We describe studies of faint structures in the outskirts of galaxies, inclusing tails, shells, haloes, and thick disks, made possible by optical and near-IR imaging at the deepest levels feasible. Such structures provide unique clues to the early formation history of galaxies as well as their subsequent evolution. Specialised data analysis techniques now allow us to reach levels of surface brightness below 30 arcsec/sq. arcsec (measured as 3 sigma over 10x10 arcsec2 areas) in images obtained from space, or from small or large ground-based telescopes. I will present highlights of our group's recent work, including the detection of satellites and tails and the nature of ultra faint galaxies. I will also show how we can use modern machine learning techniques to handle the immense quantities of images and spectra to be obtained with upcoming telescopes and surveys, including from LSST, Euclid, or JWST.