Next Article in Journal
Characteristics of Growth, Development, Response to Mineral Nutrition and Yield Stability of Spring Barley Varieties on Chernozem Soils
Previous Article in Journal
Leukocyte Differentiation and Scattergram Abnormalities in Rabbits Using the Mindray BC-60R Hematology Analyzer
Previous Article in Special Issue
Technology Configurations and Adherence in Virtual Reality-Based Mindfulness: A Systematic Review of Headsets, Wearables, Software, and Therapeutic Applications
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 16
Issue 18
10.3390/app16188906
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Measurement Differences Between Markerless Motion Capture Systems During Sit-to-Stand Performance: Implications for Clinical Interpretation and Physical Therapy Education

by
Christopher Voltmer
*,
Casey Imperio
and
Mark Apostol
Doctor of Physical Therapy Program, School of Health Sciences, Touro University, 225 Eastview Drive, Central Islip, New York, NY 11722, USA
*
Author to whom correspondence should be addressed.
Appl. Sci. 2026, 16(18), 8906; https://doi.org/10.3390/app16188906
Submission received: 17 July 2026 / Revised: 27 August 2026 / Accepted: 4 September 2026 / Published: 8 September 2026
(This article belongs to the Special Issue Current Advances in Rehabilitation Technology)
Browse Figures
Versions Notes

Abstract

Background: Markerless motion capture systems are increasingly used in clinical and educational settings to assess functional movement. However, differences in system design and data processing may influence measurement outputs, raising questions about the systematic differences in these technologies. Methods: A within-subject, repeated-measures design was used to compare range of motion (ROM) measurements obtained from two markerless motion capture systems during a sit-to-stand task. Fifty healthy adults (18–50 years) performed the task under three surface conditions (firm, compliant, and commode) while both systems recorded movement simultaneously. Joint ROM at the trunk, hips, knees, and ankles was analyzed using linear mixed-effects modeling to examine device, surface, and interaction effects. ROM was standardized to account for differences in absolute ROM magnitude and variability across biomechanically distinct joint measures. Results: A linear mixed-effects model showed significant main effects of both Device and Surface (all p’s < 0.001). There were also significant interactions between Device × Surface (p = 0.032), Device × Joint (p < 0.001), and Surface × Joint (p < 0.001), indicating that differences between devices and surfaces varied across biomechanical measures. The device-related discrepancies in standardized ROM (zROM) were greatest for hip adduction and trunk forward lean, whereas smaller between system differences were observed for ankle dorsiflexion and knee flexion. The firm surface demonstrated relatively larger zROM measurements for ankle dorsiflexion, knee flexion, and hip flexion compared to the compliant and commode surfaces. Conclusions: Markerless motion capture systems produced systematically different measurements during a functional task, and these differences were influenced by movement type and task conditions. The findings suggest that kinematic data derived from different markerless motion capture systems should be interpreted cautiously when comparing movement performance. Consistent use of a single system may be important for clinical tracking, and educational programs should emphasize critical interpretation of technology-derived movement data.

