Researchers have introduced Oxygen-TryOn, a novel foundation model designed specifically for virtual try-on applications across various fashion categories. Unlike general-purpose image editors, Oxygen-TryOn utilizes a dedicated data engine and specialized training to achieve photorealistic synthesis of subjects wearing diverse items. The model supports multiple reference images, full or half-body views, and free multi-item composition, while maintaining subject identity and item appearance. It outperforms existing proprietary and open-source systems on benchmarks for both single and multi-item virtual try-on. AI
IMPACT Sets a new standard for virtual try-on technology, potentially impacting e-commerce and fashion industries.
RANK_REASON Research paper introducing a new model with benchmark results. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
