Assessing the accuracy of eye-tracking through passive filter and active shutter-glasses.

Poster Presentation 43.406: Monday, May 22, 2023, 8:30 am – 12:30 pm, Pavilion
Session: Eye Movements: Individual differences, novel measurement

Sophie Kenny1 (), Jonathan Tong1, Amanda Estephan1; 1VPixx, Canada

Stereoscopic 3D and other binocular stimuli are often presented dichoptically on a single display through “active” systems, using shutter glasses, or “passive” systems, using glasses with polarizing filters. Since both methods operate by partially blocking or filtering out light, researchers who wish to couple these methods with infrared eye-tracking may be concerned about the clarity of eye-tracking signals (e.g. corneal reflections, pupil location and size). We characterized and quantified the impact of passive and active stereo specialized eyewear on eye-tracking data collected on a tabletop, research-grade TRACKPixx3 eye tracker. Participants completed a calibration and fixation task under three fully counterbalanced conditions:1) wearing a pair of active LCD shutter glasses, 2) wearing a pair of passive glasses with polarization filters, and 3) without specialized eyewear. When wearing passive glasses, some participants required a recalibration attempt to achieve a criterion of less than 1 degree of error. However, while wearing shutter glasses, several participants exceeded four calibration attempts. In contrast, when participants wore no specialized eyewear, calibration was successful after one attempt. Following calibration, participants performed a 36-point fixation task, with fixation targets spanning the central 20-degrees, horizontally, and 12-degrees, vertically. We compared error measurements between each condition. The best performance was achieved when wearing no specialized glasses, followed by passive filter glasses, and shutter glasses generally gave rise to the most significant error. When researchers require high-precision eye-tracking with dichoptic single-display stimuli, passive filter eyewear can be expected to produce comparable results to no eyewear. However, with shutter glasses, researchers can expect increased eye-tracking errors. We recommend that, as appropriate for their experiment paradigm, researchers could plan to relax the calibration criteria threshold to retain a maximum of participants or increase their sample size if high tracking precision is required.