Evaluating Automatic Parameter Control Methods for Locomotion in Multiscale Virtual Environments

Abstract

Virtual environments with a wide range of scales are becoming commonplace in Virtual Reality applications. Methods to control locomotion parameters can help users explore such environments more easily. For multi-scale virtual environments, point-and-teleport locomotion with a well-designed distance control method can enable mid-air teleportation, which makes it competitive to flying interfaces. Yet, automatic distance control for point-and-teleport has not been studied in such environments. We present a new method to automatically control the distance for point-and-teleport. In our first user study, we used a solar system environment to compare three methods: automatic distance control for point-and-teleport, manual distance control for point-and-teleport, and automatic speed control for flying. Results showed that automatic control significantly reduces overshoot compared with manual control for pointand-teleport, but the discontinuous nature of teleportation made users prefer flying with automatic speed control. We conducted a second study to compare automatic-speed-controlled flying and two versions of our teleportation method with automatic distance control, one incorporating optical flow cues. We found that pointand-teleport with optical flow cues and automatic distance control was more accurate than flying with automatic speed control, and both were equally preferred to point-and-teleport without the cues. CCS CONCEPTS • Human-centered computing → Human computer interaction (HCI); User studies.

Combining Motion Matching and Orientation Prediction to Animate Avatars for Consumer‐Grade VR Devices

“…However, the quality of certain types of motion (e.g., reaching an object) may be reduced because of positional misalignment: upper body IK will frequently fail to reach the target. Visual artifacts on the legs are less noticeable when animating self‐avatars, as users usually do not look at their legs [LLKH20]. Thus, we suggest using a small α to avoid problems with the IK and the location of the virtual body for self‐avatars.…”

“…Some studies indicate that synchronized leg animation based on feet tracking increases presence and SoC [FALH20; GCC20; PJ19], whereas other studies have not found significant differences for leg animations inferred with and without foot tracking [GMB*22]. Lee et al [LLKH20] also reached a similar conclusion that a prerecorded animation for the lower body and a synchronized animation predicted by a neural network had a similar effect on presence when observed indirectly during walking in place (WIP). However, synchronized leg animation provided higher presence and Body Ownership when users looked down at their lower body during WIP.…”

“…using only HMD data [CKWv16; LPHK19]. Recent studies use Deep Neural Networks (DNN) to predict the status of leg movement of WIP with the HMD tracking data and then blend leg animations based on the predicted status [LLKH20; HPAB19]. Our work can handle real‐walking, which is proven better than WIP in terms of motion sickness and SoE [SUS95].…”