Advances in computational algorithms and fabrication technologies have paved new ways to explore unconventional camera designs and break limits constrained by traditional optics. In particular, the basic optical design of a camera with a focusing lens or a system of lenses has placed a constraint on the minimum camera thickness. We break the thickness constraint by replacing the lens with a thin, light modulating mask placed just above an imaging sensor and use optimization techniques and machine learning to recover high-quality images. This lensless design of cameras can be used in a range of applications from microscopy to photography to machine vision.