1. Introduction

Accurate assessment of human movement is a foundational component of physical therapy practice and informs clinical decision making, identification of movement impairments, progression of rehabilitation interventions, and evaluation of patient outcomes. Functional tasks such as the sit-to-stand (STS) are frequently used to assess lower extremity strength, balance, coordination, postural control, and mobility across a wide range of clinical and non-clinical populations [1,2,3,4]. The STS task is considered biomechanically demanding because it requires coordinated trunk flexion, lower extremity force generation, dynamic postural control, and anticipatory balance adjustments during the transition from a stable seated posture to upright standing [1,5]. Impairments in STS performance have been associated with functional decline, increased fall risk, mobility limitations, sarcopenia, decreased gait speed and reduced independence in older adults and clinical populations [2,6,7,8,9]. As a result, accurate assessment of STS movement quality and joint kinematics is clinically important in rehabilitation settings [5].
Traditionally, assessment of movement quality in physical therapy has relied heavily on observational analysis and clinician interpretation. Although observational assessment remains a fundamental component of clinical practice, visual analysis may be influenced by subjectivity, variability in clinician experience, and limitations in the ability to accurately quantify complex multi-joint movement patterns [10]. In response, motion capture technologies have increasingly been incorporated into rehabilitation, sports medicine, and educational settings to provide objective, quantifiable measures of movement performance. Historically, three-dimensional marker-based motion capture systems have been considered the gold standard for biomechanical analysis; however, these systems often require laboratory environments, reflective markers, specialized personnel, and extensive processing time, limiting their accessibility in routine clinical settings [11,12].
Markerless motion capture systems have emerged as a more accessible alternative capable of providing real-time movement analysis without the need for reflective markers or extensive laboratory infrastructure. These systems use camera-based tracking algorithms, depth-sensing technologies, and biomechanical modeling approaches to estimate joint kinematics and movement performance during functional activities. Markerless systems have gained increasing attention within rehabilitation and physical therapy education due to their portability, lower cost, reduced setup time, and ability to provide immediate movement feedback in clinical and educational environments [13,14,15,16]. More recently, advances in artificial intelligence and computer vision technologies have accelerated the integration of markerless motion capture into movement assessment, remote monitoring, and rehabilitation applications [17,18]. Previous research has demonstrated acceptable validity and reliability of several markerless motion capture systems for assessment of gait, balance, postural control, and functional movement when compared to traditional laboratory-based systems [11,12,13,16].
Despite these advancements, important questions remain regarding the comparability of measurements generated across different markerless motion capture systems. Variability in hardware configuration, camera placement, tracking algorithms, filtering methods, sampling frequency, and biomechanical modeling approaches may contribute to systematic differences in reported movement outcomes between systems, even when assessing the same task. The prior biomechanical and rehabilitation technology literature has emphasized that differences in processing pipelines and measurement methodologies can influence interpretation of movement outcomes and comparisons across systems [19,20]. These potential differences are clinically important because clinicians, educators, and researchers may assume that outputs generated by different systems are interchangeable or directly comparable. If substantial measurement differences exist between systems, interpretation of movement performance, longitudinal tracking of patient progress, and educational interpretation of biomechanical data may be influenced by the specific technology being used rather than true differences in movement performance.
Most studies evaluating markerless motion capture have focused on establishing criterion validity through comparison with marker-based motion capture systems, which remain the accepted reference standard for biomechanical analysis [16,21,22,23,24]. These investigations are essential for determining the absolute accuracy of individual markerless systems. However, as markerless motion capture technologies become increasingly integrated into clinical practice, rehabilitation, and physical therapy education, clinicians and researchers are increasingly selecting between commercially available markerless platforms rather than between markerless and marker-based systems. Consequently, understanding whether different commercially available markerless systems provide comparable measurements represents a distinct and clinically relevant research question with important implications for rehabilitation, physical therapy education, and movement science.
Therefore, the purpose of this study was to evaluate differences in ROM measurements obtained from two commercially available markerless motion capture systems during STS transfers performed under varying surface conditions. It was hypothesized that systematic differences would exist between systems due to differences in hardware configuration, tracking algorithms, biomechanical modeling, and signal processing approaches. Additionally, it was hypothesized that these differences would vary across both surface and joint variables.

2. Methods

2.1. Study Design

A within-subject, repeated-measures design was used to compare ROM measurements obtained from two markerless motion capture systems during a functional sit-to-stand task. Each participant performed the task under multiple surface conditions while being simultaneously recorded by both systems, allowing for direct comparison of measurement outputs within the same movement trials. Approval to conduct this study was obtained through the University’s Institutional Review Board (IRB #26327).

2.2. Participants

A convenience sample of 50 adults (32 males, 18 females), ages 22–41 (M = 25.6, SD = 3.84) was recruited from a university community (See Table 1 for all demographic information). Inclusion criteria required participants to be able to perform STS transfers independently without assistance. Exclusion criteria included current musculoskeletal injury, use of an assistive device, neurological or cognitive impairment, visual impairment, medications affecting balance or coordination, and pregnancy.
An a priori power analysis was conducted in G*Power (Version 3.1.9.6) using a repeated-measures, within-subject ANOVA framework to approximate the planned study design. The parameters entered were one group, six measures, an effect size of 0.2, a power level of 0.9, and an alpha level of 0.05. The analysis showed that 36 participants were required; however, to account for attrition and incomplete data, the researchers decided on a final sample size of 50 participants.
The final analyses were conducted using linear mixed-effects modeling to account for within-subject dependency across repeated surface, device, and joint-angle measurements. To account for this, an additional simulation-based sensitivity analysis was conducted to evaluate the adequacy of the achieved sample size under the final model structure. Simulation-based power analyses are able to accommodate both the fixed- and random-effects structure of mixed models [25]. The simulation was conducted in RStudio (Version 2026.08.0+187) using the simr package [26] and retained the final structure of the reported LMM, including participant as a random intercept, and Device, Surface, Joint, and all two-way interactions as fixed effects. Device-related fixed effect coefficients were proportionally scaled while preserving the relative pattern across surfaces and joint measurements. Based on 1000 Monte Carlo simulations, the sample of 50 participants provided 89.1% power (95% CI: 87.0–91.0%) to detect a Device-related effect pattern at 80% of the magnitude observed in the fitted model, using a Type II F-test with Satterthwaite degrees of freedom. A random seed of 2026 was specified for reproducibility.

