KwaiMind achieves the strongest overall scores among evaluated open-source editors on ImgEdit, GEdit, both language splits of REDEdit, and Ecom-Bench visual quality, and the highest aggregate CTR ranking score among compared systems.
Bo-Heng Zhang, Fan Yang, Jia Sun et al.· 0 citations
DM-Align is introduced, which derives a complementary gradient direction to guide the model toward human-preferred samples, and eliminates the need for multi-step reward evaluation and complex ODE-SDE conversions inherent in traditional RL.
Jiu-Zhou Lin, Jun-Long Wu, Feilong Zuo et al.· 1 citation
DASH-OPD (Discrepancy-Aware Switching with Hysteresis for OPD) is proposed, the first agentic OPD method to perform adaptive, bidirectional executor switching and introduces hysteresis that prevents rapid switching triggered by transient discrepancy fluctuations.
Yuchen Xia, Qianguo Sun, Chao Song et al.· arXiv.org· 2 citations
Extensive experiments on both insertion and replacement demonstrate that TextRefine consistently outperforms the evaluated image editing baselines in textual fidelity, placement reliability, and glyph quality while better preserving source-image content.
Hong-Lie Wang, Jia Sun, Zi-Jun Li et al.· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.