves omission by operating directly in the shared cross-attention Key space. Utilizing a lightweight Vision-Language Model (VLM) preview, we isolate a differential key ($\Delta K$) capturing the pure semantic signature of missing concepts, and proactively inject it during the early semantic pla