The demand for appealing images is expanding in image-driven industries such as advertising, social media, entertainment, healthcare, and historical preservation in modern society. Image inpainting and outpainting techniques are critical in image restoration and enhancement, addressing issues such as degradation and missing areas. Deep learning methods, such as GANs, show promising results in terms of producing aesthetically beautiful and semantically relevant outcomes, but issues such as mode collapse exist, demanding ongoing study for more robust and accurate solutions. To address these challenges, in this research, we introduce a novel image outpainting technique that uses a unique technique called “priority decay loss”. In contrast to conventional GANs and outpainting strategies, which rely on one or more fixed loss functions throughout multiple training iterations, our method for outpainting combines dynamic loss prioritizing using two crucial parameters: gamma and alpha. While gamma shows the fundamental priority of a loss, alpha determines the rate of priority decay. By allowing us more control over the training direction and the option to integrate various loss functions, this strategy offers variable training techniques.