betaFeedback

WorksComputer Science

Apple Vision Pro: The good, the bad, the ugly

Abstract

Apple Vision Pro (AVP), the headset Apple introduced as its first spatial computer,'' built by a single, vertically integrated vendor, entered the market in February 2024 and has since accumulated a fast-growing body of primary research spanning surgery, education, accessibility, design, and human-computer interaction. In this paper, we adopt the tripartite structure of Leone's 1966 film to organize a systematic review of this literature: we cataloged the validated, working applications of AVP (the good), the persistent technical and ergonomic limitations of a still-early-generation headset (the bad), and the ethical concerns and structural harms that follow from deliberate design, marketing, and deployment choices (the ugly). The review is built from a Scopus export of 218 records mentioning AVP, which we narrowed to 104 papers (2024–2026) with a primary focus on AVP, using explicit inclusion/exclusion criteria that we report in full, alongside a PRISMA-informed flow diagram of the screening process. We find the corpus heavily skewed toward validating benefits (medicine and surgery, education, and software tooling account for over half of the included papers) and comparatively thin on documented risks or misuse, a pattern we treat as a finding in its own right rather than a limitation to be explained away. We further show that several recurring issues, namely price, simulated social presence, and the gaze-tracking privacy architecture, resist a clean separation into bad'' or ``ugly'' using the published evidence alone, since both readings remain compatible with the same data. We close by identifying where the evidence base most needs to grow: independent risk and ethics research, bystander- and equity-focused study designs, and disclosure that would let researchers distinguish an engineering backlog from a structural choice.

Introduction

When Sergio Leone’s The Good, the Bad, and the Ugly [1] premiered in 1966, its tripartite structure captured a simple idea: that the same unfolding situation can contain genuine value, unresolved difficulty, and moral compromise at once, without those three registers collapsing into one another. We borrow that structure in this study for a concrete case in spatial computing: Apple Vision Pro (AVP), the mixed-reality (MR) headset Apple introduced as its first “spatial computer” [2], a term predating the product by two decades.

AVP is a useful case for this kind of review because, unlike “artificial intelligence (AI),” a label that spans many companies, architectures, and constantly shifting systems, AVP names one company’s product, running one operating system on one hardware generation at a time. Its enumerable set of components includes a forward-facing camera and depth-sensor array, an eye-tracking pipeline, an outward-facing “EyeSight” display and “Persona” avatar system, a curated App Store, and a launch price of $3,499 [3].

Apple updated the hardware once during the period this review covers, releasing an M5-chip model on 22 October 2025 with a redesigned, counterbalanced “Dual Knit Band,” a GPU that renders 10% more pixels on the same micro-OLED displays, and passthrough refresh rates up to 120Hz, at an unchanged $3,499 starting price; Apple’s own materials describe this as an upgraded Vision Pro rather than a numbered second generation, though third-party coverage commonly calls it that [4].

Since its release, a fast-growing body of academic research has evaluated what this specific device does well, where it still falls short, and what it enables people to do to one another, deliberately or otherwise; § 1.3 quantifies the growth and composition of this literature within our own included corpus.

This paper makes two contributions. First, we conduct a systematic screening of the AVP literature indexed in Scopus, documented in § 1.1 with a PRISMA-informed flow diagram, and organize what survives screening into a good/bad/ugly catalog: § 2 covers the good (validated, working applications), § 3 the bad (technical and ergonomic limitations still being iterated on), and § 4 the ugly (ethical concerns and structural harms arising from deliberate choices). Each section closes with a summary table.

Second, we identify genuine overlaps between the bad and the ugly catalogs. Price, simulated social presence, and the gaze-tracking privacy architecture each admit two readings that the published evidence cannot adjudicate between.

This paper also presents, in § 5, what this indeterminacy, together with the shape of the current evidence base, implies for where AVP research should go next, and, in § 6, concludes this study.

Review Method

The reviewed literature was identified and screened, and the included papers were subsequently classified into the good, bad, and ugly catalog that follows.

Literature search strategy

We searched Scopus for records mentioning Apple Vision Pro in Title, Abstract, or Keywords, and exported all matches, yielding 214 records. We exported the results twice. The first export, on 10 September 2026, initially returned 242 records; a corrected re-export returned 214. We did not preserve a record of why the two exports differed, but none of the 28 dropped records is part of the included corpus. An updated export on 9 October 2026 contained the same 214 records plus four newly indexed ones, for a total of 218. All screening reported below was performed on this updated export.

Literature inclusion/exclusion criteria

We screened every record’s title, abstract, and document type against four criteria, applied by close reading rather than keyword matching:

  1. Document type: we excluded Letters, Notes, Conference reviews, Books, and Book chapters, retaining Articles, Conference papers, Reviews, and Data papers for further screening;

  2. Primary focus: a record was retained only if AVP itself, rather than headsets in general, was the central subject of its stated contribution (a paper benchmarking AVP alongside Meta Quest 3 and HoloLens 2 as one of several comparison devices was excluded on this ground even when methodologically sound); comparative studies were retained where AVP was the device under evaluation and other devices served only as baselines against which AVP’s performance was measured, and excluded where AVP appeared as one of several devices being surveyed with no single device central to the paper’s contribution;

  3. Language and genre: we excluded records whose abstract text was not in English and pure opinion or editorial pieces with no study, system, or evaluation behind them; and

  4. Completeness: records with no usable abstract were excluded (those that only mention the AVP but the study does not include it).

Additionally, of the 218 records identified, 2 were duplicate publications of the same study and were removed before screening, leaving 216 records for the first screening, in which the first screener applied the criteria above. That stage excluded 106 records (76 not primarily about AVP, 24 by document type, and 6 with no usable abstract) and retained 110. In the second screening, a second reviewer checked all of the first screener’s decisions and, in a separate blind pass on the same updated export, screened every record without sight of those decisions.

The second reviewer flagged eight retained papers in which AVP might not be the central subject, and the first screener reread them against the criteria. Six were excluded at this stage: one essay on smart glasses in general, two systems built for multiple headsets, and three studies that used AVP only as a research instrument. The other two [5, 6] were retained as judgment calls under criterion 2.

Each measures AVP’s performance against a no-headset baseline and against one optical-see-through headset. AVP is the device under evaluation in both, but the presence of a second measured headset places them at the boundary of that criterion. Because the eight records were resolved through discussion, their reclassification is a consensus decision rather than an independent one. Although the blind pass would permit one, we did not compute an inter-rater agreement statistic, which we note as a limitation in § 5.

The two stages together excluded 112 records (82 not primarily about AVP, 24 due to document type, and 6 with no usable abstract), leaving 104 papers included.

Figure 1 summarizes this PRISMA-informed screening flow [7], and we return to the composition of the resulting corpus below (§ 5) as a finding in its own right about the state of the field. This 104-paper corpus is supplemented by 14 further references, cited throughout the text, that support methodological or contextual claims (instrument and model definitions, manufacturer specifications, a security advisory, and pricing sources) rather than being part of it.

PRISMA-informed screening flow diagram

Figure 1. Screening flow for the Scopus export underlying this review, from the initial search (n = 218), through duplicate removal (n - 2 = 216) and first-screening exclusion (n - 106 = 110) and second-screening exclusion (n - 6 = 104), to the included corpus analyzed.

Analytical framework

We coded every included paper’s primary contribution into exactly one of four buckets, operationalizing the tripartite structure introduced above for this specific device rather than for head-mounted displays in general.

This good/bad/ugly/mixed coding reflects a paper’s stated primary contribution, not its statistical outcome: a study whose central aim was to validate a benefit was coded good even when its own results were null or mixed on that benefit (e.g. [8] in § 2), provided the paper’s framing, design, and discussion remained oriented toward that validating aim. We flagged in § 5 that this rule likely biases the good count upward relative to a coding scheme keyed to the statistical outcome rather than the stated aim.

A “good” classification

A paper was coded good when its central contribution was a validated or working application: a feasibility study, usability evaluation, or system description reporting that AVP accomplished a stated task.

A “bad” classification

A paper was coded bad when its central contribution documented a technical, ergonomic, or software-ecosystem limitation, a constraint the authors themselves frame as a target for future engineering rather than as a deliberate design or deployment choice.

An “ugly” classification

A paper was coded ugly when its central contribution concerned an ethical, social, or structural consequence of a choice already made in how AVP is built, marketed, priced, or deployed, a consequence that engineering iteration alone would not resolve.

A “mixed” classification

A fourth outcome, mixed, was recorded when a paper’s own reported findings substantively supported more than one of these registers rather than a single one dominating; such papers are cited in every section to which their evidence speaks; § 5 discusses several concrete examples.

This coding produced 68 good, 7 bad, 2 ugly, and 27 mixed papers (68 + 7 + 2 + 27 = 104); Table 1 summarizes the split. Reference numbers are assigned in first-citation order in the text, not by bucket, so they do not align directly with this split. A mixed paper’s first citation typically falls in whichever section happens to introduce it, even though the same paper is also cited later in one or two other sections.

BucketCountRepresentative refs
Good68[9, 10, 11]
Bad7[12, 13, 14, 15]
Ugly2[16, 17]
Mixed27[18, 19, 20, 21]

Table 1. Good/bad/ugly/mixed coding of the 104 included papers, with representative references.

Figure 2 characterizes the resulting corpus by publication year: from 11 primary-focus papers in the device’s first partial calendar year of availability (2024) to 47 in 2025, with 46 more already recorded through 9 October 2026, a trajectory consistent with a research community still in the early, accelerating phase of studying a new artifact.

Growth of primary-focus Apple Vision Pro research by publication year

Figure 2. Growth of primary-focus Apple Vision Pro research in the included corpus (n = 104), by publication year. The 2026 figure is partial-year, covering records retrieved through 9 October 2026.

Alongside this good/bad/ugly coding, we classified each included paper by its primary application domain (medicine and surgery, education, or software tooling) to characterize the composition of the corpus independently of the tripartite catalog and to structure the sub-themes reported within § 2 to § 4. Figure 3 shows the resulting distribution: medicine and surgery, education and training, and the software and developer ecosystem together account for over half of the 104 included papers, a concentration we return to when discussing where the evidence base most needs to grow.

Application domains of the 104 papers included in this review

Figure 3. Application domains of the 104 papers included in this review (2024–2026).

Because almost every included paper’s data collection predates the October 2025 M5 refresh, this corpus overwhelmingly reflects the original M2-based hardware; the ergonomic and performance findings cataloged in Section 4 in particular should not be assumed to transfer unchanged to the M5 model, since a heavier, differently balanced headset plausibly shifts exactly the comfort and workload outcomes that literature measures: the M5 model weighs 750–800g with its redesigned, counterbalanced Dual Knit Band, versus 600–650g for the original model’s Solo Knit Band and Dual Loop Band options [3, 22], and it supports a higher passthrough refresh rate.

The good

The literature converges on a clear pattern: wherever a task benefits from precise spatial registration, embodied interaction, or head-worn stereoscopic visualization, the AVP has been shown to work, often on the first attempt. The following sub-themes summarize where the corpus reports validated, positive, or genuinely promising outcomes.

