Over the past few years, numerous morphing techniques have emerged that target the vulnerabilities of Face Recognition Systems (FRS). The Development of Generative Adversarial Networks (GANs) and diffusion-based high-quality morphing generation techniques has presented an even greater challenge because these techniques can capture even the most subtle details with remarkable precision. Face Morphing Attack Detection (MAD) has been extensively studied from conventional texture-based methods to more sophisticated Convolutional Neural Network (CNN) based approaches. Among these available approaches, the use of residual noise computed using image-denoising methods has gained attention owing to its generalizability. This paper presents a deep and critical analysis of existing residual noise-based Single-image-based Morphing Attack Detection (S-MAD) in terms of performance and generalization to unseen morphing generation techniques and unseen medium. To this extent, we compare and discuss six popular residual noise-based S-MAD algorithms on three different types of face morphing generation methods and two different data mediums such as digital and print-scan. A comprehensive comparison of residual-noise-based S-MAD methods is presented using three different evaluation protocols. Therefore, this study serves as a valuable resource for gaining insight into the performance of residual-noise-based S-MAD algorithms.