2.3. Instrumentation

Two markerless motion analysis systems, the DARI motion capture and Kinotek, were utilized to capture and analyze participants’ movement during STS transfers across three surfaces. The DARI Motion capture device (DARI Motion Inc., Overland Park, KS, USA) uses a network of eight FLIR BFS-PGE-04S2C-CS cameras (Teledyne FLIR, Richmond, BC, Canada; 720 × 540-pixel resolution; 60 frames per second) and proprietary software (Captury Live version 249i, build 40aea7a74 Loud Hummingbird for ubu18) to perform full-body kinematic tracking. This device captures real-time, detailed movement data via skeletal tracking without the need for physical markers, allowing for precise measurement of joint range of motion (ROM) and overall biomechanical performance [27]. The DARI used in this study was a fixed device and calibrated using standardized DARI calibration protocols.
The Kinotek device is a portable, markerless motion capture system that uses a single Microsoft Azure Kinect v2 depth sensing camera to capture real-time movement and estimates joint kinematics and ROM via automated tracking of anatomical landmarks [28]. The camera has a sampling rate of 30 frames per second and combines RGB video and depth sensing technology with machine learning-based pose estimation, allowing for identification of 3D skeletal landmarks and segment orientations. Utilizing proprietary software, the estimated skeletal landmarks are used to derive the joint kinematics. Both joint-angle calculations and planes of motion are software defined and not modifiable by the user [23]. Kinotek operates as a cloud-based platform with proprietary processing algorithms. A discrete software or algorithm version associated with the analyzed data was not available to the authors. For the current study, both systems simultaneously recorded ROM at the trunk, hips, knees, and ankles during STS transfers.

2.4. Procedure

Before data collection, participants were screened for eligibility via online screening form. Eligible participants were contacted by the researchers, who scheduled their testing session. First, participants were consented and given a basic overview of the study. Next, participants filled out basic demographic information and provided past medical history. Testing was conducted in a controlled indoor environment where the DARI system was permanently affixed. The Kinotek camera was mounted on a tripod and positioned 7 feet from the participant at hip/ASIS height and at a 6° angle. Floor markings were used to standardize the position of the camera, participant, and testing surface throughout data collection. Participants completed the sit-to-stand task across three surface conditions: a firm surface (16 in), a compliant surface (18 in, 16 in compressed), and a commode surface (16 in).
Before completing the task, a researcher demonstrated an appropriate STS transfer and instructions were verbally recited. After ensuring there were no questions, participants performed one STS transfer per surface while both motion capture systems recorded simultaneously. Rest periods were provided as needed.

2.5. Data Analysis

Descriptive and inferential analyses were conducted using JAMOVI statistical software version 2.6 [29]. The data included ROM measurements for each joint angle, surface condition, and device. The included joint angles differed in absolute ROM magnitude and variability, so to facilitate accurate comparisons across biomechanically distinct measurements, ROM values were standardized within each joint using z-score transformations. Z-scores were calculated by subtracting the overall mean ROM for each joint from the observed ROM and dividing by the corresponding standard deviation. Raw ROM descriptive statistics for each joint angle measurement by device and surface condition are presented in Supplementary Table S1.
A linear mixed-effects model (LMM) was conducted to examine differences in ROM measurements across devices and surfaces, while accounting for repeated observations within participants. The dependent variable was standardized ROM (zROM), and fixed effects included device (DARI vs. Kinotek), surface (firm vs. compliant vs. commode), and joint (left/right ankle dorsiflexion, left/right hip adduction, left/right hip flexion, left/right knee flexion, trunk/spine flexion), along with their interaction terms. Participant was included in the model as a random intercept to account for within-subject dependency across the repeated measurements. A priori statistical significance was set at α = 0.05.
There were three primary effects of interest: (1) the main effect of device, which examined if ROM measurements differed between the DARI and Kinotek; (2) the main effect of surface, which examined if ROM measurements differed across seating conditions; and (3) interaction effects assessing if device-related differences in ROM measurements varied as a function of surface condition and joint angle. Pairwise comparisons with appropriate multiplicity correction procedures were conducted for significant effects.

3. Results

An LMM was conducted on zROM values to assess the effects of device (DARI vs. Kinotek), surface (firm vs. compliant vs. commode), and joint while accounting for repeated observations within participants. Fixed effects for device, surface, and joint, as well as all two-way interactions, were included in the model, with participant included as a random intercept. The model converged successfully and demonstrated significantly improved fit relative to an intercept-only model, ΔR2 = 0.181, χ2(37) = 500.137, p < 0.001. The full model explained 17.9% of the variance through fixed effects alone (marginal R2 = 0.179). When including participant-level random effects (conditional R2 = 0.286), 28.6% of the variance was explained via the model. The full model also demonstrated improved fit indices relative to the nested model (AIC = 7041.659 vs. 7467.796; BIC = 7277.700 vs. 7485.499).
The analysis showed significant main effects for both Device, F(1, 2613) = 16.117, p < 0.001, and Surface F(2, 2613) = 78.976, p < 0.001 (See Table 2). There was no significant main effect of Joint; however, this was expected due to the use of within-joint z-score standardization, which removed magnitude differences between joints. This finding should not be interpreted as evidence that all raw ROM values were equivalent across joints, but rather as a product of the within-joint standardization performed on the data pre-analyses. Post hoc comparisons showed that the DARI demonstrated significantly higher zROM measurements compared to the Kinotek (Mdiff = 0.13, p < 0.001). This finding represents the overall standardized device effect. Therefore, this result should not be interpreted as DARI producing greater absolute ROM for all joint-angle measurements. Turning to surface comparisons, STS transfers performed on the firm surface demonstrated significantly higher zROM measurements compared to both the commode (Mdiff = 0.44, p < 0.001) and compliant surface (Mdiff = 0.43, p < 0.001). There were no significant differences in zROM measurements between the commode and compliant surface (p > 0.05).
There were also significant two-way interactions for Device × Surface (F(2, 2613) = 3.455, p = 0.032), Device × Joint (F(8, 2613) = 47.339, p < 0.001), and surface × joint (F(16, 2613) = 7.258, p < 0.001; see Table 3). Post hoc comparisons for the significant Device × Surface interaction demonstrated that the DARI demonstrated relatively higher zROM measurements compared to the Kinotek under the firm surface condition (Mdiff = 0.192, p = 0.010). Smaller differences between the two devices were observed during the compliant and commode conditions, suggesting that device-related differences between motion capture systems may be influenced by seating stability. The significant Device × Joint interaction suggests that system-related differences varied across biomechanical variables (see Figure 1). Post hoc comparisons showed that the largest device-dependent differences were seen in zROM measurements for left hip adduction, right hip adduction, and trunk/spine forward lean (all p’s < 0.05). However, there were comparatively smaller or no significant differences in zROM measurements between systems for ankle dorsiflexion or knee flexion. Finally, the significant Surface × Joint interaction suggests that the effect of surface on zROM varied across biomechanical variables (Figure 2). Post hoc comparisons showed that higher zROM measurements were observed for ankle dorsiflexion, hip flexion, and knee flexion during STS transfers performed on the firm surface compared to both the compliant surface and commode (all p’s < 0.001). Hip adduction and trunk forward lean had smaller surface-related differences in zROM measurements (all p’s > 0.05).

