Study on the Relationship Between Observation Posture and Vergence When Viewing 360-Degree Images

VR headsets such as the VIVE Pro EYE are well known to provide observers with a high level of immersion because they can display 360-degree images (Fig. 1). Making use of this high level of immersion, VR headsets are widely used for simulations, training, and other applications [1][2]. What kind of spatial perception occurs when viewing 360-degree images through VR goggles? We have investigated this issue by measuring the vergence of the human eye [3].

The vergence angle refers to the angle formed by the line of sight; the larger the angle, the closer the object being focused on. In the real world, people indirectly use this angle to perceive space as part of their three-dimensional spatial information (Fig. 2). While vergence point theoretically always positioned at the image display surface, since VR goggles display two-dimensional images (Fig. 3). Consequently, when viewing 360-degree images through VR goggles, the vergence angle remains the same regardless of which object within the 360-degree images is being fixated on, making it impossible to use vergence as a source of depth information. Although vergence is determined by the object being fixated on, it is known that the vergence position can shift due to other factors as well. A representative factor is ocular accommodation; the phenomenon in which vergence is guided to the point of accommodation is called accommodative vergence [4]. Furthermore, in our study, when observing 360-degree images through VR goggles, vergence increased as the depth position of the target became closer [3]. In this case, the depth information of the target is based on perspective (Fig. 4). This suggested that vergence shifts depending on the target’s depth position, which was induced by perspective. In this experiment, when comparing still images, vergence shifted in response to the target’s depth position, which was induced by perspective. Furthermore, it was demonstrated that vergence also shifts in response to the target’s depth position even when the target is in motion[5].

In this study, we examined changes in vergence when the observer’s posture changes by moving the target behind the subject until the subject turns to look at it. Reference [3] also examined vergence for targets moving behind the subject. However, because that study analyzed vergence during the movement, it included a mixture of vergence values from situations where the target was moving away from the subject and those where it was approaching. In this study, we reanalyzed the data by extracting only the vergence values at the target’s initial position and those after the target stopped at first.

Subjects observed the 360-degree images while wearing VR goggles (HTC VIVE PRO EYE). The displayed 360-degree images showed two balls; the ball on the left moved back and forth, while the ball on the right always remained stationary (Fig. 5). For the left ball, we created a short-distance condition, in which it approached the subject and then retreated, and a long-distance condition, in which it moved behind the subject and then retreated (Fig. 6, Fig. 7). In the short-distance condition, after the left ball appeared, it remained stationary for 2 seconds. And then it approached subjects for 6 seconds. After approaching, it remained stationary for 2 seconds, retreated for 6 seconds, and returned to its initial position. In the long-distance condition, the left ball remained stationary for 2 seconds after the left ball was displayed, just as in the short-distance condition. And it then approached the subject for12 seconds and continued moving until it reached the area behind the subject. After moving behind the subject, it remained stationary for 2 seconds. It then returned to its initial position for 12 seconds, following the reverse route of its approach.The left ball in the long-distance condition followed a trajectory that was an extension of the left ball’s trajectory in the short-distance condition. Furthermore, the distance traveled by the left ball in the long-distance condition was twice that of the left ball in the short-distance condition. Consequently, the travel speeds of the left balls in the long-distance and short-distance conditions were the same. Each trial lasted 16 seconds in the short-distance condition and 28 seconds in the long-distance condition. The right ball always remained stationary. Subjects were shown the images for the short-distance condition first, followed by the images for the long-distance condition. The images for each condition were shown five times. The participants’ task was to fixate on the left ball, and their vergence angle, output from the VIVE PRO EYE, was acquired at 120 Hz by the PC used for images output. The number of subjects was eight. They were in their 20s and had normal vision.

In the short-distance condition, the subjects’ vergence were measured during the latter 1 second of the first 2 seconds pause (1st measurement period), and the vergence during the latter 1 second of the second 2 seconds pause (2nd measurement period) (Fig. 8). In the long-distance condition, as in the short-distance condition, the vergence angle was measured during the latter 1 second of the initial stationary period (1st measurement period) and during the latter 1 second of the period when the left ball was stationary behind the subject (2nd measurement period). The difference in vergence between the mean 1st and 2nd measurement period was calculated. Since the initial positions of the left ball were identical in both conditions, it was predicted that the difference in vergence between the 2nd and 1st measurement periods would be greater in the short-distance condition than in long-distance condition, if the perspective information effectively would work. A mixed-effects model test was conducted with the difference in vergence between the mean 1st and 2nd measurement period as the dependent variable, the distance traveled by the left ball as the fixed effect, and the subjects as a random intercept. The results of the analysis showed no significant effect of the distance traveled by the left ball. Fig. 8 showed the estimated mean of the difference in vergence between the 2nd and 1st measurement period for each condition. Although the difference in vergence was greater in the short-distance condition than in the long-distance condition, indicating a change in vergence, no statistically significant difference was observed. This suggested that when the left ball moved enough that the subject had to turn their head, no clear fluctuation in vergence was observed. Why were no fluctuations in vergence observed in this experiment? One possible cause was the effect of changes in the subject’s posture. Under the long-distance condition, the left ball moved behind the subject, so the subject had to turn their head to continue fixating on it. At this time, the subject had to fixate on the left ball while rotating both their body and head simultaneously. During head movement, the entire visual field on the subject’s retina moves rapidly, so it was possible that the accuracy of perceiving changes in the depth position of the left ball decreased. As shown in the Fig. 8, the difference in vergence itself was greater under the short-distance condition than under the long-distance condition, indicating that the perspective effect inherent in 360-degree images was reflected in the vergence. However, because the precision of subjective depth position decreased during head rotation, leading to greater variance, it was possible that no statistically significant difference was observed. Based on these findings, when postural changes—including head rotation—were required, the variation in vergence in response to the depth position of the fixation target was limited, and no clear changes were observed.

In this experiment, no clear fluctuations were observed in the vergence of the fixation target in 360-degree images during body movements, including head rotation. In the future, we plan to compare whether a similar trend is observed for objects in real space and examine whether the results of this study represent a phenomenon specific to 360-degree images.

In this study, vergence while continuously fixating on a moving ball was measured. Considering perspective information, vergence was expected to be smaller in the condition where the ball moved backward than in the condition where it approached the viewer. However, no significant difference was observed between the two conditions. One possible reason for this lack of difference is that rotating the head was necessary to fixate on the ball as it moved backward. This is because, during head rotation, the accuracy of the subjective depth position of the ball decreased, and vergence did not fluctuate instantaneously in response to the depth position of the fixation target.