Auto-stereoscopic displays can be classified into holographic and nonholographic devices. In this review paper we compare several systems in each group on the basis of their information content. Two approaches currently under investigation at Stanford University are discussed. Several investigators have noted a chronological decrease in the diffraction efficiency of sequentially recorded images in a multiple exposure hologram, but no satisfactory explanation for this phenomenon, which we denote by Holographic Reciprocity Law Failure (HRLF), has been presented. We propose a mechanism, based on latent image formation in silver halide films, to explain HRLF and a procedure to equalize the efficiency of all simultaneously reconstructed images. As many as 16 images have been superimposed and reconstructed with equal brightness. Digital holography is an attractive method for displaying discrete volumetric data bases, but present computational methods based on implementing DFT routines for the calculation of Fresnel holograms require large memory and long computational times. We have investigated an alternative method, based on calculating a convolution operation directly (which also uses DFT techniques), but which allows parallel computation of patches of the hologram. This method offers a tradeoff between processors, memory requirements and computation time.