Smartphone virtual reality (VR) can o er immersive experience while being a ordable and easy to use. To enhance the VR experi- ence under limited smartphone computation and battery resources, solutions have been proposed for e cient rendering and content delivery. However, e orts towards optimizing the distinct head- mounted display (HMD) are unfortunately limited. This paper un- veils the opportunity of optimizing smartphone VR by leveraging human vision in HMD. In particular, we shift the default xed full brightness in VR video/game Apps to a dark adaptation based dynamically scaled brightness. By exploiting the time-varying sen- sitivity of human eyes in dark HMD, we can reduce VR display energy while maintaining brightness perception. The proposed system, Strix, is empowered by a dark adaptation model trained from classic experimental data, a varying trend of perceptual full brightness derived from the dark adaptation model, and a smooth brightness transition scheme balancing energy and experience. Ex- perimental results show that Strix can achieve 25% system energy reduction without negatively impacting brightness perception.