A phase encoded image is encrypted using the double random phase encoding technique, (DRPE). The effects of using a variable dynamic range of phase distribution during phase encoding (pre-encryption) are examined. We begin by phase encoding the input image using the full phase range, from -pi to pi. We perform numerically perfect encryption and we then introduce errors into the decrypting phase-keys in the form of a pseudo-random distribution (position and phase) of incorrect pixels values. By quantifying the resulting error in the attempted decryptions, for increasing amounts of error in the decrypting phase-keys, we examine the effects of reducing the phase range to, +/- (pi - Delta). In this way we attempt to improve the phase encoding procedure for use with the DRPE technique. When the pixel values calculated, during an attempted decryption, fall outside the phase range used to phase encode the assigned input image, we examine different methods of redistributing the value of that pixel to an assigned value within the allowed phase range. The effects of the phase quantisation used in the keys are also examined. (C) 2008 Elsevier B.V. All rights reserved.