Digital holographic systems are a class of two step, opto-numerical, three-dimensional imaging techniques. The role of the digital camera in limiting the resolution and field of view of the reconstructed image, and the interaction of these limits with a general optical system is poorly understood. The linear canonical transform describes any optical system consisting of lenses and/or free space in a unified manner. Expressions derived using it are parametrised in terms of the parameters of the optical system, as well as those of the digital camera: aperture size, pixel size and pixel pitch. We develop rules of thumb for selecting an optical system to minimise mean squared error for given input and digital camera parameters. In the limit, our results constitute a point spread function analysis. The results presented in this paper will allow digital holography practitioners to select an optical system to maximise the quality of their reconstructed image using a priori knowledge of the camera and object.