PaperlayerView exact revision on arXiv

Source-linked AI summary

Resize Me! Exploring the User Experience of Embodied Realistic Modulatable Avatars for Body Image Intervention in Virtual Reality

Nina Döllinger, Erik Wolf, David Mal, Stephan Wenninger, Mario Botsch, Marc Erich Latoschik, Carolin Wienrich

arXiv:2203.05060v1cs.HC

TL;DR

Body image disturbances associated with obesity motivate complementary VR interventions that can alter how users experience and perceive their bodies. The paper presents and evaluates a personalized, photorealistic avatar system with real-time body-weight modulation in a formative study of 12 participants. The evaluation found solid overall UX and informed design guidelines for future body image intervention systems.

  • Problem

    Existing VR body image research has rarely combined interactive body-weight modification with avatar embodiment and personalization while also examining body awareness.

  • Method

    The authors developed a personalized photorealistic avatar system and evaluated its body scan, VR exposure, interaction methods, body awareness, and body-weight perception with 12 participants.

  • Results

    The prototype demonstrated solid overall UX, particularly for body scanning and the VR experience, while participant feedback identified constructive areas for improvement.

  • Takeaways & Limitations

    The derived design guidelines support future development and evaluation of VR systems for body image interventions.

  • Takeaways & Limitations

    Clinical applicability remains untested because the design phase used students without diagnosed body image disturbance, predominantly with healthy-range BMI.

Abstract

from arXiv · show

Obesity is a serious disease that can affect both physical and psychological well-being. Due to weight stigmatization, many affected individuals suffer from body image disturbances whereby they perceive their body in a distorted way, evaluate it negatively, or neglect it. Beyond established interventions such as mirror exposure, recent advancements aim to complement body image treatments by the embodiment of visually altered virtual bodies in virtual reality (VR). We present a high-fidelity prototype of an advanced VR system that allows users to embody a rapidly generated personalized, photorealistic avatar and to realistically modulate its body weight in real-time within a carefully designed virtual environment. In a formative multi-method approach, a total of 12 participants rated the general user experience (UX) of our system during body scan and VR experience using semi-structured qualitative interviews and multiple quantitative UX measures. By using body weight modification tasks, we further compared three different interaction methods for real-time body weight modification and measured our system's impact on the body image relevant measures body awareness and body weight perception. From the feedback received, demonstrating an already solid UX of our overall system and providing constructive input for further improvement, we derived a set of design guidelines to guide future development and evaluation processes of systems supporting body image interventions.

1 INTRODUCTION

The paper motivates VR-based body image interventions as complements to established treatments and introduces a personalized, photorealistic avatar system with real-time body-weight modulation. A formative evaluation examined usability, user experience, interaction methods, body awareness, and body-weight perception.

  • Motivation: Obesity carries physical and psychological burdens, including body image disturbance associated with stigmatization.The paper describes obesity as a chronic disease associated with secondary diseases, mortality risks, and disturbed body image.
  • Motivation: VR avatar interventions offer a way to complement established body image treatments through embodied experiences.Prior VR methods use human 3D models and may influence perception and behavior, especially through avatar embodiment.
  • System and aims: The proposed system creates personalized photorealistic avatars that users can embody and modify in body weight in real time.The system was designed for accessibility, ease of use, and later evaluation in a clinically relevant setting.
  • Evaluation: The study used qualitative interviews and quantitative UX measures to derive design guidelines for future body image intervention systems.The formative evaluation targeted both overall system experience and initial effects on body awareness and body-weight perception.
  • Evaluation: Twelve participants completed avatar generation, virtual exposure, body-weight estimation, and interactive modification tasks using gestures, a joystick, or virtual buttons.The evaluation also collected interviews and quantitative ratings of calibration, interaction methods, presence, embodiment, body awareness, and avatar perception.

2 RELATED WORK

Related work establishes VR avatars as tools for studying body-weight perception and supporting body image interventions. The paper addresses limited integration of dynamic modification, embodiment, personalization, interaction, and body-awareness assessment.

