Applies Neural Style Transfer using a pre-trained VGG19 network. Unlike typical deep learning, NST optimizes pixel values rather than weights — minimizing a combined content and style loss to produce images that carry one image's structure and another's texture.