We report a crosstalk analysis method for multiple-image encryption, which focuses on the individual effects of different multiplexed images in the crosstalk noise. The different noise characteristics are analyzed using a simulation model for each image. Using the resulting crosstalk noise characteristics, we can explain the differences in the quality of multiplexed images that cannot be explained using conventional approaches. Furthermore, we describe a simple image-modulation method to compensate for differences in image quality. We then experimentally show that using the image-modulation technique, we can significantly reduce the variation in image quality within a set of images. This technique leads to improvements in the quality of the poorest images in the set.