Exploring the StyleGANEX for Morphing Faces with Indian Ethnicity

This research explores an approach to face morphing using StyleGANEX, a cutting-edge Generative Adversarial Network (GAN). By projecting source and target images into StyleGANEX’s latent space and interpolating their latent vectors, we achieve a seamless transition between faces, emphasizing realism and visual fidelity. The pre-processing stage integrates facial landmark detection for precise image cropping, ensuring optimal facial feature alignment. The resultant morphed images effectively blended characteristics from both source and target, demonstrating potential applications in entertainment and advanced image manipulation techniques. In this analysis, we specifically blended Indian faces to generate morphing in Indian ethnicity. We observed that StyleGANEX generates high-quality visual results on Indian faces when compared with StyleGAN due to the optimization in latent spaces. Further, StyleGANEX is faster, requires less time for the same number of optimization steps in morphing.