This project explores the innovative application of generative AI in the fashion industry, focusing on image-to-image translation to create realistic and diverse fashion designs. The fashion image architectures, StyleGAN and Variational Autoencoders (VAEs) are trained on the DeepFashion dataset, which includes approximately 11,000 images across 15 fashion categories. StyleGAN is recognized for its high resolution, stylized outputs, while VAEs capture uncertainty within fashion image distributions. The models’ performances are compared using the Inception Score (IS), and an interactive interface is developed, enabling users to select a fashion category and generate corresponding images dynamically. The quality and diversity of the generated images are further assessed using evaluation metrics such as Inception Score (IS) and Fréchet Inception Distance (FID). This project aims to advance AI in creative industries and provide insights into the effective use of generative architectures for fashion image synthesis.