Restoring old images using deep learning. The architecture entails a DnCNN model comprised of 17-Layers of a homogenous CNN with batch normalization and ReLU. Initial code is adopted from Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising by K. Zhang, et. al