Abstract: This study explores the optimization of loss functions in conditional Generative Adversarial Networks (cGANs), specifically the pix2pix model, for the deconvolution of astronomical images ...