  • Body image disturbance: Body image disturbance involves negative, distorted, or inaccurate body perception and may include body-weight misperception, dissatisfaction, and reduced body awareness.Established treatments include psychoeducation, self-monitoring, mirror exposure, and video feedback.
  • Avatar-based VR: VR avatars can simulate rapid body-shape or body-weight changes, supporting investigation of body-weight perception and potential intervention approaches.Prior systems include both generic avatars with different weights and dynamic body-weight modification methods.
  • Avatar-based VR: The proposed system dynamically modifies an embodied avatar while preserving a small facial region to better retain user identity.Previous systems commonly presented static modified avatars or used models that either covered the whole body or omitted the head region.
  • Personalization and embodiment: Personalization and embodiment influence avatar body-weight estimation, with personalization effects depending on matching avatar shape and texture.Embodiment includes self-location, body ownership, and agency, and visuomotor coherence can evoke it.
  • Research gap: The paper extends mirror-confrontation approaches by letting users interactively adjust avatar shape to engage with their body image.The authors identify limited prior work combining simulated body-weight modification with embodiment or user interaction and assessing body awareness as a possible mediator.

3 SYSTEM DESCRIPTION

The VR system uses a Unity-based implementation with a Valve Index headset, handheld controllers, and body-mounted trackers for motion capture. The hardware provides visual display and tracking for avatar animation.

  • Technical implementation: The system was implemented in Unity 2019.4.15f1 LTS and integrated with SteamVR 1.16.10 and its Unity plugin.The implementation used a Valve Index headset and associated tracking hardware.
  • Technical implementation: The Valve Index headset provides 1440×1600 pixels per eye, a 120° field of view, and a 90 Hz refresh rate.These display specifications describe the visual hardware used during the VR experience.
  • Motion tracking: Motion tracking uses two handheld controllers, one lower-spine tracker, and two foot trackers.The trackers support capture of body movements for avatar animation.

3.1 Virtual Environments

The system uses two virtual environments: a spatially calibrated replica of the physical experiment room for preparation and a realistic exposition office for mirror exposure. Figures 1 and 2 visualize these environments and avatar weight states.

  • Virtual preparation environment: The preparation environment replicates the physical room and is spatially overlaid through automatic calibration.Preparatory tasks include ground calibration, vision testing, equipment adjustments, and embodiment calibration.
  • Virtual preparation environment: Figure 1 compares the real experiment environment with its replicated virtual preparation environment during embodiment calibration.The left and right sides show the physical user and virtual avatar, respectively.
  • Virtual exposition environment: The exposition environment is modeled as a psychotherapist’s office with an exposure area and virtual mirror.The mirror enables allocentric observation of the embodied avatar within a realistic, coherent setting.
  • Virtual exposition environment: Figure 2 compares the personalized avatar at reduced, normal, and increased body weight in front of the virtual mirror.The three states are arranged from left to right as reduced, normal, and increased body weight.

3.2 Generation and Animation of Personalized Avatars

The system generates personalized photorealistic avatars from photogrammetry and animates them in real time from tracked participant movements.

  • Participants embody a personalized avatar from an egocentric perspective while it follows their body movements in real time.
  • A 94-camera photogrammetry rig captures detailed subject images for automatic point-cloud generation.Ten cameras are zoomed in on the face, and uniform illumination is provided by four studio lights with diffuser balls.
  • Template fitting uses 23 manually selected point-cloud landmarks and a template containing approximately 21,000 vertices, an animation skeleton, blendshapes, and eye and teeth meshes.
  • SteamVR motion tracking continuously captures movements and calibrates body height, arm length, limb orientations, and associated body parts for animation.

3.3 Body Weight Modification of Avatars

The system models body-weight changes from anthropometric measurements and PCA, updates nonfacial vertices, and stitches them to a fixed face region while comparing three interaction methods.

  • The runtime system dynamically modifies avatar body weight using a statistical model of weight gain and loss.
  • The body-shape model uses 1,700 European CAESAR scans and maps anthropometric measurements to PCA coefficients.
  • Thirty PCA components represent body-shape variation, while a selector matrix extracts coordinates for vertices outside the fixed face region.
  • Desired anthropometric changes are projected into PCA space and then vertex space before differential-coordinate reconstruction stitches modified vertices to the unmodified face.
  • Fixing the face preserves user identity at large weight changes and avoids recalculating auxiliary meshes such as eyes and teeth.
  • The evaluation compares gestures, joystick movement, and virtual objects, with all methods restricted to ±35% of the user’s body weight.
  • The avatar example starts at BMI 19.8 and is modified across BMI 16–32 in two-point increments.

