Deep Fusion Generative Adversarial Networks (DF GANs) have shown to be effective tools for producing realistic pictures. In this research, we demonstrate our implementation that creates unique representations of objects and human beings using a pretrained DF GAN. With the help of labelled input photos and identification tokens, our model was able to create new representations of objects and show people doing interesting things. By using script-based prompts, our system was able to generate scenarios with many pictures that made sense. We also made accessibility easier with a Gradle-implemented user-friendly interface. We illustrate the efficiency and adaptability of our method for producing a wide range of visual material through experimental validation and qualitative evaluation.