4. Discussion

The primary finding of this study was that the two markerless motion capture systems produced systematically different range of motion (ROM) measurements across multiple joint variables during a sit-to-stand task. These differences were consistent in direction for several movements but varied depending on the plane of motion and specific joint being analyzed. Overall, these findings demonstrate systematic differences in measurements produced by the two systems when assessing movement performance during this functional task. These findings are consistent with the prior literature demonstrating that differences in hardware configuration, biomechanical modeling, filtering techniques, and tracking algorithms may contribute to variability across motion capture technologies [19,20,30,31].
For sagittal plane movements, including ankle dorsiflexion, knee flexion, and hip flexion, the DARI produced relatively higher zROM measurements compared to the Kinotek. In contrast, the Kinotek demonstrated relatively higher zROM measurements for hip adduction and trunk forward lean. These findings suggest that the differences between the two motion capture systems may be movement-dependent, as system-related variability differed across biomechanical measures. Previous investigations comparing markerless and marker-based systems have similarly reported that measurement agreement may vary by movement plane and joint variable, particularly during complex functional tasks requiring multi-joint coordination [11,12,32].
The current study was not designed to explore the underlying cause of these device-dependent differences, and any differences may reflect system-specific measurement and processing approaches. Recent reviews have emphasized that differences in pose estimation algorithms, training datasets, and biomechanical modeling techniques may contribute to variability in markerless motion capture outputs [33]. Differences in the technical configurations of the DARI and Kinotek may provide additional context to these findings, however. The DARI utilizes a fixed eight-camera configuration, which allows movement to be captured from multiple viewing perspectives, as well as a fixed bone-length biomechanical model with distances between estimated joint centers being constrained during movement. In contrast, the Kinotek consists of a single depth-sensing camera and relies on anatomical landmark estimation from a single viewing angle. Thus, differences in both camera configuration and biomechanical modeling could contribute to differences in joint-center estimation and segment orientation between the two systems. Differences in the operational definition of specific measurements may also contribute to device-related differences. DARI defines the trunk measurement using the relationship between the pelvic and thoracic regions. In contrast, although Kinotek reports trunk forward lean as a distinct measurement from lumbar, thoracic, and overall spinal flexion, the specific anatomical landmarks used to calculate this measurement are not available to the authors. Therefore, differences in the operational definition of trunk forward lean may have contributed to the observed differences between systems. This further highlights that similarly labeled kinematic variables across markerless motion capture systems may not represent identical measurements and should therefore be interpreted cautiously when comparing measurements between systems. However, the specific source of observed differences cannot be determined, as the proprietary algorithms used by both manufacturers were not evaluated in this study.
Movement patterns were also influenced by surface condition, with relatively higher zROM values for the sagittal plane movements compared to both the compliant and commode conditions. This pattern was observed across both the DARI and Kinotek devices, suggesting that task constraints may influence movement performance during STS transfers. The prior STS literature has demonstrated that environmental and task-related factors can alter lower extremity kinematics and postural control demands during transfer performance [1,5,34,35]. However, the interaction effects showed that the magnitude of device-related differences in zROM measurements was dependent on biomechanical variables. Device-related differences were more pronounced for hip adduction and trunk forward lean under specific surface conditions. These findings highlight that device-related differences were not uniform and may depend on both the movement being assessed and the environmental context in which the movement is performed [36,37,38].
The present findings should be interpreted within the context of the study design and intended objective. Previous investigations have primarily focused on establishing the criterion validity of individual markerless motion capture systems through comparison with marker-based reference standards [16,21,22,23,24]. In contrast, the present study evaluated systematic differences in ROM measurements between two commercially available markerless motion capture systems rather than the criterion validity or absolute accuracy of either platform. Consequently, the observed differences should not be interpreted as evidence that one system provides more accurate measurements than the other. Instead, these findings suggest that measurements obtained from different commercially available markerless systems may differ systematically, even when evaluating the same functional task. This distinction has important implications for clinicians, educators, and researchers when interpreting movement data across institutions, comparing findings between studies, or monitoring performance longitudinally using different markerless technologies.
From a clinical perspective, systematic differences between markerless motion capture systems should be considered when assessing or tracking patient performance over time. Differences in ROM values may reflect variations in measurement approach rather than true changes in movement, which could influence clinical decision making if not considered. This consideration may be particularly important when motion capture technologies are used longitudinally to monitor patient progress and guide clinical decision making [15,31]. Therefore, consistent use of a single system may be important when monitoring progress or evaluating outcomes.
From an educational perspective, these findings could have implications for how technology-based movement data are discussed in physical therapy education. Motion capture technologies are being increasingly integrated into physical therapy curricula, and thus, awareness that different systems may produce different measurements for the same task may be relevant when teaching future PTs how to interpret biomechanical data [14,16].

