Bundle-Fit Display and See-Through Models for Augmented Reality Head-Mounted Displays
A calibration model for augmented-reality head-mounted displays that keeps virtual content world-locked despite complex see-through optics. Rather than tracing rays through the optical hardware (white-box) or fitting a single viewpoint-blind function (black-box), students learn to exploit the fiber-bundle structure of the pixel-to-ray correspondence: fit an intermediate model on the base space and compose it with the bundle projection, so the display and see-through models stay accurate across pupil positions (pupil swim) while remaining optics-agnostic and cheap to evaluate via warm-started runtime optimization.
BundleFit: Display and See-Through Models for Augmented Reality Head-Mounted Displays Yufeng Zhu
BundleFit proposes display and see-through models for augmented-reality head-mounted displays (AR HMDs) used in world-locked rendering, where virtual objects must stay registered to real ones as the …