Surgical visualization and intraoperative guidance.

The single largest and most consistent cluster of positive findings concerns intraoperative use of the AVP as a wearable, wireless display that consolidates multiple imaging feeds directly into the surgeon’s line of sight. In ophthalmology, prospective feasibility studies report high procedure-completion rates and reduced surgeon workload when the headset served as the primary visualization platform for cataract surgery, endoscopic dacryocystorhinostomy, intraocular surgery with live telementoring, and oculoplastic procedures, with usability ratings consistently in the “excellent” range [9, 23, 24, 25].

Hepatobiliary and abdominal surgeons report similar gains: an AVP-based real-time image-fusion navigation system increased puncture accuracy from 42.5% to 90% in a comparative experiment [26], separate from the same study’s eight-patient clinical series, which was completed without major complications. augmented reality (AR)-assisted cholangioscopy and laparoscopic radiofrequency ablation were completed without technical failure or complications [27, 28], and the device played a documented role in guiding lesion identification during the first Da Vinci-assisted total gastrectomy performed with its aid [10].

In spine and orthopedic procedures, AR-guided plate positioning and biportal or standard endoscopic lumbar surgery were completed successfully with head-mounted visualization replacing or supplementing conventional monitors, without added cognitive workload or complications [29, 30, 5]. Minimally invasive general surgery, urologic Mini-ECIRS, and chest-wall oncologic reconstruction case series likewise report low perceived workload, uneventful outcomes, and workflow compatibility [31, 32, 33], while a co-designed multimodal assistant demonstrates that sterile, glanceable AR interaction can be built around interventional radiology’s specific procedural constraints [34].

A scan-derived digital-twin framework further shows sub-2mm registration accuracy for surgical robotic components rendered on the device [35], and immersive endoscopy reduces neck and shoulder muscle activation relative to a non-ergonomic monitor arrangement [36]. A neurosurgical planning application built natively for the headset, with controller-free manipulation of patient-specific 3D models and synchronized multi-user case discussion, received a total satisfaction score of +1.7 on a scale from −2 to +2 from 15 neurosurgeons [37].

Surgical, procedural, and clinical training.

Beyond live intervention, AVP-based simulators and stepwise AR overlays show measurable learning gains across several specialties. A stepwise AR learning-support system for surgical education and an AVP-integrated laparoscopic surgery teaching program each report improved trainee performance, with the latter documenting a 15% improvement in operative skill and a 9% increase in exam scores after iterative latency optimization [38, 39]. Domain-specific MR trainers outperform lecture- or screen-based instruction: an ultrasound-interpretation tool for veterinary sonography improved both immediate comprehension and six-week retention relative to lecture-trained peers [40], a naloxone-administration trainer increased knowledge, confidence, and intent to help across a heterogeneous lay and professional sample [41], and a resident vascular-surgery case study found superior spatial awareness of a 3D aortic model compared with standard workstation viewing [42].

A hybrid femoral-nailing simulator combining real instruments with AR guidance produced clear procedural-efficiency gains that were retained at follow-up [43]. At the same time, quantitative eye-tracking benchmarking supports the device’s suitability for training cardiac surgeons in fine-grained anatomical landmark selection [44].

Diagnostic imaging, radiology, and pathology workflows.

Outside the operating room, pilot deployments in radiology and pathology point to strong short-term feasibility. A prospective pilot with 25 radiology physicians (22 residents, 2 fellows, 1 attending) found rapid onboarding (under five minutes), high perceived utility for multidisciplinary conferences, and majority-positive ratings of ease of use [45]. Cinematic rendering software for volumetric computed tomography (CT) and magnetic resonance imaging (MRI) review was rated highly usable by 14 medical experts on the platform [11], and a gigapixel whole-slide pathology viewer that combines embodied navigation with multimodal AI assistance demonstrated technical feasibility for reducing the panning and context-switching burden of conventional 2D digital pathology [46].

In interventional settings, AR and VR preprocedural planning changed the surgical approach in nearly half of 50 hepatic oncology cases and yielded high scores in anatomic understanding and procedural confidence [47]. At the same time, comparative phantom studies of AR-guided percutaneous biopsy and ultrasound-guided needle placement found accuracy and workflow benefits comparable to or exceeding those of conventional guidance, with novice operators gaining the most [48, 49].

A pilot comparing headset- and desktop-based interpretation of diverticulitis on CT found statistically equivalent diagnostic accuracy between the two [50]. Elsewhere, an AR-guided functional near-infrared spectroscopy (fNIRS) probe-positioning system achieved sub-centimeter, hands-free placement accuracy for neurorehabilitation applications [51], and a review outlines the device’s potential as a portable, home-based perimetry tool for glaucoma monitoring [52].

Education and embodied cultural learning.

A recurring finding across the education literature is that embodied, spatially situated interaction produces measurable learning gains over flat or GUI-based alternatives. A controlled study of a Chinese-artifact learning platform found significantly higher learning gains, engagement, and interest than a traditional GUI equivalent [53], and a spatial-computing chair-design curriculum produced more intuitive, creative design outcomes than an iPad-based control condition [54].

Museum and heritage applications extend this pattern: an MR system for exploring a national museum’s storage and curatorial practices [55], an eXtended reality (XR) documentary format developed explicitly in response to the device’s release for visible-storage exhibitions [56], and a retrieval-augmented, voice-and-gesture interface for conversing with digitized historical archives [57] all demonstrate novel access to cultural content that flat interfaces do not afford.

A reproducible cardiovascular-anatomy platform paired with a retrieval-augmented voice tutor met its real-time interaction targets across eight anatomical systems while substantially undercutting the licensing cost of proprietary alternatives [58]. Field-based and blended-learning studies add nuance without displacing the overall positive signal: mixed-reality note-taking during outdoor excursions improved content extraction for physical and textual information relative to pen-and-paper and mobile notes [59], and a comparison of classical media against XR eLearning found immersive presentation boosted engagement with visually rich content, with most participants keen to use XR for learning again [18].

Adoption-modeling and controlled-comparison research adds a further, more mixed data point: a study of AVP’s AI-assisted features using PLS-SEM, partial least squares structural equation modeling, found most theorized predictors of continued use confirmed, though informativeness gratification did not significantly predict use [60, 61], and a counterbalanced comparison of Persona-mediated mixed-reality vocal instruction against tablet-based video and in-person teaching found only a non-significant trend toward reduced performance anxiety, with no significant difference in sight-singing outcomes overall [8].

Accessibility and assistive applications.

Several teams treat the device’s depth-sensing, tracking, and passthrough capabilities as infrastructure for accessibility research rather than as an end in themselves. A technical feasibility study used AVP’s depth detection and semantic segmentation to identify trip hazards exceeding 1.5 cm in height, a step toward real-time fall-prevention support [62], and a companion VR study used the platform to build ecologically realistic obstacle scenarios for studying older adults’ walking safety, isolating an interaction effect between static and dynamic hazards [63].

A co-design study with adults with Cerebral Visual Impairment used the headset across a two-week diary study and iterative development sessions to address object location, text reading, face recognition, and sensory stress management in a population underrepresented in assistive technology research [64]. An AR interface built on the device’s tracking and object-detection capabilities to support semi-autonomous prosthetic-hand control significantly lowered workload and raised usability scores for participants open to wearing a headset [65]. An in-the-wild evaluation with Deaf and hard-of-hearing users likewise revealed concrete opportunities to make spatial computing more inclusive for this population [66].

Design, architecture, and industrial engineering workflows.

In design and engineering contexts, immersive 1:1-scale evaluation is repeatedly credited with judgments that flat screens do not support. An architectural design study found that while tablet interaction accelerated idea generation, spatial immersion on the headset enabled a deeper perceptual understanding of mass, light, and enclosure, shaping design choices beyond what screen-based tools afford [67]. A biomimetic transportation design study similarly found that immersive scale and depth perception enhanced the assessment of proportions during AI-assisted vehicle styling [19].

In structural engineering, an interactive topology-optimization playground built natively for the device integrates real-time human judgment into compliance-minimization workflows, aiming to reduce lead times and improve manufacturability [68]. A natively redesigned interface for teleoperating an underwater remotely operated vehicle achieved the highest usability scores and the lowest workload among the three tested interfaces, outperforming both a tablet-based control and a direct port of that interface [69].

Finally, an immersive digital supply chain twin prototype was evaluated as a step toward more transparent, collaborative planning among network partners [70].

Spatial video, novel sensing, and hardware-enabled research instrumentation.

The device’s built-in stereoscopic capture, precise eye- and hand-tracking, and depth sensing have made it a research instrument in its own right. Public datasets built around its native spatial-video pipeline, including a 300-clip stereoscopic dataset, a quality-of-experience dataset showing that spatial video delivers superior perceptual quality to matched bitrates, and a 600-clip egocentric spatial-video quality-assessment database, address a previously missing resource for codec and immersive-video research [71, 72, 73].

Outside the lab, the platform’s depth sensing supported forestry-inventory apps whose tree-diameter measurements closely tracked caliper references (R2R^2 up to 0.978) across varied terrain [74], and custom haptic gloves with vibrotactile and thermoelectric feedback were built around its hand-tracking to extend tactile realism in AR [75]. Vision scientists further show that classic psychophysical phenomena, such as visual crowding, can be replicated on consumer headsets for trial designs in which continuous gaze monitoring is not required [76].

A click-based gaze-visualization tool demonstrates that useful attention heatmaps can still be extracted for research purposes within the platform’s constraints on continuous access to eye-tracking data [20].

Gaming, entertainment, and immersive art and storytelling

Creative and entertainment applications exploit the headset’s passthrough, spatial audio, and unbounded interaction volume. Comparative interaction studies find the device’s gaze-pinch and direct-touch input modalities functionally equivalent for rapid target acquisition in a custom gaming task, giving developers latitude in interaction design [77].

A real-time hand-tracking boxing trainer achieved 96.3% gesture-recognition accuracy at 60 FPS across lighting conditions, extending precision sports coaching beyond gym settings [78].

Physics-driven interaction demonstrations, a dart-throwing simulator rated highly for realism and immersiveness, and a block-stacking task probing depth perception and kinetic control, show that native hand- and eye-tracking support convincing unbounded-volume physical simulation [79, 80]. Narrative and installation work extends the medium further: a validated “Point-Line” framework for MR cinema significantly improved narrative efficacy and lowered cognitive load relative to conventional immersive video [81], particle-based video-content extension into the surrounding room enhanced presence and concentration relative to a character-based baseline [82], and standalone art installations exploring embodied perception [83] and processed digital experience [84] illustrate the platform’s use as a venue for spatial art.

An exclusively built 3D multiplayer game [85] and a publicly released, theremin-inspired polyphonic instrument controlled by hand tracking [86] round out the entertainment cluster.

Productivity, collaboration, and natural spatial interaction

Workplace-oriented studies find that natural, gesture- and gaze-based interaction is readily and quickly adopted. Participants with little prior XR exposure engaged with 3D content on the headset without instruction, favoring embodied gesture and voice input over keyboard-based alternatives [87]. Systematic manipulation of immersion level found that selective attention improved as immersion increased during individual work [88], and interviews with ten long-term users found that self-reported productivity increased and users adapted to the device over time [21].

