In this paper, we show how specific properties of human visual perception can be used for improving the perceived image quality, often without a need for enhancing physical display parameters. The focus of our work is on depth perception and improvement of viewing comfort on stereoscopic and multiscopic displays by exploiting complex interactions between monocular depth cues (such as motion parallax) and binocular vision. We also consider stereoscopic cinematographic applications, where proper handling of scene cuts, reflective/refractive objects, and film grain requires special attention to provide the required level of perceptual quality. Finally, we discuss gaze-driven depth manipulations to enhance perceived scene depth, and we present our predictor for saccade landing position which significantly reduces undesired effects of inherent system latency in foveated rendering applications.