Here I have demonstrated effect of dimensionality reduction (using PCA) on picture quality and reversing the effect on MNIST dataset. With applications in communication technology such that we can determine how many principal components we can work with thereby less data consumption, without rendering the image unusable.