Immersive Virtual Reality has been increasingly integrated into educational contexts due to its potential for spatial engagement. However, the trade-offs between immersion and interaction performance remain a subject of debate. This paper evaluates this relationship using GeoConnect, an educational cartography application designed to support Brazilian cartography learning through interactive map-assembly tasks. We conducted a within-subjects study with 36 students aged 12–13 years to compare two functionally equivalent 3D versions of GeoConnect, evaluated as a conventional Desktop condition and an HMD-based condition. The evaluation adopts a mixed subjective–objective approach, combining standardized instruments, SUS, UEQ, and SSQ, with telemetry-based metrics, including completion time, errors, precision, and a hesitation index. Results indicate that both conditions achieved high perceived usability, with mean SUS scores of 78.19 for the Desktop condition and 79.72 for the HMD-based condition. The HMD-based condition was preferred by 88.89% of participants, suggesting a strong preference for the immersive condition. However, cybersickness was not negligible: the HMD-based condition produced a mean SSQ total score of 32.41, a median of 22.44, and the highest subscale score in Disorientation; the most frequent symptoms were fullness of head, eyestrain, and dizziness with eyes open. Objective telemetry suggested that immersive interaction increased completion time by 82.97 seconds on average, although this difference did not survive Holm–Bonferroni correction across telemetry metrics. Presentation order also influenced completion time: the HMD time cost was concentrated among participants who used the HMD-based condition first, whereas participants who used Desktop first completed the later HMD-based condition with nearly equivalent times. These findings indicate descriptive pragmatic–hedonic tendencies moderated by learning and carryover effects. We conclude that Desktop and HMD-based conditions should be treated as complementary tools in geography education rather than mutually exclusive alternatives.
Virtual Reality (VR) has become an increasingly relevant medium for psychological interventions because it can combine immersion, controlled stimuli, and therapist-guided reflection. However, VR systems explicitly grounded in Positive Psychology remain comparatively underexplored, particularly in supervised psychotherapy settings. This paper presents Positive Psychology VR, a Unity-based VR intervention composed of eight virtual environments designed around Positive Psychology constructs, including gratitude, mindfulness, savoring, personal strengths, positive reappraisal, emotional expression, and acts of kindness. We report an exploratory pilot study with 30 adults who participated in three weekly therapist-guided sessions using a Meta Quest 3 head-mounted display, followed by a three-month follow-up. Data were collected through participant characterization, open-ended session reports, psychologist observations, acceptability records, and follow-up feedback. Results indicate high acceptability, with all participants reporting that they would recommend the experience. No serious cybersickness events were reported. Qualitative findings suggest that the intervention supported self-awareness, reflection, emotional regulation, relaxation, and changes in perspective. Nature-inspired, meditation-oriented, and strengths-based environments were among the most frequently used and were often described as memorable anchors for later recall. At follow-up, most respondents reported perceived benefits that extended beyond the sessions, although individual variability was observed. These findings suggest that Positive Psychology-oriented VR can be feasibly integrated into professionally supervised psychotherapy as an adjunctive tool for reflection and emotional well-being. The study contributes design and evaluation insights for immersive therapeutic systems and highlights the need for larger, controlled, and longitudinal studies.
XR content creation increasingly combines reconstruction, generative modeling, animation synthesis, neural rendering, and real-time engine deployment. However, creating deployable synthetic 4D humans remains fragmented because human assets must preserve body structure, appearance, motion, deformation, and scene-level plausibility across time. This paper proposes SPL4SH, a Systematic Pipeline for 4D Synthetic Humans, to investigate how contemporary AI-assisted techniques can be organized into an XR-oriented production workflow. SPL4SH is built from a stage-based taxonomy covering modeling, rigging, animation, rendering, and interaction. This taxonomy supports a comparative suitability assessment of recent techniques and guides a modular proof-of-concept implementation integrating SMPLify for parametric body modeling and rigging, Kimodo for controllable motion generation, SMPLitex for SMPL-compatible texture generation, Rokoko-based retargeting, Blender-based asset integration, and Unreal Engine deployment. The evaluation combines individual module tests, full-pipeline integration, and scene-level deployment in a meeting-room environment involving human-object, human-human, and human-scene arrangements. Results show that modular 4D human generation is technically feasible: SMPL-X geometry supports rigged deformation, generated motions can be retargeted to animated characters, texture maps improve mannequin-like bodies, and final assets can be placed in real-time XR scenes. However, persistent barriers remain, including format mismatch, manual skeleton alignment, retargeting fragility, texture discontinuities, hallucinated visual regions, motion interpenetration, stiff transitions, and scene-level validation. SPL4SH contributes a grounded framework for selecting, combining, and evaluating independent generative components as XR-ready synthetic human assets.
This paper presents FacePipe, a low-cost 3D facial capture and animation solution based on a conventional RGB camera and integrated with Blender. The system was developed from design requirements derived from an exploratory review, workflow analysis, and technology-selection criteria focused on low-cost hardware availability, interoperability, usability, openness, and maintainability. FacePipe consists of a MediaPipe-based capture module, a structured local data management and export workflow, and a dedicated Blender add-on that converts captured blendshape coefficients into editable animation data. The solution was evaluated through a Hierarchical Task Analysis comparing FacePipe, FaceCap, and a MetaHuman-based workflow, and through a between-subjects usability study comparing FacePipe and FaceCap with an adapted UEQ+ questionnaire. The evaluation addresses workflow structure and perceived user experience, but it does not establish objective tracking-accuracy equivalence between FacePipe and FaceCap. Results indicate that both tools were evaluated positively, but with distinct workflow profiles. FaceCap was rated more favorably in setup, perceived dependability, and Blender integration, while FacePipe showed a descriptive advantage in file export and organization and comparable perceived usability during recording. These findings suggest that FacePipe is a viable low-cost alternative for editable facial animation workflows, especially in contexts that value widely available RGB-camera-based capture, structured local file management, open implementation, and integration with Blender.
Content-based image retrieval (CBIR) has advanced significantly with deep learning, yet effectively ranking similar images remains challenging, particularly in high-dimensional feature spaces, where pairwise distances often fail to capture contextual relationships and the semantic gap between visual features and high-level concepts persists. Manifold learning and rank-based refinement methods have emerged as complementary strategies, respectively improving feature representations and exploiting contextual information embedded in ranked lists, such as neighborhood relationships among images. However, combining these projection-based and rank-based strategies to exploit their complementary properties remains a challenging research problem. To address this, we propose a framework that combines neighbor embedding projections with rank-based manifold learning through rank aggregation. Uniform Manifold Approximation and Projection (UMAP) generates alternative low-dimensional feature representations, and ranked lists obtained from UMAP projections and rank-based re-ranking methods are combined using the Borda Count aggregation strategy. Experiments were conducted on several public datasets using deep learning features extracted from ResNet152, Swin Transformer, and DINOv2 models. Results show that the proposed approach improves retrieval effectiveness in several scenarios, particularly when the baseline representation struggles to achieve high precision. The aggregation strategy also often improves the quality of top-ranked positions, leading to competitive Mean Average Precision (MAP) and Precision values across different datasets and feature extractors. These findings suggest that combining projection-based and rank-based manifold learning strategies through rank aggregation can provide complementary contextual information for image retrieval tasks.