FaceTime’s spatial-persona telepresence feature was found to deliver a genuinely immersive experience at lower bandwidth than competing 2D-persona video conferencing [89], and a business-intelligence dashboard ported to visionOS was received with enthusiasm for its potential to promote new engagement patterns with enterprise data at a large industry conference [90].

An AR task-guidance framework combining voice input, large language models, and object detection likewise enhanced participant engagement during a real-world guided task [91].

Developer tooling and generative spatial-content creation

A cluster of platform-specific toolkits and creation apps shows how quickly a native content-creation ecosystem formed around the headset’s spatial APIs. Toolkits such as Palimpsest automate the coherent spatial arrangement of 3D content around tracked real-world objects [92], MediaMix built a full multimedia-retrieval interface around the device’s eye-tracking and gesture recognition for a video-retrieval competition [93], and RécitKit demonstrated that spatial data narratives, illustrated with an immersive historical flow map, support insight formation for both authors and viewers [94].

Generative-AI creation tools, including AI3D Desktop and VolumeMatic, allow text- or voice-driven 3D content generation directly in the headset [95, 96], and a related system, Dream Space, found in a 20-participant mixed-methods evaluation that this spatial-computing creation paradigm significantly reduced cognitive load relative to conventional 3D tools and let novices complete complex creative tasks within 20–30 minutes [97].

Immersive analytics applications extend this tooling to technical debugging, with computer vision and machine learning (CVML) engineers reporting that seamlessly integrating 2D and 3D visualization on the device improved their evaluation workflow for a spatial localization model [98], while a scene-aware spatial-measurement prototype [99] and an XR sketch-to-3D-generation system [100] further illustrate the breadth of native tools now built for the platform.

Memory and mixed-reality content

Beyond task performance, one study examines how AVP-mediated content is remembered. A passthrough object-viewing study identified the specific perceptual cues, namely lighting, shadow, texture, and tactile exploration, that shape how well real and virtual objects are later remembered, informing design for memorable mixed-reality content [101]. Table 2 summarizes these sub-themes.

Application areaMain findingsRefs
Surgical visualization and intraoperative guidanceWearable, wireless intraoperative display consolidating multiple feeds; high completion rates, low workload, improved navigation accuracy across ophthalmology, hepatobiliary, spine, general, urologic, and interventional-radiology procedures[9, 23, 24, 25, 26, 27, 28, 10, 29, 30, 5, 31, 32, 33, 34, 35, 36, 37]
Surgical, procedural, and clinical trainingStepwise AR overlays and MR simulators outperform lecture- or screen-based training in skill acquisition, exam scores, retention, and confidence[38, 39, 40, 41, 42, 43, 44]
Diagnostic imaging, radiology, and pathology workflowsRapid onboarding and strong perceived usability for image review, cinematic rendering, whole-slide pathology, and preprocedural AR/VR planning, with accuracy comparable to conventional tools[45, 11, 46, 47, 48, 49, 50, 51, 52]
Education and embodied cultural learningEmbodied, spatially situated interaction improves learning gains, engagement, and access to museum/heritage content over flat interfaces[53, 54, 55, 56, 57, 58, 59, 18, 60, 8]
Accessibility and assistive applicationsDepth sensing, tracking, and passthrough repurposed for fall-hazard detection, low-vision support, prosthetic control, and Deaf/hard-of-hearing accessibility research[62, 63, 64, 65, 66]
Design, architecture, and industrial engineering workflowsImmersive 1:1-scale evaluation improves spatial/proportion judgment and interface usability in architecture, transportation design, topology optimization, teleoperation, and supply-chain visualization[67, 19, 68, 69, 70]
Spatial video, novel sensing, and hardware-enabled research instrumentationNative stereoscopic capture, eye/hand tracking, and depth sensing enable new public datasets, forestry measurement, haptics, and portable psychophysics[71, 72, 73, 74, 75, 76, 20]
Gaming, entertainment, and immersive art and storytellingPassthrough, spatial audio, and unbounded interaction volume support functionally equivalent input schemes, precision sports training, physics-driven play, gestural musical instruments, and novel narrative/installation formats[77, 78, 79, 80, 81, 82, 83, 84, 85, 86]
Productivity, collaboration, and natural spatial interactionGesture- and gaze-based interaction is adopted quickly and without instruction; immersion improves selective attention; spatial telepresence and immersive analytics show early workplace promise[87, 88, 21, 89, 90, 91]
Developer tooling and generative spatial-content creationA native toolkit and generative-AI content-creation ecosystem lowers cognitive load and technical barriers for 3D authoring, data narratives, and debugging[92, 93, 94, 95, 96, 97, 98, 99, 100]
Memory and mixed-reality contentA passthrough object-viewing study clarifies which perceptual cues make mixed-reality objects memorable[101]

Table 2. The good: application areas where AVP has been validated, mostly in feasibility and pilot studies.

The bad

The limitations documented in this section are distinct in kind from the concerns raised in § 4. Where “the ugly” describes deliberate choices in how AVP is built, marketed, or used (choices that raise ethical questions about consent, surveillance, or exclusion), “the bad” describes the persistent technical and operational immaturity of a headset still in its first two hardware iterations.

These are not intentional harms: they are the ergonomic, perceptual, and software-ecosystem constraints that recur across the controlled studies, feasibility trials, and field evaluations reviewed here, and that most of these same papers frame as targets for iteration rather than as design failures.

Notably, only seven of the 104 included papers were coded as centrally organized around such limitations (with a further 11 of the 27 mixed-coded papers also touching on them as one of several findings, discussed as “mixed” in Section 2); far more report them as secondary findings inside otherwise favorable feasibility or usability studies, which is itself informative about how deeply these constraints are embedded in the current hardware generation regardless of application domain.

Ergonomic burden and physical side effects

Multiple independent evaluations converge on the same complaint: the headset’s weight and bulk become a liability over time or in professional settings. A case report of endoscopic spine surgery notes that “current AR headsets are relatively large and bulky,” which “may cause discomfort from the headset’s weight if used for long periods of time” [5]. An exploratory study of young adults using AVP within visionOS found that, despite participants’ high technology familiarity and screen-time exposure, they “experienced physical side effects early in the testing process,” prompting the authors to argue explicitly for more attention to negative physical outcomes alongside XR’s celebrated benefits [12].

Even in a largely positive feasibility pilot of AVP for diagnostic radiology, 3 of the same 25 participants (12%) reported eye strain after a 14-minute average session [45]. A mixed-methods evaluation combining structured user-experience feedback with technical assessment similarly reported measurable UX improvement over prior headsets alongside persistent comfort and ergonomic pain points that future hardware generations still need to address [102].

Input-method trade-offs and interaction limits.

AVP’s signature gaze-and-pinch interaction paradigm underperforms simpler alternatives on measurable tasks. A controlled comparison of AVP’s two native text-entry methods found Tap Typing significantly faster and more accurate than Gaze + Pinch, with participants also preferring it and rating it higher on usability and workload measures [13].

A study of automatically arranged multi-window layouts on AVP found no significant improvement over conventional arrangements and surfaced “fundamental challenges common to multi-window tasks in HMD environments, such as difficulties with text input, visual feedback, and physical strain” [103].

Vision-science researchers attempting to repurpose AVP as a portable psychophysics instrument likewise report concrete platform constraints: the device provides no access to continuous real-time eye-position data, and visionOS specifies stimulus size and position in device-defined “points” rather than degrees of visual angle, complicating direct comparison with laboratory-grade eye trackers [76]. A closely related workaround study built a click-based gaze-logging tool precisely because AVP’s privacy architecture “prevent[s] direct access to continuous user gaze data,” forcing researchers toward manual pinch- or controller-triggered capture with substantially lower sampling rates than true gaze tracking [20].

In a design-generation context, voice and freehand-sketch input on AVP were found to “lack the precision of traditional desktop workflows,” producing designer–AI intent mismatches during iterative 3D design tasks [19].

Performance and workload costs relative to established alternatives.

Several controlled comparisons quantify a real performance penalty for AVP relative to both unencumbered vision and competing headsets. In a suturing-technique study with 20 healthcare professionals, AVP produced significantly longer task completion times, significantly higher workload on the NASA Task Load Index (NASA-TLX) [104], significantly lower system usability, and significantly higher simulator sickness scores than both Microsoft HoloLens 2 and a no-headset baseline [14].

A companion study isolating video-see-through optics from optical-see-through and unencumbered vision found that manual dexterity, measured with the Purdue Pegboard Test [105], a timed test of fine motor speed using pins and washers, “significantly declined under the AVP condition across all subtests,” accompanied by higher cognitive load and post-task reports of discomfort and visual fatigue, a “hidden ergonomic cost” the authors argue designers of AR-assisted tools must actively counterbalance [106].

In a phantom-based interventional radiology comparison, AVP was rated more mentally demanding than an optical-see-through competitor, particularly among experienced operators [48]. In diagnostic radiology, the median time per case for detecting diverticulitis on CT was significantly longer on AVP (57 s) than on a conventional desktop workstation (31 s), despite equivalent diagnostic accuracy [50].

Software and application ecosystem immaturity.

Beyond individual task performance, reviewers repeatedly note that the surrounding software ecosystem has not matured to match the hardware. Interviews with ten long-term AVP users conducted more than a year after release found that, in spite of marketing claims, persistent limitations, “a lack of native applications, difficulties when integrating XR into current workflows, and limited possibilities to adapt and customize the XR experience,” continue to constrain everyday productivity use [21].

A graphics-visualization study attempting GPU-based volume rendering on visionOS found the platform’s native APIs “predominantly oriented towards surface shading,” requiring workarounds for a technique common in medical and manufacturing applications [107].

A domain review in archaeology concluded, after direct field testing, that AVP “does not yet represent a significant enough shift toward to have a major impact” on the discipline compared to prior XR hardware [15]. Pedagogically, a comparison of classical media against AVP-based eLearning found a clear preference for traditional formats for reading and writing tasks, with knowledge retention measurably lower for text-heavy material presented in the immersive environment [18] (the same comparison reported higher engagement for visually rich content in § 2, one study cutting both ways depending on content type, an example of the single-bucket-coding tension discussed in § 5.)

Connectivity, latency, and the cost of small-sample validation.

Streaming and power constraints round out the picture. A position paper on spatial video streaming to AVP identifies “bandwidth limitations, latency, and synchronization issues between multiple camera views” as unresolved obstacles to viable spatial-video delivery [108]. A medical-education deployment for laparoscopic surgery teaching reported that initial video latency of 120 ms had to be reduced through resolution trade-offs and hardware upgrades before the system was considered clinically usable, ultimately reaching under 60 ms only after additional investment in capture cards and workstations [39].

On power, an early clinical-adoption survey notes that although AVP is “untethered,” it remains tied to the user via “the cable to the battery,” a compromise this device shares with, but does not fully escape from, earlier tethered headsets [109].

