Perceptual Thresholds for Radial Optic Flow Distortion in Near-Eye Stereoscopic Displays

TL;DR

This study quantifies perceptual sensitivity to radial optic flow artifacts in near-eye varifocal displays and demonstrates blink suppression as a practical masking mechanism. Using adaptive psychophysics with a stereoscopic, wide-FOV display simulator and a 2-interval forced-choice task, the authors measure thresholds before, during, and after self-initiated blinks across ten participants. They report baseline sensitivity around $0.15\%$ image-size change, rising to $1.5-2.0\%$ during blinks, with suppression lasting about $70$ ms after blink onset; the maximum distortion that can be hidden during a blink is approximately $2\%$ (range $1.14\%-2.54\%$). The results provide empirical constraints to hardware design and software distortion-correction algorithms for future varifocal near-eye displays, suggesting blink timing can be leveraged to mitigate visible artifacts without perceptual loss.

Abstract

We provide the first perceptual quantification of user's sensitivity to radial optic flow artifacts and demonstrate a promising approach for masking this optic flow artifact via blink suppression. Near-eye HMDs allow users to feel immersed in virtual environments by providing visual cues, like motion parallax and stereoscopy, that mimic how we view the physical world. However, these systems exhibit a variety of perceptual artifacts that can limit their usability and the user's sense of presence in VR. One well-known artifact is the vergence-accommodation conflict (VAC). Varifocal displays can mitigate VAC, but bring with them other artifacts such as a change in virtual image size (radial optic flow) when the focal plane changes. We conducted a set of psychophysical studies to measure users' ability to perceive this radial flow artifact before, during, and after self-initiated blinks. Our results showed that visual sensitivity was reduced by a factor of 10 at the start and for ~70 ms after a blink was detected. Pre- and post-blink sensitivity was, on average, ~0.15% image size change during normal viewing and increased to ~1.5-2.0% during blinks. Our results imply that a rapid (under 70 ms) radial optic flow distortion can go unnoticed during a blink. Furthermore, our results provide empirical data that can be used to inform engineering requirements for both hardware design and software-based graphical correction algorithms for future varifocal near-eye displays. Our project website is available at https://gamma.umd.edu/RoF/.

Figures (9)

• Figure 1: A visualization of the vergence-accommodation conflict (recreated from hoffman2008vergence), which varifocal HMDs are designed to mitigate. (A) In real-world viewing conditions, when the observer focuses on an object, their vergence distance and focal distance match. (B) In VR HMD viewing conditions, when the observer focuses on a virtual object rendered on the stereoscopic display, their vergence distance matches that of the real-world condition, but their focal distance is tuned to the distance of the display, which does not match the perceived distance of the object. (C) When the vergence and focal distances match, the object being looked at appears in focus and is clear (blue bunny), while objects located at other distances from the observer appear blurred (red bunnies on either side). (D) When the vergence and focal distances do not match (and the observer focuses on a far-away object), all virtual objects appear in focus and are clear due to the fixed focal power of the display. If the user instead looked at an object at a nearby object, all of the virtual content would appear out of focus.
• Figure 2: A flowchart showing the procedure for a single trial in our 2IFC experiment. During each "Interval" block, the virtual scene was rendered on the display either with or without the radial optic flow distortion, which was triggered after the user initiated a blink (see \ref{['fig:teaser']}). Only one of the two intervals had the distortion effect, chosen randomly for each trial. The intermission screen was used to mitigate any afterimage effects. On the response screen, participants pressed a button to indicate which of the two intervals they believed contained the radial optic flow effect. Before the next trial began, we used AEPsych owen2021adaptive to assign the new values for the optic flow magnitude and delay between the blink and the optic flow onset in order to converge towards the participant's perceptual threshold for radial optic flow.
• Figure 3: An image of Papertown, the 3D virtual scene that participants viewed. This scene was chosen because its rich and natural properties (many 3D objects with varying shape, texture, size, and color properties) make it representative of typical content that is viewed in VR HMDs. Note that participants viewed a still image of the scene, so it did not contain any dynamic elements.
• Figure 4: Distribution of blink durations (voluntary and involuntary) across all participants. The high concentration of blinks in the $100-200$ ms range aligns with prior research on blink durations holmqvist2017eye, which suggests that participants did not have abnormal blink behaviors in our experiment.
• Figure 5: Average duration of voluntary and involuntary blinks. Involuntary (natural) blinks were significantly longer than involuntary blinks in our experiment. Although the absolute difference in duration between voluntary and involuntary blinks is not large, this statistical difference suggests that the engineering requirements for radial flow distortion may change depending on the type of blink a user initiates.