A very fast progress generative adversarial network altered the landscape of synthetic images that could almost, highly realistically and viscerally, visually convince people with realistic-looking human faces in synthetic media. GANs play a major role as a primary tool inside deepfakes. Thus, in the present research work, this synthesis is associated with image manipulation with both seamless facial morphing and detection inside deep-fake topics. The work explores new methods to traverse the generator latent space so that the face image transition is smooth through DCGANs. Meanwhile, in parallel to the exploration of generator latent spaces, the XceptionNet model is utilized to identify deepfake images and provide robust solutions toward the identification and mitigation of synthetic media’s impact.