Cost recurs as a stated constraint on study design rather than an equity argument: a hepatic-ablation navigation study lists “cost and fatigue” alongside its positive accuracy results as open challenges to broader deployment [26], and a volume-rendering evaluation explicitly frames AVP’s “initial high cost” as a present barrier to wider platform access for developers and researchers, one it expects to erode only as the device matures in later hardware generations [107]. Consistent with this, the great majority of clinical AVP evaluations surveyed for this review remain single-case reports or small (single-digit-to-low-double-digit) pilot cohorts, a pattern consistent with, though not conclusively caused by, the device’s price, limiting the sample sizes and statistical power that early adopters can field.

Consumer adoption research offers a complementary, demand-side view rather than additional evidence on cost. A technology acceptance model survey of 272 sport fans found that purchase intention toward AVP was shaped indirectly, through perceived usefulness and ease of use, by team identification with the sport in question [110, 111]. Like the small-sample clinical literature, this work treats willingness to buy, rather than clinical or educational outcome, as its dependent variable.

Taken together, these findings describe a device still in active technical adolescence: capable enough to enter operating rooms, classrooms, and laboratories, but not yet comfortable, precise, well-supported, or affordable enough to do so at scale without measurable costs to its users. Table 3 summarizes these limitations.

LimitationMain findingsRefs
Ergonomic burden and physical side effectsHeadset weight/bulk causes discomfort over extended use; early physical side effects reported in young, tech-familiar users; measurable eye strain even in short sessions.[5, 12, 45, 102]
Input-method trade-offs and interaction limitsNative Gaze + Pinch input is slower and less accurate than Tap Typing; multi-window text entry and visual feedback remain difficult; restricted/no continuous gaze-tracking API complicates research and precision use; voice/sketch input lacks desktop-level precision.[13, 103, 76, 20, 19]
Performance and workload costs relative to established alternativesControlled studies show AVP produces slower task completion, higher cognitive workload, lower usability, higher simulator sickness, and measurable manual-dexterity decline compared to unencumbered vision or competing headsets (e.g., HoloLens 2, desktop workstations).[14, 106, 48, 50]
Software and application ecosystem immaturityLong-term users report a lack of native apps and workflow integration; visionOS graphics APIs are not well suited to some technical tasks (e.g., volume rendering); domain reviews find the platform not yet transformative; text-heavy eLearning content shows reduced retention in XR.[21, 107, 15, 18]
Connectivity, latency, and the cost of small-sample validationSpatial-video streaming faces bandwidth, latency, and synchronization challenges; clinical deployments require substantial latency-engineering effort; the device remains tethered to an external battery; high cost is repeatedly cited as a barrier and coincides with a corpus dominated by small pilot studies and single-case reports.[108, 39, 109, 26, 107, 110]

Table 3. The bad: recurring technical and ergonomic limitations of AVP, and their main findings.

The ugly

Where the previous section cataloged limitations Apple and third-party developers are still working to solve, the ugly concerns a different register of problem: harms arising not from what the AVP cannot yet do well, but from what it already does, and from the choices developers, marketers, employers, and bystanders make around it.

These are not bugs awaiting a firmware update; they are foreseeable consequences of design and deployment decisions: a camera array that is active whenever the device is worn, biometric-grade eye tracking, a photorealistic avatar system, and a launch price that sorts who gets to use, study, or be represented in this technology at all, combined with the promotional incentives of a nascent, heavily hyped consumer category.

The literature is markedly thinner here than on the good or the bad: of 104 primary-focus AVP papers surveyed, only two were coded as centrally organized around misuse, risk, or exclusion (a further 6 of the 27 mixed-coded papers also touch on these concerns without them being the paper’s stated primary contribution), and fewer still document realized rather than anticipated harm. That imbalance is itself part of the story, taken up explicitly below.

Gaze as an attack surface. Apple markets AVP’s eye-tracking-driven gaze-and-pinch input as a privacy-preserving improvement over hand- or acoustic-based interaction, since raw gaze coordinates are not exposed to third-party apps. GAZEploit shows this protection can be circumvented: by analyzing eye movements rendered onto a user’s virtual Persona avatar during video calls or screen sharing, an attacker can remotely reconstruct what a wearer is typing on a gaze-controlled virtual keyboard, and the authors identify more than 15 top-rated App Store applications vulnerable to the technique [16].

The feature meant to make the avatar feel present and expressive, its rendering of the wearer’s eyes for others to see, becomes the side channel that leaks credentials and private text. Apple patched the specific vulnerability that GAZEploit exploited in visionOS 1.3, released on 29 July 2024 (CVE-2024-40865, affected component: Presence), by suspending Persona rendering while the virtual keyboard is active [112].

The fix closes the one-channel GAZEploit demonstrated; it does not resolve the broader architectural tension the finding exposes, since AVP’s privacy model still withholds continuous gaze data from third-party developers while rendering the wearer’s eye movement outward through Persona in every other context. That a narrow, reactive patch was needed at all, and that it took independent academic security research, not an internal audit, to surface the exposure, is itself evidence for the § 5 point that this kind of finding is currently rare precisely because so little independent security research on AVP exists.

Researchers building iTrace, a click-based gaze-visualization tool, show that even without continuous gaze access, user-triggered gaze samples (captured with a pinch or game-controller click) are sufficient to reconstruct attention heatmaps. They restrict their tool to research use in acknowledgment of how sensitive the recovered data is [20].

Early-adopter reviews echo the concern: privacy is one of the few affordances adopters describe as actively discouraging rather than encouraging adoption [113].

Simulated presence, real absence. AVP’s Persona avatars and its EyeSight display, which projects a digital rendering of the wearer’s eyes outward so bystanders can see where they are looking, are designed to preserve social presence while the wearer is immersed. A controlled comparison shows that the feature improves perceived social acceptability and presence relative to a headset without it, yet participants still rated it as falling short of unmediated face-to-face interaction [114].

Read as a societal-harm question rather than a usability one, this is a notable admission: a technology marketed as compatible with ordinary social connection instead substitutes a simulacrum of eye contact for the real thing, and does so most visibly in exactly the shared, co-present situations where genuine attention matters most.

Hype rhetoric and the suppression of critical scrutiny

A recurring finding is that discourse around AVP has itself shaped how critically the device gets evaluated. A rhetorical analysis of YouTube reviews finds that anticipatory communities around the launch converged on a small set of strategies, namely excitement, exploration, exemplification, and exaggeration, that intensified anticipation while constraining more critical evaluation once the device shipped [17]. Long-term users interviewed more than a year after launch push back against this rhetoric directly, describing persistent gaps between marketing’s productivity promises and what everyday use actually delivers [21].

Sentiment analysis of scraped online commentary likewise reports “substantial variability” in real-world reception rather than the consensus enthusiasm implied by early coverage [115]. None of this is a technical failing of the headset; it is a pattern of promotional communication that a premium, still-young category is especially prone to, and it matters ethically because it delays the public’s ability to reason clearly about the device’s risks.

Price as exclusion, not just a barrier.

The bad section treats AVP’s price as a technology-adoption obstacle; read differently, the same fact is a structural equity problem. Market-behavior research applying the AIDA model [116], a marketing framework tracking attention, interest, desire, and action toward a purchase, finds purchase intentions above 75% among high-income users, while price sensitivity substantially limits adoption among middle-income users [6], meaning the population able to buy, use, and be represented in AVP research is itself skewed toward the affluent.

Because most of the “good” evidence in this corpus derives from small convenience samples of people who already own or can access an AVP, price does not merely slow diffusion; it shapes whose needs the emerging literature is even in a position to study.

A conspicuously thin evidence base.

Finally, the shape of the corpus is itself a finding. Only two of the 104 included papers were coded as primarily ugly in focus, compared with 68 coded as good; most published AVP research to date consists of feasibility and pilot studies validating benefits in medicine, education, and design, rather than investigations of risk, misuse, or exclusion. We found no dedicated study in this corpus of, for example, employer or institutional surveillance uses of the headset, despite seven papers examining productivity and enterprise deployment, a gap worth naming rather than papering over with speculation.

Several of the concerns above resist being cleanly filed under either "bad" or "ugly". Price can be read as a temporary technical-adoption barrier (bad) or as a structural exclusion that determines who gets to benefit from and be represented in AVP research (ugly), depending solely on whether the observer’s lens is the manufacturer’s cost curve or the resulting research population. Diminished social presence is similarly double-faced: a limitation of current avatar and passthrough fidelity (bad) and, at the same time, a societal-harm consequence of choosing to mediate embodied co-presence through a commercial device at all (ugly).

The gaze-tracking privacy architecture is a third such case: withholding continuous gaze data from developers while rendering eye movement outward through Persona can be read as an unfinished engineering compromise (bad) or as a considered trade-off favoring avatar expressiveness over wearer privacy (ugly). We do not resolve these ambiguities here; we return to them, without resolving them, in § 5. Table 4 summarizes these issues.

ConcernMain findingsRefs
Gaze as an attack surfaceEye-tracking and Persona-avatar rendering intended to improve privacy and expressiveness instead create a side channel usable for remote keystroke inference; even sanctioned, user-triggered gaze sampling yields sensitive attention data. Apple patched the specific keystroke-inference channel in visionOS 1.3 (CVE-2024-40865); the underlying outward-rendering design is unchanged.[16, 112, 20, 113]
Simulated social presenceEyeSight and Persona substitute a digitally rendered proxy for direct human contact; even where it improves perceived presence relative to no proxy, it remains short of face-to-face interaction, reframing isolation as a societal cost of mediated presence.[114]
Hype-driven suppression of scrutinyPromotional rhetoric surrounding AVP’s launch and marketing shapes public and user discourse, constraining critical evaluation of the device’s risks and overstating its everyday productivity value.[17, 21, 115]
Price as structural exclusionPurchase intention and adoption are sharply stratified by income, meaning the device, and the research literature built on it, systematically underrepresents lower- and middle-income users.[6]
A conspicuously thin evidence baseOnly 2 of 104 included papers were coded as centrally focused on misuse/risk/ethics, versus 68 on validated benefits, 7 on technical/ergonomic limitations, and 27 spanning more than one register; documented (as opposed to anticipated) harms, and topics such as workplace surveillance, remain largely unstudied.This review (Table 1)

Table 4. The ugly: ethical concerns and structural harms arising from how AVP is built, priced, and deployed.

Discussion

Read as a whole, the catalog in § 2 to § 4 says as much about the shape of AVP research to date as it does about AVP itself. Three patterns recur across all three sections and are worth naming explicitly.

First, the corpus comprises early adopters studying early adoption. The great majority of included papers are single-case reports, small pilot cohorts, or short controlled studies run by teams that already had access to a device that launched at $3,499 and has since risen to $3,699, a constraint several papers cite directly as limiting their generalizability.

This is not a flaw specific to any one study; it is a structural feature of research on an expensive, single-vendor artifact in its first three years of availability, and it means the good cataloged in § 2, however consistent across specialties, should be read as evidence of feasibility rather than of population-level effectiveness.

Relatedly, our own coding rule (§ 1.3) codes a study by its stated aim rather than its statistical outcome, which means a handful of null or mixed results (e.g. [8]) are counted toward the good total above; this is a second, self-inflicted source of upward bias in the same direction as the early-adopter sampling problem, and future coding of this literature should consider outcome-weighted alternatives.

