Depth cues, such as binocular disparity, play a critical role in supporting accurate and precise reaching within near space, facilitating reliable interaction with objects and surfaces [27]. Augmented reality (AR) presents unique challenges for such interactions, as mismatched focal distances between real and virtual objects can distort perceived depth. Prior work has examined the perceptual consequences of focal distance mismatches under controlled viewing conditions, as well as AR task performance in more natural settings; here we consider these factors together in a visually guided reaching task with multiple available depth cues for combinations of real and virtual objects. Participants performed a precision visual motor task in an AR environment, placing a virtual ring around either real or virtual posts at varying distances. Our results show that accuracy was degraded when placing a virtual ring on real posts, with fewer successful placements compared to virtual posts. We demonstrate that this increase in error was due to a systematic underestimation of virtual object distances relative to their physical counterparts. These findings indicate that distortions in perceived depth between real and virtual objects persist during active reaching, despite the availability of multiple depth cues including binocular disparity, proprioception, occlusion, and motion parallax. Measuring and correcting for these distortions is important for optimizing the effectiveness and usability of AR devices.

Binocular disparity provides metric depth information, while monocular cues like occlusion offer depth order. In augmented reality (AR), conflicts between these cues can occur when virtual objects fail to be occluded by real-world surfaces, creating a depth cue conflict and subsequently impacting depth perception. The integration of occlusion and binocular disparity was investigated under this cue conflict in AR using distance-matching paradigms. These paradigms were applied within reach space (0.35–0.5 m) and beyond reach space (0.9–1.5 m). Observers matched the distance of a virtual letter ’A,’ superimposed on a physical surface, using a virtual probe. In addition to the probe, manual reach responses were also made with the index finger for the within reach space condition. Results revealed consistent underestimation of the letter’s distance when it was rendered beyond the surface, with errors proportional to distance. These biases persisted even when proprioceptive information was available, highlighting the robust influence of occlusion cue conflicts on perceived depth. Thus, designers must carefully plan and position virtual augmentations to avoid such errors and their impact on user interaction.

Stereoscopic displays present similar but different images/videos separately to the left and right eyes. Here we report the results of a subjective assessment of computer-generated high dynamic range (HDR) stereoscopic video compression targeting lightweight, low-latency visually lossless performance. The ISO/IEC 29170–2 non-flicker paradigm (Annex A) was adapted for stereoscopic videos and used to evaluate the VESA DSC 1.2a and VDC-M 1.2.2 codecs operating at and beyond their recommended maximum compression levels. Both codecs exhibited visually lossless performance at their target compressions levels. There were differences in compression artifact detection rates with some image sequences between stereoscopic and 2D presentation although the direction of the difference was inconsistent. The results confirm the effectiveness of these codecs for Stereoscopic 3D (S3D) videos and suggest that observers are relatively insensitive to compression artifacts in moving stereoscopic imagery.

In stereoscopic displays different images are presented separately to the left and right eyes. This requirement may increase the bandwidth demand as well as increase the occurrence of visible compression-related artefacts. Here we report the results of a large-scale subjective assessment of high dynamic range (HDR) stereoscopic image compression. The ISO/IEC 29170-2 flicker paradigm was adapted for stereoscopic images and used to evaluate two VESA (Video Electronics Standards Association) image compression codecs: DSC 1.2a and VDCM 1.2.2. We compared the performance on stereoscopic images versus 2D images for both codecs.