This work introduces LayerEditor, a unified framework that couples instruction-guided image editing with controllable generation of reusable RGBA layers, and proposes Compositional Edit-Prior Distillation, which transfers its editing prior to both the edited canvas and the image composed from the predicted layers.
Wen Xue, Yong Liu, Xue-Heng Li et al.· 0 citations
Multi-teacher on-policy distillation (MOPD) extends on-policy distillation (OPD) by routing domain-specific data to specialized teachers, integrating their capabilities through dense token-level supervision on student-generated trajectories. However, domain routing only distinguishes the supervision sources for differe...
Shi-Yu Tian, Yong Liu, Xi-Man Liu et al.· 3 citations
This work presents TryOnReward, a fine-grained reward model tailored for VTON that significantly outperforms generic judges in human preference alignment, and when serving as the RFT reward function, it consistently yields human-preferred try-on results across multiple baselines.
Xue-Heng Li, Yong Liu, Xiao-Long Fu et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.