Second, the field has so far allocated its attention asymmetrically. Sixty-eight of 104 included papers were coded as centering on a validated benefit; two on ethical risk or misuse; a further seven on a technical/ergonomic limitation; and 27 as mixed, spanning more than one register (Table 1). We do not read this as evidence that AVP is safer than other emerging technologies: published feasibility and usability studies are the wrong instrument for detecting structural harms in the first place, since those harms often look identical to ordinary technical immaturity until someone deliberately investigates whose interests a given design choice serves.

What the asymmetry does tell us is where the marginal research contribution is largest: independent security research (of which GAZExploit is so far a rare example [16]), bystander-centered study designs (rather than wearer-centered usability studies), and research that treats non-owners and non-users as a population worth studying, not just an absence in the sampling frame.

Third, several of the issues cataloged above do not cleanly resolve into bad or ugly using the kinds of evidence current studies collect. Price is a case in point: § 3 treats it as an adoption barrier that constrains sample sizes across the corpus, one that later hardware generations may ease; § 4 treats the same figure as a stratifying mechanism, since purchase intention exceeds 75% among high-income users while price sensitivity suppresses adoption in the middle-income segment, skewing who is represented in AVP research [6].

Both readings are consistent with the same published data, because only Apple knows whether the current price reflects a declining cost curve or a deliberate premium-positioning choice.

The price trajectory observed since launch is suggestive on this point: AVP held at $3,499 through the October 2025 M5 hardware refresh, then rose to $3,699 (256GB), $3,899 (512GB), and $4,199 (1TB) in June 2026, a change Apple attributed to rising memory and storage component costs rather than to a product refresh [117]. A flat-then-rising trajectory is evidence against, not for, the “declining cost curve” reading that would ease the adoption barrier § 3 describes; two years in, price has moved in the direction the structural-exclusion reading in this section anticipates, not the direction the technical-adoption-barrier reading in § 3 would predict.

Simulated social presence shows the same pattern: EyeSight and Persona measurably fall short of face-to-face interaction, which can be read as a limitation of current sensor and rendering technology or as a designed feature of mediating co-presence through a proprietary avatar system at all [114], and nothing in the usability data distinguishes a temporary shortfall from a permanent trade-off.

The gaze-tracking privacy architecture raises a related tension: no continuous gaze-position API is exposed to third-party apps or researchers, yet the same architecture renders the wearer’s eye movements into the outward-facing Persona stream, the exact channel that GAZEploit exploits to reconstruct a wearer’s typed passwords [16].

Apple’s own response tells us how it handled this one instance, a targeted fix in visionOS 1.3 that suspends Persona during keyboard entry rather than a redesign of what gaze data Persona renders outward at all, but it does not tell us whether the underlying decision to render gaze onto Persona in every other context is a considered trade-off privileging avatar expressiveness or an oversight nobody has yet been paid to look for again; that remains an internal decision no published study can observe.

What would resolve these cases is not another usability study but disclosure: a component-level roadmap from Apple, or an independent regulatory and security review, that converts an inference about intent into a fact.

These patterns suggest three concrete directions for future work: risk- and ethics-focused primary research, commissioned rather than incidental to studies designed for other purposes; bystander- and equity-centered study designs that do not require owning or wearing the device to participate; and continued attention, in future reviews of this literature, to findings like the ones above where the same published result is genuinely compatible with more than one reading, since resolving them depends on disclosure the studies themselves cannot supply.

Limitations.

This review has its own methodological limits worth stating plainly. We searched a single database (Scopus); given how much of this corpus is ACM- and IEEE-published conference material, indexing gaps relative to the ACM Digital Library or IEEE Xplore directly are a real coverage risk rather than a formality, and a multi-database replication would be a natural next step.

Screening was performed by two reviewers, one of whom screened every record and one of whom checked those decisions and ran a separate blind pass, but the eight borderline records were resolved by discussion, and although the blind pass would permit it, no inter-rater agreement statistic was computed.

The two papers retained as judgment calls under criterion 2 (§ 1.2), each comparing AVP with a no-headset baseline and one other headset, would be removed under a stricter single-device rule.

Screening was conducted at the title/abstract/document-type level rather than at the full-text level, consistent with the scope of an emerging-technology rapid review but a step short of a full systematic review. We did not conduct a quality appraisal of individual included studies (e.g., risk-of-bias scoring) or a meta-analysis, both of which the largely single-case and small-pilot nature of this literature would, in any case, make difficult to interpret.

Finally, restricting the corpus to English-language records excludes any non-English primary research on AVP that may exist.

Conclusion

We have reviewed the primary-focus academic literature on Apple Vision Pro published between its 2024 launch and 9 October 2026, screening 218 Scopus records down to 104 papers and organizing them using a good/bad/ugly structure adapted from Sergio Leone’s tripartite framing in The Good, the Bad, and the Ugly.

The good is substantial and consistent: across surgery, diagnostic imaging, education, accessibility, design, and research instrumentation, AVP has repeatedly proven effective, often on a team’s first attempt.

The bad is equally consistent and, on the evidence, actively being engineered against: weight and comfort, input-method trade-offs, workload costs relative to established alternatives, and an immature software ecosystem recur across otherwise unrelated studies.

The ugly is real but thin: gaze-based attack surfaces, income-stratified access, and hype rhetoric that outpaces critical scrutiny, and its thinness is itself a finding about where the field’s attention has not yet gone.

Beyond the catalog, we have shown that some of the issues it documents do not resolve into a single verdict from the published evidence alone: price, simulated social presence, and the gaze-tracking privacy architecture each admit both a “bad” (technical-immaturity) and an “ugly” (structural-choice) reading that the data cannot adjudicate between. We treat this indeterminacy as a substantive finding rather than a shortcoming of our own analysis, and as a marker of where independent disclosure, not further usability testing, is what the field most needs next.

Together, the catalog and this observation are offered not as a final verdict on whether Apple Vision Pro is good, bad, or ugly, but as an account of why that question, for this kind of technology, is one the published literature alone cannot yet settle, and a map of what would need to change, in the evidence and in the disclosure available to researchers, for it to.

Data availability The Scopus export underlying the screening reported here: https://doi.org/10.5281/zenodo.23669705, the screening decisions are available in section § 1.1.

Declaration of competing interest

The authors declare no conflicts of interest.