5. Limitations

Several limitations should be considered when interpreting the findings of this study. First, the sample consisted of a relatively young and healthy population, which may not reflect movement patterns observed in older adults or clinical populations. Specifically, older adults and individuals with mobility or neurological impairments may have greater movement variability, compensatory movement strategies, and even slower movement speeds when performing STS transfers [5,39]. This variability may influence not only how movement is captured, but also how it is modeled by these two devices, potentially resulting in a different pattern or magnitude of systematic differences than those observed in the current sample. Thus, the present findings should not be generalized to clinical populations or older adults. Second, only a single functional task (sit-to-stand) was evaluated, and findings may not generalize to more complex or dynamic movements. Future research should evaluate these devices across other functional movements in clinical populations with greater movement variability and functional impairment.
Additionally, this study did not include a marker-based motion capture system as a reference standard. Prior work comparing squat ROM measurements between the Kinotek and Vicon marker-based system showed that correspondence in measurements between systems was dependent on the measurement, with larger differences in knee flexion and ankle dorsiflexion ROM between the systems, and similar hip flexion measurements [23]. The DARI has also been compared to marker-based systems such as Vicon [40] and Qualisys [30] across different movements and kinematic variables. Similar to the Kinotek study, measurement differences and magnitude of agreement between the DARI and marker-based systems varied depending on the movements and joint angles. It is important to note that neither DARI studies did not assess a squat or similar movement to an STS. Also, to our knowledge, no study has attempted to determine which device, the DARI or Kinotek, is closer to a gold standard marker-based system. Therefore, the criterion validity and absolute measurement accuracy of either markerless system cannot be determined. However, this was consistent with the intended objective of the investigation, which was to evaluate systematic differences in ROM measurements between two commercially available markerless motion capture systems rather than establish criterion validity. Future studies incorporating both marker-based and multiple markerless systems would provide complementary information regarding both absolute measurement accuracy and inter-system comparability.
Surface conditions were not counterbalanced. All participants completed the movement in the same order (firm, compliant, then commode), which may have introduced potential order effects such as fatigue, adaptation, or task familiarization. As a result, some observed surface-related differences may reflect testing order rather than surface condition alone. Specifically, because the firm condition was always completed first, and the commode condition last, systematic changes in movement strategy may have been introduced. Indeed, some of the differences attributed to surface condition, particularly between the firm and commode conditions, could reflect the fixed testing sequence. However, the observed findings demonstrated distinct biomechanical patterns across the joint variables, rather than uniform directional changes across all measures. Surface-related effects were most pronounced for ankle dorsiflexion, knee flexion, and hip flexion variables, whereas hip adduction and trunk forward lean demonstrated comparatively smaller zROM differences across seating conditions. In addition, the strongest interaction observed in the model was the Device × Joint interaction, which suggests that system-related differences varied as a function of the biomechanical variable being assessed, rather than surface condition alone. Together, these findings suggest that the observed effects are unlikely to be attributable solely to sequencing effects. However, future studies should counterbalance testing order to better explore the influence of surface condition on ROM.
The absolute magnitude for hip adduction ROM was small across conditions, especially for the DARI and thus, measurement error may represent a greater proportion of the observed ROM. Minimum detectable angular thresholds were not specified by the manufacturers of either system; therefore, it cannot be determined whether these small frontal-plane hip excursions approached the sensitivity limits of either device. Thus, the comparatively large z-score standardized difference observed for hip adduction should be interpreted cautiously.
The use of within-joint z-score standardization allowed for comparison of biomechanically distinct joint-angle measurements to be evaluated within a single model, though this transformation does not preserve the absolute magnitude of between-device differences. Consequently, these standardized differences should not be interpreted as proportional or equivalent absolute measurement differences across the joint-angle measurements. For example, the same absolute difference in degrees may be relatively small for a joint-angle measurement with a large ROM but substantial for a measurement with a small ROM. To facilitate the interpretation of absolute magnitude of the observed device and surface differences, raw ROM values have been provided in Supplementary Table S1. The present study did not establish a priori thresholds for clinically acceptable between-system differences, and thus, statistical differences observed in this data should not be interpreted as evidence of clinical significance.
Additionally, differences in surface height may have contributed to observed changes in ROM. Finally, only two motion capture systems were examined, and results may not generalize to other technologies with different hardware configurations or analytical approaches.

