Get Started
Home
Topics
Search
Library
Research questionHow can single-image diffusion personalization preserve identity while enabling controllable attribute-level context changes?A single reference image can cause test-time adaptation to absorb background artifacts into the learned concept and memorize its spatial arrangement. This makes it difficult to preserve identity while changing contextual attributes.
Computer Vision
Diffusion Models
Evaluation & Benchmarks
Image Generation
Machine Learning
Latest papersRecent research connected to this question, newest first.SEAL: Semantic-aware Single-image Sticker Personalization with a Large-scale Sticker-tag DatasetThe source addresses sticker personalization using diffusion models and proposes an architecture-agnostic adaptation module that operates without modifying the U-Net backbone. It also introduces a six-attribute tagged sticker dataset; the reported evidence covers identity preservation and contextual controllability in the sticker domain.research paper · Sep 4, 2026
Related questions
How can text-to-image diffusion models preserve prompt alignment without the extra sampling cost of classifier-free guidance?How can prompt embeddings be optimized during diffusion inference to balance aesthetics, prompt alignment, and resource use?How can generative image systems preserve subject identity under viewpoint changes, degradation, and iterative edits?How can reference-guided image generation control distinct attributes of multiple objects in complex scenes?