References

References

  1. [1]S. Leone, Il buono, il brutto, il cattivo [The Good, the Bad and the Ugly]. Film, Produzioni Europee Associate / United Artists (1966).
  2. [2]S. Greenwold, Spatial Computing, B.S./M.Eng. thesis, Massachusetts Institute of Technology (2003). URL https://acg.media.mit.edu/people/simong/thesis/SpatialComputing.pdf
  3. [3]Apple Inc., Apple Vision Pro - Tech Specs, Apple Support, Original (M2) model; device weight 21.2–22.9 ounces (600–650 g), varying with Light Seal and headband configuration (Solo Knit Band or Dual Loop Band) (2024). URL https://support.apple.com/en-us/117810
  4. [4]Apple Inc., Apple Vision Pro upgraded with the M5 chip and Dual Knit Band, Apple Newsroom, announced 15 October 2025; available 22 October 2025 (2025). URL https://www.apple.com/newsroom/2025/10/apple-vision-pro-upgraded-with-the-m5-chip-and-dual-knit-band/
  5. [5]D. Park, S. Park, S. Hashmi, Y.-P. Lee, N. Bhatia, M. Oh, Enhancing endoscopic spine surgery with intraoperative augmented reality: A case report, International Journal of Surgery Case Reports 131 (2025) Art. 111342. doi:10.1016/j.ijscr.2025.111342.DOI
  6. [6]P. He, T. Cui, Exploration on the Application of Head-Mounted Display Devices Based on the AIDA Model, Proceeding of the International Conference on Intelligent Engineering and Next-Gen Healthcare Systems (IEHNS 2026) (2026). doi:10.1109/IEHNS68708.2026.11607603.
  7. [7]M. Page, J. McKenzie, P. Bossuyt, I. Boutron, T. Hoffman, C. Mulrow, et al., The PRISMA 2020 statement: an updated guideline for reporting systematic reviews, BMJ 372 (2021) n71. doi:10.1136/bmj.n71.DOI
  8. [8]M. Oehler, L. Bruns, B. Saurbier, T. Voong, Effects of Instructional Modality on Performance Anxiety and Stress in Sight-Singing, 2025 IEEE 6th International Symposium on the Internet of Sounds, IS2 2025 (2025). doi:10.1109/IS264627.2025.11284581.DOI
  9. [9]T. Korn, C. Santiago, Y. Fuller, B. Said, Apple Vision Pro as the Intended Primary Intraoperative Visualization Platform for Complete Cataract Surgery: A Prospective Single-Center Feasibility Study, Clinical Ophthalmology 20 (2026) 1–16. doi:10.2147/OPTH.S633394.
  10. [10]Z. Liu, J. Gu, Y. Zhang, L. Bai, Y. Guan, B. Ni, H. Zhang, M. A imaiti, P. Zhang, X. Shen, S. Wang, B. Yue, X. Xia, Z. Zhang, Total laparoscopy-assisted total gastrectomy with Da Vinci robotic system conducted by RENJI-GISH: a preliminary clinical study and case report, World Journal of Surgical Oncology 23 (1) (2025) Art. 86. doi:10.1186/s12957-025-03735-z.
  11. [11]G. Luijten, L. Faray de Paiva, S. Krueger, A. Brost, L. Mazilescu, A. Santos, P. Hoyer, J. Kleesiek, S.-T. Schmitz, U. Neumann, J. Egger, From screen to space: evaluating Siemens’ Cinematic Reality application for medical imaging on the Apple Vision Pro, Frontiers in Virtual Reality 6 (2026) Art. 1666614. doi:10.3389/frvir.2025.1666614.DOI
  12. [12]P.-E. Chontas, A. Iftene, S.-C. Buraga, Evaluating the Effects XR Has on Users: An Exploratory Study, Communications in Computer and Information Science, 2371 CCIS (2025) 150–167doi:10.1007/978-3-031-83845-3_9.DOI
  13. [13]+ Pinch, Proceedings of the Human Factors and Ergonomics Society 69 (1) (2025) 1848–1853. doi:10.1177/10711813251367736.
  14. [14]H. Javaheri, V. F. Rey, P. Lukowicz, G. Stavrou, J. Karolus, O. Ghamranejad, Assessing the Feasibility of Using Apple Vision Pro While Performing Medical Precision Tasks: Controlled User Study, JMIR XR and Spatial Computing 2 (2025) e73574. doi:10.2196/73574.DOI
  15. [15]H. Azizbekyan, P. J. Cobb, E. Cobb, The apple vision pro: Useful mixed/augmented reality (mr/ar) headset for archaeology or not quite there yet?, Advances in Archaeological Practice 13 (2) (2025) 346–355.DOI
  16. [16]H. Wang, Z. Zhan, H. Shan, S. Dai, M. Panoff, S. Wang, GAZEploit: Remote Keystroke Inference Attack by Gaze Estimation from Avatar Views in VR/MR Devices, CCS 2024 – Proceedings of the 2024 ACM SIGSAC Conference on Computer and Communications Security (2024) 1731–1745doi:10.1145/3658644.3690285.DOI
  17. [17]R. Shivener, J. Tham, Hypercollaboration: Shaping Public Engagement with Emerging Technologies Through Anticipatory Communities and Embodiment, Rhetoric Society Quarterly (2026). doi:10.1080/02773945.2026.2684344.DOI
  18. [18]P.-E. Chontas, G.-G. Constantinescu, Classical Media Versus Extended Reality in eLearning: A Study on User Engagement and Knowledge Retention on Apple Vision Pro, Communications in Computer and Information Science 2834 CCIS (2026) 551–563doi:10.1007/978–3–032–16448–3_31.DOI
  19. [19]Y.-H. Lee, Investigating Designer Interaction with AI Text-to-Image Systems in Vision Pro for Transportation Design, Lecture Notes in Computer Science 16717 LNCS (2026) 107–120doi:10.1007/978–3–032–31421–5_7.DOI
  20. [20]E. Mehmedova, S. Berzueeta-Guzman, S. Wagner, iTrace: Click-Based Gaze Visualization on the Apple Vision Pro, Human-Centric Intelligent Systems 6 (2) (2026) 269–282. doi:10.1007/s44230–026–00143–8.DOI
  21. [21]V. Biener, F. Winston, D. Schmalstieg, A. Plopski, Long-Term Experiences from Working with Extended Reality in the Wild, Proceedings - 2025 IEEE International Symposium on Mixed and Augmented Reality, ISMAR 2025 (2025) 667–676. doi:10.1109/ISMAR676309.2025.00076.
  22. [22]Apple Inc., Apple Vision Pro (M5) - Tech Specs, Apple Support, M5 model; device weight 26.4–28.2 ounces (750–800 g), including Light Seal and counterbalanced Dual Knit Band (2025). URL https://support.apple.com/en-us/125436
  23. [23]N. El Jalbout, R. Bansal, M. Shoji, S. Cheng, A. D’Urso, H. Hosalkar, C. Liu, H. Dediu, T. Korn, D. Kikkawa, R. Broderick, B. Korn, Apple Vision Pro as a wearable display for endoscopic dacryocystorhinostomy: Surgical experience, efficiency, and cost, AJO International 3 (3) (2026) Art. 100288. doi:10.1016/j.joint.2026.100288.DOI
  24. [24]R. Mastropasqua, M. Ruggeri, A. Quarta, R. Tartaro, L. Vecchiarino, L. Corboli, L. Brescia, M. Orione, A. Russo, T. Avitabile, L. Mastropasqua, New Mixed Reality Headset: First Exploratory Use in Intraocular Surgery and Telementoring, Ophthalmology and Therapy 14 (7) (2025) 1597–1609. doi:10.1007/s40123–025–01117–y.DOI
  25. [25]M. Orione, G. Rübegni, R. Tartaro, A. Alberghina, M. Fallico, C. Orione, A. Russo, G. Tosi, T. Avitabile, Utilization of apple vision pro in ophthalmic surgery: A pilot study, European Journal of Ophthalmology 35 (2) (2025) 715–721. doi:10.1177/11206721241273574.DOI
  26. [26]T. Lan, S. Liu, Y. Dai, J. Luo, J. Han, Y. Jin, Real-time image fusion and Apple Vision Pro in laparoscopic microwave ablation of hepatic hemangiomas, npj Precision Oncology 9 (1) (2025) Art. 79. doi:10.1038/s41698–025–00867–z.DOI
  27. [27]S. Abu Zahra, A. Riaz, A. Kundu, R. Lewandowski, A. Gordon, Augmented reality-assisted cholangioscopy for biliary interventions: initial clinical experience in four cases, Surgical Endoscopy 40 (8) (2026) 7189–7196. doi:10.1007/s00464–026–12032–w.DOI
  28. [28]Y.-C. Tsai, C.-C. Hsiao, C. Chung-Wei Lin, M.-C. Chern, S.-W. Huang, Apple Vision Pro-guided Laparoscopic Radio Frequency Ablation for Liver Tumors: The Pioneer Experience, Surgical Innovation 32 (3) (2025) 312–314. doi:10.1177/15533506251316001.DOI
  29. [29]H. Ma, C. Sun, Z. Li, J. Chen, R. Li, Z. Wang, Y. Tian, T. Zhu, Z. Qian, Augmented Reality-Assisted Guide Plate Positioning in Lumbar Surgery Using Apple Vision Pro, Proceedings - 2025 IEEE VRW (2025) 1272–1273. doi:10.1109/VRW66409.2025.00282.DOI
  30. [30]H.-J. Hanks, M. Kim, R. Lin, V. Chen, A. Ton, E. Mills, H.-H. Wu, S. Hashmi, Y.-P. Lee, N. Bhatia, W. Liawrungreuang, M.-H. Wu, J.-W. Hur, D. Park, Incorporating mixed reality head mounted display technology in biportal endoscopic lumbar surgery: an early feasibility study, Frontiers in Surgery 13 (2026) Art. 1772853. doi:10.3389/fsurg.2026.1772853.DOI
  31. [31]R. Broderick, G. Spurzern, J. Jeffery Reeves, H. Hollandsworth, B. Sandler, G. Jacobsen, C. Longhurst, S. Horgan, First use of augmented reality headset in minimally invasive general surgery: seeing is believing, Surgical Endoscopy 39 (9) (2025) 6055–6060. doi:10.1007/s00464-025-11985-x.DOI
  32. [32]R. Neto, F. Vicentini, R. Ugino, A. Danilovic, G. Marchini, F. Torricelli, C. Batagello, A. Pellanda, A. Silva, W. Nahas, E. Mazzucchi, Mixed Reality Ultrasound-Guided Mini-ECIRS with Apple Vision Pro - First Case Report, International Braz J Urol 51 (2) (2025) Art. e20240610. doi:10.1590/S1677-5538.IBJU.2024.0610.
  33. [33]A. Radevic, J. Austerlitz, R. Gilbertson, S. Krishna, M. Colello, D. Johnson, R. Singh, A. Perez, J. Lawrenz, M. Topf, Novel Intraoperative Augmented Reality-Guided Specimen Localization in Chest Wall Oncologic Surgery: A Case Report, Journal of Medical Extended Reality 3 (2026). doi:10.1177/29941520261426873.DOI
  34. [34]B. Xu, W. Wu, Co-Designing a Multimodal AR Assistant for Interventional Radiology Workflows, Conference on Human Factors in Computing Systems - Proceedings (2026) Art. 169. doi:10.1145/3772363.3799005.DOI
  35. [35]A.-N. Vo, D. Subramani, J. Schmidt, J. Arensmeyer, P. Feodorovici, A Granular Interaction Framework for Surgical Robotic Digital Twins: Enabling Realistic Segment-Level Manipulation in Extended Reality, International Journal of Computer Assisted Radiology and Surgery (2026). doi:10.1007/s11548-026-03677-1.DOI
  36. [36]S. Nagano, K. Momose, T. Kato, Y. Ishii, S. Yamaguchi, T. Hagi, K. Yamashita, T. Saito, K. Tanaka, T. Makino, T. Takahashi, Y. Kurokawa, H. Eguchi, Y. Doki, K. Nakajima, Immersive endoscopy: potential usefulness from an ergonomic perspective, Minimally Invasive Therapy and Allied Technologies (2026). doi:10.1080/13645706.2026.2693303.
  37. [37]C. Günther, D. Winkler, S. Jung, M. Hoffmann, K. Scheidt, F. Kropla, E. Güresir, R. Grunert, Next viewer-development of a software for the spatial computing data glasses Apple Vision Pro for patient-specific preoperative planning and navigation, Frontiers in Digital Health 8 (2026) 1–11. doi:10.3389/fdgth.2026.1764807.DOI