6. Future Research

Future research should further examine measurement comparability across motion capture systems in more diverse populations and across a wider range of functional tasks. Additional work is needed to better understand the factors contributing to measurement differences, including system configuration and data processing methods. Investigating how these differences influence clinical interpretation and decision making over time may further clarify the role of motion capture technology in both practice and education.

7. Conclusions

This study evaluated differences in ROM measurements between two commercially available markerless motion capture systems during sit-to-stand performance and demonstrated systematic differences in range of motion measurements across multiple joint variables. These differences varied by movement and task condition, suggesting that systematic measurement differences may depend on the biomechanical variable and testing conditions. Surface conditions also influenced movement outcomes, and in some cases, affected the magnitude of differences observed between systems. These results highlight the importance of considering both task demands and measurement approach when interpreting movement data. From a clinical perspective, consistent use of a single motion capture system may be important when tracking performance over time, as differences between systems may reflect variation in measurement rather than true changes in movement. From an educational standpoint, these findings highlight the importance of teaching future PTs that measurement outputs may vary depending on the technology being used and that these differences should be considered when interpreting technology-derived data. Overall, this study highlights the importance of evaluating measurement differences between commercially available markerless motion capture systems as a complementary line of investigation alongside traditional criterion validation studies. Together, these approaches will enhance the interpretation and application of markerless motion capture technologies in rehabilitation, physical therapy education, and movement science.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/app16188906/s1, Table S1: Raw Range of Motion Means and Standard Deviations (SD) by Joint angle, Device, and Surface condition.

Author Contributions

The authors confirm contribution to the paper as follows: C.V.: concept development, data collection, content expertise, and manuscript preparation. C.I.: methodology refinement, study design, statistical analyses, and manuscript preparation. M.A.: data collection, content expertise, and critical manuscript revision. All authors have read and agreed to the published version of the manuscript.

Funding

This study received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Touro University Institutional Review Board (#26327 on 17 June 2025).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Acknowledgments

The authors used ChatGPT (OpenAI, San Francisco, CA, USA), version GPT-5.3 Instant, to assist with language refinement and organizational support during manuscript preparation. All scientific content, data interpretation, statistical analyses, and final manuscript decisions were reviewed and approved by the authors.

Conflicts of Interest

The authors report no conflicts of interest.

References

  1. Janssen, W.G.M.; Bussmann, H.B.J.; Stam, H.J. Determinants of the sit-to-stand movement: A review. Phys. Ther. 2002, 82, 866–879. [Google Scholar] [CrossRef] [Scilit]
  2. Lord, S.R.; Murray, S.M.; Chapman, K.; Munro, B.; Tiedemann, A. Sit-to-stand performance depends on sensation, speed, balance, and psychological status in addition to strength in older people. J. Gerontol. Ser. A Biol. Sci. Med. Sci. 2002, 57, M539–M543. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  3. O’sullivan, S.B.; Schmitz, T.J.; Fulk, G. Physical Rehabilitation; F.A. Davis: Philadelphia, PA, USA, 2019. [Google Scholar]
  4. Shumway-Cook, A.; Woollacott, M.H. Motor Control: Translating Research into Clinical Practice; Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2007. [Google Scholar]
  5. Sadeh, S.; Gobert, D.; Shen, K.-H.; Foroughi, F.; Hsiao, H.-Y. Biomechanical and neuromuscular control characteristics of sit-to-stand transfer in young and older adults: A systematic review with implications for balance regulation mechanisms. Clin. Biomech. 2023, 109, 106068. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  6. Alcazar, J.; Alegre, L.M.; Van Roie, E.; Magalhães, J.P.; Nielsen, B.R.; González-Gross, M.; Júdice, P.B.; Casajús, J.A.; Delecluse, C.; Sardinha, L.B.; et al. Relative sit-to-stand power: Aging trajectories, functionally relevant cut-off points, and normative data in a large European cohort. J. Cachexia Sarcopenia Muscle 2021, 12, 921–932. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  7. Bohannon, R.W. Reference Values for the Five-Repetition Sit-to-Stand Test: A Descriptive Meta-Analysis of Data from Elders. Percept. Mot. Ski. 2006, 103, 215–222. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  8. Steinbrink, G.M.; Martinez, J.; Swartz, A.M.; Strath, S.J. Sit-to-Stand Power Is a Stronger Predictor of Gait Speed than Knee Extension Strength. J. Funct. Morphol. Kinesiol. 2024, 9, 103. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  9. Yoshiko, A.; Ogawa, M.; Shimizu, K.; Radaelli, R.; Neske, R.; Maeda, H.; Maeda, K.; Teodoro, J.; Tanaka, N.; Pinto, R.S.; et al. Chair sit-to-stand performance is associated with diagnostic features of sarcopenia in older men and women. Arch. Gerontol. Geriatr. 2021, 96, 104463. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  10. Whatman, C.; Hing, W.; Hume, P. Physiotherapist agreement when visually rating movement quality during lower extremity functional screening tests. Phys. Ther. Sport 2012, 13, 87–96. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  11. Clark, R.A.; Pua, Y.-H.; Fortin, K.; Ritchie, C.; Webster, K.E.; Denehy, L.; Bryant, A.L. Validity of the Microsoft Kinect for assessment of postural control. Gait Posture 2012, 36, 372–377. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  12. Pfister, A.; West, A.M.; Bronner, S.; Noah, J.A. Comparative abilities of Microsoft Kinect and Vicon 3D motion capture for gait analysis. J. Med. Eng. Technol. 2014, 38, 274–280. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  13. Bonnechère, B.; Jansen, B.; Salvia, P.; Bouzahouene, H.; Omelina, L.; Moiseev, F.; Sholukha, V.; Cornelis, J.; Rooze, M.; Van Sint Jan, S. Validity and reliability of the Kinect within functional assessment activities: Comparison with standard stereophotogrammetry. Gait Posture 2014, 39, 593–598. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  14. Lam, W.W.T.; Tang, Y.M.; Fong, K.N.K. A systematic review of the applications of markerless motion capture (MMC) technology for clinical measurement in rehabilitation. J. Neuroeng. Rehabil. 2023, 20, 57. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  15. Needham, L.; Evans, M.; Cosker, D.P.; Wade, L.; McGuigan, P.M.; Bilzon, J.L.; Colyer, S.L. The accuracy of several pose estimation methods for 3D joint centre localisation. Sci. Rep. 2021, 11, 20673. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  16. Scataglini, S.; Abts, E.; Van Bocxlaer, C.; Van den Bussche, M.; Meletani, S.; Truijen, S. Accuracy, Validity, and Reliability of Markerless Camera-Based 3D Motion Capture Systems versus Marker-Based 3D Motion Capture Systems in Gait Analysis: A Systematic Review and Meta-Analysis. Sensors 2024, 24, 3686. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  17. Kidziński, Ł.; Yang, B.; Hicks, J.L.; Rajagopal, A.; Delp, S.L.; Schwartz, M.H. Deep neural networks enable quantitative movement analysis using single-camera videos. Nat. Commun. 2020, 11, 4054. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  18. Mündermann, L.; Corazza, S.; Andriacchi, T.P. The evolution of methods for the capture of human movement leading to markerless motion capture for biomechanical applications. J. Neuroeng. Rehabil. 2006, 3, 6. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  19. Ceseracciu, E.; Sawacha, Z.; Cobelli, C. Comparison of Markerless and Marker-Based Motion Capture Technologies through Simultaneous Data Collection during Gait: Proof of Concept. PLoS ONE 2014, 9, e87640. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  20. Muro-de-la-Herran, A.; Garcia-Zapirain, B.; Mendez-Zorrilla, A. Gait Analysis Methods: An Overview of Wearable and Non-Wearable Systems, Highlighting Clinical Applications. Sensors 2014, 14, 3362–3394. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  21. Chougule, A.; Dowsett, M.; Ekundayomi, D.; Machesney, A.; Mather, T.; Gompels, B.; McDonnell, S. Accuracy and Validity of 3D Markerless Motion Capture Compared to Marker-Based Systems for Lower-Limb Biomechanical Assessment: A Systematic Review. Sensors 2026, 26, 3956. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  22. Perrott, M.A.; Pizzari, T.; Cook, J.; McClelland, J.A. Comparison of lower limb and trunk kinematics between markerless and marker-based motion capture systems. Gait Posture 2017, 52, 57–61. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  23. Pottorf, O.A.; Greenberg, E.T.; Garcia, M.C.; Dewan, B.M. Comparison of peak joint angles from non-concurrent squat and upper-limb assessment using a markerless Kinotek system and a marker-based Vicon system. PeerJ 2026, 14, e21219. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  24. Yoma, M.; Llurda-Almuzara, L.; Herrington, L.; Jones, R. Reliability and validity of lower extremity and trunk kinematics measured with markerless motion capture during sports-related and functional tasks: A systematic review. J. Sports Sci. 2025, 43, 1703–1730. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  25. Johnson, P.C.D.; Barry, S.J.E.; Ferguson, H.M.; Müller, P. Power analysis for generalized linear mixed models in ecology and evolution. Methods Ecol. Evol. 2015, 6, 133–142. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  26. Green, P.; McLeod, C. Simr: An R Package for Power Analyses of Generalized Linear Mixed Models by Simulation, R Package version 1.0.5.; CRAN (The Comprehensive R Archive Network): Vienna, Austria, 2016.
