Vytvoření aplikace pro automatické generování vlastního stylu oblečení

This thesis explores the development process of a web application with special focus on machine learning models. Generative models transforming textual inputs into visual form has experienced a large spike in popularity in recent years. Altogether with an exponential growth of general digital content, creating ideal environment resulting in opportunity of utilizing such models with the aim to generate unique visual content. The research seeks usage of custom text-to-image model, analysing its efficacy in producing realistic and contextually relevant images based on textual input. The specific area of interest for text-to-image model application is fashion design. As well as generative models, fashion design domain has undergone an extensive digitalization, as large portion of sales and consumer communication shifted to online platforms. The development process encompasses the integration of these models into a user-friendly web interface, allowing users to craft unique styled clothes reflecting their personal preferences solely through textual input almost in real time. Through evaluation of user-feedback and comparison with other models, the efficacy, usability, and potential of further improvement is measured. The findings might contribute to enhancement in understanding generative models particularly its practical integration into end-to-end web applications, paving the way for future innovations and content creation.