  38. [38]M. Ishikhi, J. Yamanishi, K. Kinoshita, S. Mitani, Development and Evaluation of a Stepwise Learning Support System for Surgical Education Using Augmented Reality, IEEE Transactions on Electronics, Information and Systems 146 (7) (2026) 664–671. doi:10.1541/ieejeiss.146.664.
  39. [39]K. Shen, C. Yang, C. Wang, X. Ge, J. Xing, B. Sun, J. Yu, Z. Xie, Exploration and practice of Apple Vision Pro in medical education: application in laparoscopic surgery teaching, Chinese Journal of Digestive Surgery 24 (12) (2025) 1566–1571. doi:10.3760/cma.j.cn115610-20251011-00620.
  40. [40]M. Golledge, K. Seymour, M. Seymour, S. de Graaf, Interactive Mixed Reality Simulation Enhances Student Knowledge and Ultrasound Interpretation in Sheep Pregnancy Diagnosis, Veterinary Sciences 13 (1) (2026) Art. 80. doi:10.3390/vetsci13010080.DOI
  41. [41]W. Jayawardene, R. Magnuson, C. Kumbalatara, M. Kase, A. Park, A. Goodson, A. Barrows, R. Bolinski, J. Willett, Usability of Mixed Reality for Naloxone Training: Iterative Development and Field Testing of ReviveXR, Healthcare (Switzerland) 13 (12) (2025) Art. 1449. doi:10.3390/healthcare13121449.DOI
  42. [42]C. Dandu, B. Creisher, M. Davis, M. Lin, S. Schenke, E. Aboian, Integration of high-resolution mixed reality models into resident training, Journal of Vascular Surgery Cases, Innovations and Techniques 12 (2) (2026) Art. 102133. doi:10.1016/j.jvscit.2026.102133.DOI
  43. [43]T. Stauffer, B. Gutier, R. Basht, Q. Lohmeyer, M. Meboldt, Augmented reality enhanced physical simulation for femoral nailing training, International Journal of Computer Assisted Radiology and Surgery (2026). doi:10.1007/s11548-026-03646-8.DOI
  44. [44]E. Tomiuk, J. Miró, L. Duong, Evaluation of eye-tracking capabilities in Apple Vision Pro for training in hybrid ventricular septal defect procedures, Progress in Biomedical Optics and Imaging - Proceedings of SPIE 13408 (2025) Art. 134080K. doi:10.1117/12.3047251.DOI
  45. [45]J. Volin, V. Viswanathan, E. Krupinski, N. Safdar, R. Silverman, V. Narayan, E. Kikano, Feasibility of Using the Apple Vision Pro for Diagnostic Radiology: User Experience and Perceived Clinical Utility, Journal of Imaging Informatics in Medicine (2025). doi:10.1007/s10278-025-01745-7.DOI
  46. [46]J. Veerla, P. Guttikonda, H. Shang, M. Nasr, C. Torres, J. Luber, Beyond the Monitor: Mixed Reality Visualization and Multimodal AI for Enhanced Digital Pathology Workflow, Proceedings - International Symposium on Biomedical Imaging 2026-April (2026). doi:10.1109/ISBI61048.2026.11515391.DOI
  47. [47]S. Zahra, M. Antalek, A. Kundu, A. Nadeem, S. Schenke, K. Finos, N. Chatterjee, U. Bagci, R. Lewandowski, A. Gordon, Augmented and Virtual Reality for Preprocedural Planning in Hepatic Transarterial Interventional Oncology: A Pilot Study in 50 Cases, Journal of Vascular and Interventional Radiology 37 (11) (2026) Art. 108979. doi:10.1016/j.jvir.2026.108979.DOI
  48. [48]A. Jacob, A. Böhner, A. Henkel, T. Aziz-Safaie, L. Oelmeier, N. Lenzen, A.-M. Odenthal, M. Bahaeldin, A. Schmidt, D. Kurt, J. Sieber, L. Bischoff, Y. Layer, D. Kravchenko, M. Voigt, P. Kupczyk, T. Dell, N. Mesropyah, A. Isaak, C. Pieper, J. Leutkens, D. Kuetting, Optical See-Through and Video See-Through Head-Mounted Displays for Percutaneous Biopsies: A Comparative Phantom Study, CardioVascular and Interventional Radiology 49 (5) (2026) 970–979. doi:10.1007/s00270-026-04660-3.DOI
  49. [49]N. Tregobov, T. Jiang, P. Edgcumbe, A. Alhuwaymil, J. Yan, A. Aludaini, P. Munk, H. Ouellette, P. Mallinson, A. Sheikh, Augmented Reality and Ultrasonography-Guided Needle Placement: Feasibility and Accuracy in a Phantom Model, Journal of Diagnostic Medical Sonography 42 (4) (2026) 293–297. doi:10.1177/87564793261418020.DOI
  50. [50]P. Murphy, J. An, L. Wojdyla, A. Searleman, A. Khurana, T. Loehfem, K. Fowler, C. Longhurst, Diagnostic Performance of a Next-Generation Virtual/Augmented Reality Headset: A Pilot Study of Diverticulitis on CT, Journal of Imaging Informatics in Medicine 38 (5) (2025) 3271–3278. doi:10.1007/s10278-024-01292-7.DOI
  51. [51]H. Kawaguchi, H. Yamasaki, AR-Guided Probe Positioning System for fNIRS Using a Head-Mounted Display, Progress in Biomedical Optics and Imaging - Proceedings of SPIE 13835 (2026) Art. 1383506. doi:10.1117/12.3080193.DOI
  52. [52]M. Masalkhi, J. Ong, E. Waisberg, E. Dervan, A. Lee, Apple Vision Pro’s new technology as a head-mounted perimetry device for glaucoma and other potential applications, Pan-American Journal of Ophthalmology 6 (3) (2024) Art. 89. doi:10.4103/pajo.pajo_32_24.DOI
  53. [53]K. Chen, T. Zang, Z. Cao, J. Zheng, G. Sun, H. Li, ArtificialLive: Mixed Reality Artifact and Embodied Learning, Conference on Human Factors in Computing Systems - Proceedings (2026) Art. 96. doi:10.1145/3772363.3799038.DOI
  54. [54]L. Xia, X. Li, Y. Qin, D. Li, L. Fan, Enhancing Design Education Through Spatial Computing: A Comparative Study of Traditional and Immersive Technologies in Chair Design Projects, Proceedings - Frontiers in Education Conference (2024) FIEdoi:10.1109/FIE61694.2024.10893086.DOI
  55. [55]S. Lee, J. Choi, H. Bae, D. Lee, J. Kim, S. Kim, Y. Choi, VisionStorage Classroom: MR-Based Educational System for MMCA Cheongju Korea, Proceedings - 2025 IEEE International Symposium on Mixed and Augmented Reality Adjunct, ISMAR-Adjunct 2025 (2025) 968–970. doi:10.1109/ISMAR-Adjunct68609.2025.00271.DOI
  56. [56]D. Kim, M. Jannay, W. Choi, S. Kim, J. Kim, Lee Jungseop XR: A XR Documentary for Apple Vision Pro, Proceedings - 2024 IEEE ISMAR-Adjunct (2024) 650–651 doi:10.1109/ISMAR-Adjunct64951.2024.00194.DOI
  57. [57]M. Vlachos, S. Konstantinidis, A. Dzhrathatspanian, M.-Y. Kan, Conversing with the Past: Immersive Access to Cultural Digital Archives with LLMs and Spatial Interfaces, International Conference on Intelligent User Interfaces, Proceedings IUI (2026) 143–147 doi:10.1145/3742414.3794770.DOI
  58. [58]S. Patil, V. Lalwani, B. Mahmud, S. Carr, J. Woodcock, K. Grellinger, G. Hong, A Mixed-Reality Approach to Cardiovascular Anatomy Education, Future Internet 18 (6) (2026) Art. 314. doi:10.3390/fi18060314.DOI
  59. [59]K. Lim, A. Liang, Y. Fang, B. Kuok, Annotating the Field: Investigating the Affordances of Mixed Reality for Learning Beyond the Classroom, Virtual Worlds 4 (2) (2025) Art. 23. doi:10.3390/virtualworlds4020023.DOI
  60. [60]R. Al-Maroof, R. Tawafak, W. Al-Rahmi, K. Alhashmi, I. Alyoussef, Predicting the actual use of artificial intelligence features of Apple Vision Pro using PLS-SEM, Contemporary Educational Technology 17 (3) (2025) ep580. doi:10.30935/cedtech/16208.DOI
  61. [61]J. Hair, G. Hult, C. Ringle, M. Sarstedt, A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), 3rd Edition, Sage, 2022.
  62. [62]M. Koenig, T. Waechter, Can Mixed Reality Support Fall Prevention?: Assessing the Feasibility of Mixed Reality for Trip Hazard Detection, Proceedings of the 22nd International Conference on Intelligent Environments (2026) IE 2026 doi:10.1109/IE69249.2026.11539054.DOI
  63. [63]K. Zhang, Q. Li, J. Zhang, Evaluating the Effects of Multiple Environmental Risk Factors on Older Adults’ Walking Safety: An Exploratory Virtual Reality Experiment, Lecture Notes in Computer Science 16729 LNCS (2026) 425–442 doi:10.1007/978-3-032-30044-7_26.DOI
  64. [64]B. Gamage, N. Mcdowell, D. Kovacic, L. Holloway, T.-T. Do, A. Lowery, N. Price, K. Marriott, Smart Glasses for CVI: Co-Designing Extended Reality Solutions to Support Environmental Perception by People with Cerebral Visual Impairment, ASSETS 2025 (2025) Art. 2. doi:10.1145/3663547.3746383.DOI
  65. [65]B. Guan, Z. Wang, R. Godoy, M. Owen, M. Liakopoulos, Combining Augmented Reality With Semi-Autonomous, Lightography-Based Control to Improve Usability of Prostheses, IEEE Access 14 (2026) 50537–50548. doi:10.1109/ACCESS.2026.3677129.DOI
  66. [66]R. Mathew, K. Wu, W. Dannels, An In-the-Wild Accessibility Evaluation of Apple Vision Pro for Deaf or Hard of Hearing Users, Proceedings of the ACM Symposium on Virtual Reality Software and Technology, VRST (2025) Art. 21 doi:10.1145/3756884.3765997.DOI
  67. [67]T. Tay, L. Liang, D. Nizam, C. Cai, G. Lipkowitz, Development and User Evaluation of Multimodal Spatial Computing Applications for 3D Generative Workflows in Early-Stage Architectural Design Processes, Proceedings of the International Conference on Computer-Aided Architectural Design Research in Asia 1 (2026) 549–558.DOI
  68. [68]A. Aragón, H. Algra, ARCADE: an interactive playground for immersive topology optimization, Structural and Multidisciplinary Optimization 68 (12) (2025) Art. 251. doi:10.1007/s00158-025-04152-z.DOI
  69. [69]H. Lied, T. Sitompul, O. Alsos, Remote Teleoperating ROV Using Apple Vision Pro: A Study on Usability and Workload, Proceedings of VRCAI 2025 (2026) Art. 6 doi:10.1145/3779457.DOI
  70. [70]J. Andersson, G. Cenk, T. Engel, Evaluating the User Experience of an Immersive Digital Supply Chain Twin: A Mixed-Methods Approach, Communications in Computer and Information Science 2528 CCIS (2025) 259–267 doi:10.1007/978-3-031-94168-9_25.
  71. [71]M. Izadimehr, M. Ghanbari, G. Chen, W. Zhou, X. Hao, M. Dasari, C. Timmerer, H. Amirpour, SVD: Spatial Video Dataset, MM 2025 (2025) 12988–12994. doi:10.1145/3746027.3758246.DOI
  72. [72]A. Gholami, W. Zhou, S. Baldoni, C. Timmerer, F. Battisti, H. Amirpour, Perceptual Quality Assessment of Spatial Videos on Apple Vision Pro, IXR 2025 (2025) 20–28. doi:10.1145/3746269.3760422.DOI
  73. [73]X. Zhu, H. Duan, L. Yang, Y. Zhu, X. Min, G. Zhai, P. Callet, ES-VQA: Perceptual Quality Assessment of Egocentric Spatial Videos, Proceedings - IEEE ICME 2025 (2025). doi:10.1109/ICME59968.2025.11209505.DOI
  74. [74]T. Ofner-Graf, V. Sarkleti, P. Svazek, A. Tockner, S. Witzmann, L. Moik, R. Kraßnitzer, C. Gollob, T. Ritter, M. Kühmaier, K. Stampfer, A. Nothdurft, Potential of Apple Vision Pro for Accurate Tree Diameter Measurements in Forests, Remote Sensing 17 (1) (2025) Art. 141. doi:10.3390/rs17010141.
  75. [75]X. Bao, P. Guerrero, A. Abad, Design and Development of Haptic Gloves and Temperature Feedback Module for Apple Vision Pro, Proceedings of IEEE GEM 2025 (2025). doi:10.1109/GEM66882.2025.11155731.
