ANIMATED REMOTE APPAREL FITTING

Smart overview of the Invention

Online shopping for clothing offers convenience and a wide selection, yet it presents challenges in determining garment fit and appearance. Unlike physical stores where try-ons are possible, online shoppers face difficulties in assessing whether clothing suits their body type or style. This uncertainty often leads to dissatisfaction, increased returns, and hesitancy in purchasing clothing online. Variations in sizing across brands further compound the issue, making it hard for consumers to choose items with confidence.

Innovative Techniques

The described techniques enable virtual try-on experiences by generating videos that depict clothing items on a digital representation of the shopper. Using a machine-learning model, the system processes inputs such as an image of the shopper, a clothing item, and a reference video showing movements. The generated video allows users to visualize themselves wearing the garment in motion, offering a more immersive experience compared to static images. This approach helps consumers make more confident purchasing decisions.

Machine Learning Models

The system employs advanced machine-learning models, including conditioning networks and a diffusion transformer, to create conditioning sequences from the input images and videos. A second model uses these sequences to produce a video of the subject wearing the clothing item. An auto-regressive video refiner enhances the visual quality, while user customization through text prompts allows minor edits to movements or clothing items. The diffusion transformer is trained using a multi-phase approach to optimize garment-to-body associations.

Enhanced Consumer Experience

By combining sophisticated machine learning with intuitive user inputs, the virtual try-on system provides a dynamic and user-friendly experience. Consumers can see how garments fit and move, offering a comprehensive understanding of the clothing before purchase. This improved visualization reduces the likelihood of returns and increases satisfaction with online shopping. The system also supports arbitrary garment try-ons and is robust to various capture techniques, allowing users to try on garments seen on others.

System Environment

The digital medium environment includes a remote provider system and a user-associated computer, connected via the Internet. The provider system uses a digital service manager module to support functionalities like accessibility and real-time processing. Users access these services through communication systems like browsers, enabling the remote fitting service to generate and return try-on videos. This setup ensures scalability and efficient delivery of virtual try-on experiences to enhance online shopping.