4 EVALUATION

A formative evaluation examined the system’s scan and VR user experience, interaction methods, and body-image-related measures using qualitative interviews and quantitative assessments.

  • The prototype evaluation combined qualitative questions and quantitative measures covering body scan, VR exposure, interaction methods, and body image.
  • Twelve healthy student participants completed the evaluation, which used a conservative formative strategy before later clinical feasibility testing.
  • A counterbalanced 1x3 within-subjects design compared gestures, joystick, and objects through active modification tasks, with passive estimation tasks before and after.
  • Semi-structured interviews assessed expectations, comfort, process clarity, avatar feelings, preferred modification method, estimation difficulty, body awareness, and body affect.
  • Quantitative measures included presence, embodiment, simulator sickness, avatar affect, workload, calibration and interaction efficiency, body awareness, and weight estimation.
  • Method rankings combined workload, estimation difficulty, vividness, contentment, and preference using reversed weighted scores.
  • Passive estimation used nine avatar modifications across ±20% in 5% intervals, while active tasks required matching numeric target weights.

5 RESULTS

The evaluation found generally positive scan and VR experiences, while revealing mixed avatar perceptions and task-specific usability differences. Body weight estimation was difficult, and modification methods did not significantly differ in estimation accuracy.

  • 5.1 Body Scan Experience: Most participants found the body scan simple and clear, and all said they would undergo it again.Four participants described it as straightforward or easy, while others found it interesting, felt observed, or needed time to adjust.
  • 5.1 Body Scan Experience: Most participants perceived body-measure assessment as neutral or similar to a doctor’s appointment, though one found weight measurement private and uncomfortable.Eight participants reported the assessment as neutral or medically familiar; three did not expect it in a normal laboratory study but did not find it awkward.
  • 5.2 VR Experience: Simulator sickness scores increased by 16.21 points, remaining below the 20-point indication threshold, with no significant pre–post difference.The Wilcoxon test yielded Z = 1.14, p = .254.
  • 5.2 VR Experience: Avatar reactions were mixed: six participants described the experience positively or neutrally, whereas six described it as strange or irritating.Criticism focused particularly on absent facial expressions, eye movements, and hand gestures.
  • 5.2 VR Experience: Joystick interaction was preferred by 8 participants for controllability and simplicity, while 4 preferred gestures for intuitiveness, flexibility, and directness; none preferred objects.Body-awareness ratings also tended to be higher with joystick interaction than with gestures or objects.
  • 5.3 Body Image-Related Measures: The three modification methods did not differ significantly in average or absolute body-weight misestimation, while estimating avatar weight was generally difficult.Eight participants described estimation as difficult, citing repetition and reduced physical-body perspective; direction effects varied by estimation method.

6 DISCUSSION

The system showed a generally solid user experience across scanning, calibration, VR exposure, avatar embodiment, and body-weight modification, while revealing privacy, realism, personalization, and interaction trade-offs for future clinical use.

  • 6.1 Body Scan Experience: Participants found body scanning simple, interesting, acceptable, and compatible with clinical settings, despite substantial technical effort.They reported willingness to be scanned again and did not describe the associated measurements as unpleasant.
  • 6.1 Body Scan Experience: Feelings of being watched and left alone constrained the scanning experience, motivating privacy-preserving camera arrangements and more continuous personal contact.The authors suggest reducing or rearranging cameras and maintaining dialogue during scanning.
  • 6.2 User Experience of VR Experience: Calibration received positive feedback, with low calibration times and workload, although markerless tracking could further reduce effort and invasiveness.The current procedure required only a short period holding a T-pose.
  • 6.2 User Experience of VR Experience: Presence was acceptable, but involvement and realism were lower, potentially because experimenter interaction and implausible body-weight modification disrupted the VR experience.The paper calls for further research on presence and its sub-dimensions in VR body-image interventions.
  • 6.2 User Experience of VR Experience: Virtual body ownership was lower than in prior work with personalized photorealistic avatars, possibly because minor visual inaccuracies and real-time shape changes reduced body concordance.Participants sometimes described their avatars as uncanny or not fully recognizable, raising questions about the necessity and feasibility of highly photorealistic textures.
  • 6.2.1 User Experience of Body Weight Modification: Joystick and gesture interactions were similarly rated, with slight preference for the joystick, whereas virtual objects were more demanding, difficult, less vivid, and less preferred.The authors also note that richer modifications by body part or tissue composition could improve expressiveness but would increase interaction difficulty.

Guidelines for Body Weight Modifications

The study derives practical guidelines for realistic body-weight modification and estimation in embodied VR, while identifying individual and methodological factors requiring further research.

  • Body Weight Modifications: Body-weight modifications should avoid unrealistic ranges or severe deviations from users’ BMI to reduce alienation.
  • Body Weight Modifications: Body-weight changes should be independently controllable across body parts to reflect differing tissue compositions.
  • Body Weight Modifications: Hardware input devices or body gestures should be preferred over virtual objects or buttons for weight modification.
  • Body Image-Related Outcomes: Reported effects on body awareness and affect were highly individual, with participants describing either increases or decreases in both.The authors call for larger samples and further study of individual characteristics and task difficulty.
  • Body Weight Estimations: Body-weight estimates did not differ across interaction methods, while passive estimation was more accurate than active modification.The authors relate this difference to the greater familiarity of estimating appearance-based weight than actively modifying a virtual body, while noting that task phrasing may also matter.
  • Body Weight Estimations: Body-weight estimations should occur early because perceptibility of the real body may decrease during an intervention.
  • Body Weight Estimations: Estimations should present the body equally from multiple perspectives and use healthy individuals’ average accuracy to limit system-property effects.
  • Future Research: Future research should clarify differences between active modification and passive estimation, including their possible effects on body image.

7 CONCLUSION

The evaluated prototype provided an enjoyable VR experience and generated design guidelines for future body-image intervention systems. Therapeutic application remains premature, requiring further work on avatar design, task differences, and target-group evaluation.

  • Conclusion: The prototype supported embodiment of a rapidly generated personalized photorealistic avatar with real-time body-weight modulation.The system showed positive qualities particularly during body scans and the overall VR experience.
  • Conclusion: The derived design guidelines can facilitate future development of VR body-image therapy support systems.
  • Conclusion: More research is needed before therapeutic application, including work on photorealism, active versus passive weight tasks, and the intended target group.Future studies should also examine individual characteristics related to body-image distortion, body dissatisfaction, and body awareness.

1 EVALUATION INTERVIEW QUESTIONS

The evaluation interviews used semi-structured questions covering the body scan, measurements, avatar interaction, physical experience, task instructions, and the overall process.

  • Body Scan Experience: Body-scan questions addressed expectations, clarity of instructions, feelings during scanning, willingness to repeat it, desired changes, and experimenter-related comfort.
  • Body Measurements: Body-measurement questions examined reasons for the procedure, pleasantness, possible improvements, and whether experimenter gender affected comfort.
  • Avatar Interaction: Avatar-interaction questions asked about feelings toward the personal avatar and the pleasantness of receiving instructions without a visible speaker.
  • Overall Process: The interview also asked whether participants could imagine sharing the virtual environment and what they would change about the overall process.
  • Physical Experience: Physical-experience questions covered appearance changes, awareness of the physical body, altered body perception, direct consequences, and possible support for conscious bodily experience.
  • Task Instructions: Task-instruction questions assessed expectations, verbal and textual instructions, avatar appearance, weight estimation, weight adjustment, and interaction-method preferences.

4. If you imagine a visual representation of the instructing voice, how would it look like?

The interview addressed whether participants could imagine sharing the virtual environment with another person during avatar-appearance changes and invited broader feedback on the process.

  • 4. If you imagine a visual representation of the instructing voice, how would it look like?: Participants were asked whether they could imagine sharing the virtual environment with another person while changing their avatar’s appearance.
  • 4. If you imagine a visual representation of the instructing voice, how would it look like?: The interview concluded by asking what participants would change about the overall process and whether they noticed anything else.
Loading 2203.05060v1…