  76. [76]K. Kolvoka, K. Dowd, M. Grubb, Capabilities and limitations of the Apple Vision Pro as a portable psychophysics tool: Replicating crowding and polar angle asymmetries, Journal of Vision 26 (7) (2026) 13. doi:10.1167/jov.26.7.13.
  77. [77]A. Paul, F. Eishita, Eyes or Hands?: Evaluating Gaze-Pinch and Touch for Game Interaction on Apple Vision Pro, Proceedings - 2026 IEEE 5th International Conference on Intelligent Reality (2026) ICIR 2026 doi:10.1109/ICIR70631.2026.11628254.DOI
  78. [78]A. Karelin, D. Brazhenko, G. Kliukovkin, Y. Chernenko, Real-Time Hand Tracking and Collision Detection for Immersive Mixed-Reality Boxing Training on Apple Vision Pro, Sensors 25 (16) (2025) Art. 4943. doi:10.3390/s25164943.DOI
  79. [79]M. Ghanbari, W. Zhou, C. Stejerean, C. Timmerer, H. Amirpour, SDART: Spatial Dart AR Simulation with Hand-Tracked Input, MM 2025 (2025) 13543–13545. doi:10.1145/3746027.3754484.DOI
  80. [80]M. Ghanbari, H. Amirpour, C. Timmerer, M. Izadimehr, W. Zhou, C. Stejerean, STACK: Spatial Tower Assembly using Controlled Kinetics, 2025 International Conference on Visual Communications and Image Processing (2025) VCIP 2025 doi:10.1109/VCIP67698.2025.11396840.DOI
  81. [81]S. Xiao, Expansion of immersive film narrative structure by Mixed Reality technology - The case of Vision Pro, Proceedings of VAICT 2026 (2026) 148–155. doi:10.1109/VAICT69205.2026.11519117.DOI
  82. [82]Y. Kobuku, T. Ishikawa, Evaluation of Video Augmentation Effects and Immersive Experience in MR Environments, Lecture Notes in Computer Science 16397 LNCS (2026) 480–491 doi:10.1007/978-3-032-14495-9_37.
  83. [83]Y. Ding, X. Lyu, W. Ma, S. Yang, Z. Gao, Dreaming of Butterflies, Proceedings - 2026 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2026 (2026) 1467–1468. doi:10.1109/VRW70859.2026.00408.DOI
  84. [84]H. Choi, S. Cha, Processed Experience: A Contemporary Discourse on Existence with Spatial Installation, ARTECH 2025 (2026) Art. 33. doi:10.1145/3773699.3776573.DOI
  85. [85]M. Daka, J. Zhong, D. Antiya, Revolutionizing Multiplayer Gaming: A Deep Dive into VisionXO, a 3D Multiplayer Tic-Tac-Toe Game, Communications in Computer and Information Science 2260 (2025) 242–246. doi:10.1007/978-3-031-85923-6_20.
  86. [86]G. Lee, iXeRemin: Designing a Polyphonic MR Instrument for Artistic Performance and Interaction, Proceedings of the International Conference on New Interfaces for Musical Expression (2026).DOI
  87. [87]P.-E. Chontas, S.-C. Buraga, A. Iftene, Adapting to Natural Interaction: Exploring Gesture Control Innovations in XR with Apple Vision Pro, Procedia Computer Science 270 (2025) 2997–3006. doi:10.1016/j.procs.2025.09.424.DOI
  88. [88]C. Wienrich, D. Obremski, Immersion, Attention, and Collaboration in Spatial Computing: A Study on Work Performance with Apple Vision Pro, IEEE Transactions on Visualization and Computer Graphics 31 (5) (2025) 2995–3002. doi:10.1109/TVCG.2025.354915.DOI
  89. [89]R. Cheng, N. Wu, M. Zawello, E. Chai, S. Chen, B. Han, A First Look at Immersive Telepresence on Apple Vision Pro, Proceedings of the ACM SIGCOMM Internet Measurement Conference, IMC (2024) 555–562. doi:10.1145/3646547.3689006.
  90. [90]M. Brhmer, G. Gloystein, B. Zhou, A. Gray, S. Pillai, B. Medina, V. Setlur, Evaluating an Immersive Analytics Application at an Enterprise Business Intelligence Customer Conference, Proceedings - 2025 IEEE Conference on Human Factors in Immersive Analytics, HFIA 2025 (2025) 11–14. doi:10.1109/HFIA68651.2025.00007.
  91. [91]J. Sorysz, N. Fannoun, J. Karolits, B. Zhou, S. Suh, P. Lukowicz, User Experience in Task Guidance: Comparing Mixed Reality Devices with Traditional Vision-Based Systems, Proceedings - 2025 IEEE ISMAR-Adjunct 2025 (2025) 727–728. doi:10.1109/ISMAR-Adjunct68609.2025.000160.DOI
  92. [92]G. Lipkowitz, Palimpsest: a spatial user interface toolkit for cohering tracked physical entities and interactive 3D content, UIST Adjunct 2024 (2024) Art. 42. doi:10.1145/3672539.3686780.DOI
  93. [93]R. Arnold, R. Kempf, R. Waltenspül, H. Schuldt, MediaMix: Multimedia Retrieval in Mixed Reality, Lecture Notes in Computer Science 15524 LNCS (2025) 310–317. doi:10.1007/978-981-96-2074-6_37.DOI
  94. [94]V. Setlur, S. Ridet, RécitKit: A Spatial Toolkit for Designing and Evaluating Human-Centered Immersive Data Narratives, Proceedings - 2025 IEEE HFIA 2025 (2025) 6–10. doi:10.1109/HFIA68651.2025.00006.DOI
  95. [95]Y. Chang, AI3D Desktop: Creating 3D using AI in Skeuomorphic XR and the Future of Work, Proceedings - SIGGRAPH 2024 Appy Hour (2024) Art. 1. doi:10.1145/3664294.3664363.DOI
  96. [96]Y. Chang, VolumeMatic and the AI3D.foundation: Text to interactive volume app spatial computing creator app with generated content, Proceedings - SIGGRAPH 2024 Appy Hour (2024) Art. 10. doi:10.1145/3664294.3664361.DOI
  97. [97]J. Fu, M. Grierson, R. Jiang, S. Fu, M. He, M. Xu, Interface design and interaction optimization for spatial computing 3D content creation and immersive environment generation using Apple Vision Pro, Frontiers in Computer Science 7 (2025) Art. 1591289. doi:10.3389/fcomp.2025.1591289.DOI
  98. [98]T. Lin, J. Yuan, K. Miao, T. Katolikyian, I. Walker, M. Cavallo, Towards a Better Evaluation of 3D CVML Algorithms: Immersive Debugging of a Localization Model, Computer Graphics Forum 44 (3) (2025) Art. e70111. doi:10.1111/cgf.70111.DOI
  99. [99]Z. Pan, RUIAR: Spatial Measurement in Indoor Scenes via Intelligent Labeling and Immersive Visualization, Proceedings of ICHEC 2025 (2026) Art. 71. doi:10.1145/3786995.3787070.DOI
  100. [100]Y. Wu, Y. Li, Q. Wang, J. Ross, G. Johnson, M. Lui, Q. Jin, SpatialPrompt: XR-Based Spatial Intent Expression as Executable Constraints for AI Generative 3D Design, DIS 2026 Companion Proceedings (2026) 509–512. doi:10.1145/3809474.3809475.
  101. [101]V. Weser, A. Crocker, K. Hieftje, Mixed Reality Object Viewing: What do People Remember About the Qualities of Real and Virtual Objects Viewed through Passthrough?, Proceedings - 2026 IEEE VRW (2026) 1017–1019. doi:10.1109/VRW70859.2026.00185.DOI
  102. [102]S. Pan, R. Lin, J. Luo, Virtual Reality Headset Devices: User Experience and Device Optimization for the Apple Vision Pro, Proceedings of SPIE - The International Society for Optical Engineering 13658 (2025) 13658OG. doi:10.1117/12.3071744.
  103. [103]S. Awano, W. She, Y. Shibuya, Multi-Window Operation Utilizing Omnidirectional Space in Mixed Reality, Proceedings – 2026 IEEE 5th International Conference on Intelligent Reality (2026) ICIR 2026doi:10.1109/ICIR70631.2026.1628139.DOI
  104. [104]S. Hart, L. Staveland, Development of NASA-TLX (Task Load Index): results of empirical and theoretical research, in: P. Hancock, N. Meshkati (Eds.), Human Mental Workload, Vol. 52 of Advances in Psychology, North-Holland, 1988, pp. 139–183. doi:10.1016/S0166-4115(08)62386-9.DOI
  105. [105]J. Tiffin, E. Asher, The Purdue pegboard: norms and studies of reliability and validity, Journal of Applied Psychology 32 (3) (1948) 234–247. doi:10.1037/h0061266.DOI
  106. [106]H. Javaheri, V. Rey, D. Habusch, J. Karolus, P. Lukowicz, When AR Hinders Performance: The Hidden Costs of Video-See-Through Displays, Proceedings of the ACM Symposium on Virtual Reality Software and Technology (2025) VRSTdoi:10.1145/3756884.3765981.DOI
  107. [107]C. Hrycak, D. Lewakis, J. Kruger, Investigating the Apple Vision Pro Spatial Computing Platform for GPU-Based Volume Visualization, Proceedings – 2024 IEEE Visualization Conference – Short Papers, VIS 2024 (2024) 181–185. doi:10.1109/VIS55277.2024.00044.DOI
  108. [108]G. Chen, S. Wang, J. Chakareski, D. Soutnikolas, M. Dasari, Spatial Video Streaming on Apple Vision Pro XR Headset, HOT-MOBILE 2025 – Proceedings of the 2025 26th International Workshop on Mobile Computing Systems and Applications (2025) 115–120. doi:10.1145/3760488.3811788.DOI
  109. [109]J. Egger, C. Gasnaxer, G. Luijten, J. Chen, X. Chen, J. Bian, J. Kleesiek, B. Puladi, Is the Apple Vision Pro the Ultimate Display? A First Perspective and Survey on Entering the Wonderland of Precision Medicine, JMIR Serious Games 12 (2024). doi:10.2196/52785.
  110. [110]V. Selvaratnam, R. Snelgrove, M. Naraine, Sport Fans’ Adoption of a Mixed Reality Headset: Extending the Technology Acceptance Model with Team Identification, Sport Marketing Quarterly 35 (2) (2026) 99–117. doi:10.1177/10616934264141857.DOI
  111. [111]F. Davis, Perceived usefulness, perceived ease of use, and user acceptance of information technology, MIS Quarterly 13 (3) (1989) 319–340. doi:10.2307/249008.DOI
  112. [112]Apple Inc., About the security content of visionOS 1.3, Apple Support advisory, published 5 September 2024; visionOS 1.3 released 29 July 2024; source for CVE-2024-40865 (2024). URL https://support.apple.com/en-us/120915
  113. [113]W. Taffesse, The User Affordances of Apple Vision Pro: Insights from Early Adopters, SAGE Open 16 (3) (2026) 21582440261464433. doi:10.1177/21582440261464433.
  114. [114]M. Vergari, T. Kojic, W. Wardah, M. Warsinske, S. Möller, J.-N. Voigt-Antons, R. Spang, Digital Eyes: Social Implications of XR EyeSight, Proceedings of the ACM Symposium on Virtual Reality Software and Technology (VRST) (2024) Art. 53. doi:10.1145/3641825.3689526.DOI
  115. [115]A. Erfina, D. Sukmawan, Z. Ayu, T. Indah, L. Pratiwi, F. Wicaksana, Exploring Public Sentiment Toward the Application of Mixed Reality in Apple Vision Pro Using Convolutional Neural Networks, AIP Conference Proceedings 3432 (1) (2026) Art. 030018. doi:10.1063/5.0328464.
  116. [116]T. Barry, The Development of the Hierarchy of Effects: An Historical Perspective, Current Issues and Research in Advertising 10 (1-2) (1987) 251–295. doi:10.1080/01633392.1987.10504921.DOI
  117. [117]MacRumors, Apple Vision Pro Just Got Even More Expensive, MacRumors news, 25 June 2026; price raised to $3,699 (256GB) / $3,899 (512GB) / $4,199 (1TB), attributed to rising memory and storage component costs (2026). URL https://www.macrumors.com/2026/06/25/apple-vision-pro-just-got-even-more-expensive/

Paper details

vlnc:W_v5em6d9k
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents