ConceptPrism: Concept Disentanglement in Personalized Diffusion Models via Residual Token Optimization

ArXi:2602.19575v2 Announce Type: replace Personalized text-to-image (T2I) generation has emerged as a key application for creating user-specific concepts from a few reference images. The core challenge is concept disentanglement: separating the target concept from irrelevant residual information. Lacking such disentanglement, capturing high-fidelity features often incorporates undesired attributes that conflict with user prompts, compromising the trade-off between concept fidelity and text alignment.