This chapter discusses GAN applications in image quality enhancement, including denoising, deblurring, tone mapping, super-resolution, and restoration. Denoising frameworks like GCBD use GANs to synthesize realistic noise for training. Deblurring models (e.g., DeblurGAN and DBGAN) combine perceptual and adversarial losses with multi-scale discriminators. Tone mapping leverages paired (e.g., MIT-Adobe FiveK) or unpaired datasets (e.g., CycleGAN-based unsupervised enhancement). Super-resolution (e.g., SRGAN) employs VGG-based perceptual loss. Restoration models address inpainting and artifact removal. The chapter contrasts simulated vs. real-world datasets (e.g., RENOIR, GoPro) and highlights challenges in generalization and real-time processing. Finally, Practical implementations of SRGAN are demonstrated, including code interpretation and training details.