  27. DARI Motion. DARI Motion. Available online: https://darimotion.com/ (accessed on 23 February 2026).
  28. Kinotek. Kinotek|Functional Movement Assessment System. (n.d.). Available online: https://kinotek.com (accessed on 23 February 2026).
  29. The Jamovi Project, version 2.6; Jamovi: Sydney, Australia, 2025. Available online: https://www.jamovi.org (accessed on 23 February 2026).
  30. Das, K.; de Paula Oliveira, T.; Newell, J. Comparison of markerless and marker-based motion capture systems using 95% functional limits of agreement in a linear mixed-effects modelling framework. Sci. Rep. 2023, 13, 22880. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  31. Kanko, R.M.; Laende, E.K.; Davis, E.M.; Selbie, W.S.; Deluzio, K.J. Concurrent assessment of gait kinematics using marker-based and markerless motion capture. J. Biomech. 2021, 127, 110665. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  32. Verheul, J.; Hughes, O.; Hitchens, L.; Atherton, T.; Sauter, T.; Radwan, A.; Pryse, E.; Mitchell, A.; Burton, S. Markerless motion capture for running: Validity and reliability of whole-body, joint, and muscle kinematics. J. Biomech. 2026, 195, 113133. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  33. Wade, L.; Needham, L.; McGuigan, P.; Bilzon, J. Applications and limitations of current markerless motion capture methods for clinical gait biomechanics. PeerJ 2022, 10, e12995. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  34. Jeon, W.; Hsiao, H.-Y.; Griffin, L. Effects of different initial foot positions on kinematics, muscle activation patterns, and postural control during a sit-to-stand in younger and older adults. J. Biomech. 2021, 117, 110251. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  35. Palazzo, F.; Nardi, A.; Lamouchideli, N.; Caronti, A.; Alashram, A.; Padua, E.; Annino, G. The effect of age, sex and a firm-textured surface on postural control. Exp. Brain Res. 2021, 239, 2181–2191. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  36. Chaumeil, A.; Lahkar, B.K.; Dumas, R.; Muller, A.; Robert, T. Agreement between a markerless and a marker-based motion capture systems for balance related quantities. J. Biomech. 2024, 165, 112018. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  37. Croci, E.; Künzel, C.; Gehring, D.; Müller, A.M.; Mündermann, A. Comparison of shoulder kinematics between Theia 3D markerless motion capture and marker-based motion capture during full arm ranges of motion. J. Biomech. 2025, 191, 112905. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  38. Nishikawa, N.; Watanabe, S.; Yamamoto, K. Comparison of kinematics between markerless and marker-based motion capture systems for change of direction maneuvers. J. Biomech. 2025, 192, 112965. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  39. Andrade, P.H.S.; de Osawa, C.R.; Aveiro, M.E.S.; Rocha, C.R.; Bazan, R.; Souza, L.A.P.S.; Luvizutto, G.J. Task-Based Mapping of Compensatory Strategies and Movement Kinematics After Stroke: A Systematic Scoping Review. Physiother. Res. Int. J. Res. Clin. Phys. Ther. 2026, 31, e70215. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  40. Fleisig, G.S.; Slowik, J.S.; Wassom, D.; Yanagita, Y.; Bishop, J.; Diffendaffer, A. Comparison of marker-less and marker-based motion capture for baseball pitching kinematics. Sports Biomech. 2024, 23, 2950–2959. [Google Scholar] [CrossRef] [Scilit] [PubMed]
Applsci 16 08906 g001
Figure 1. Estimated marginal mean zROM values for DARI and Kinotek across joint variables. These values represent model-estimated marginal means derived from the LMM following within-joint z-score standardization. The figure demonstrates a significant device × joint interaction, indicating that differences between the devices varied across biomechanical measures, with the largest differences observed for hip adduction and trunk forward lean. Error bars represent 95% CI. L = left, R = right, DF = dorsiflexion, and FL = forward lean.
Figure 1. Estimated marginal mean zROM values for DARI and Kinotek across joint variables. These values represent model-estimated marginal means derived from the LMM following within-joint z-score standardization. The figure demonstrates a significant device × joint interaction, indicating that differences between the devices varied across biomechanical measures, with the largest differences observed for hip adduction and trunk forward lean. Error bars represent 95% CI. L = left, R = right, DF = dorsiflexion, and FL = forward lean.
Applsci 16 08906 g001
Applsci 16 08906 g002
Figure 2. Estimated marginal mean zROM values across surfaces and joint variables. The values represent model-estimated marginal means derived from the LMM following within-joint z-score standardization. Error bars represent 95% CI. The figure demonstrates a significant surface × joint interaction, showing that surface differentially influenced biomechanical variables. The firm surface conditions consistently produce larger zROM values for ankle dorsiflexion, knee flexion, and hip flexion compared to the compliant and commode conditions. L = left, R = right, DF = dorsiflexion, and FL = forward lean.
Figure 2. Estimated marginal mean zROM values across surfaces and joint variables. The values represent model-estimated marginal means derived from the LMM following within-joint z-score standardization. Error bars represent 95% CI. The figure demonstrates a significant surface × joint interaction, showing that surface differentially influenced biomechanical variables. The firm surface conditions consistently produce larger zROM values for ankle dorsiflexion, knee flexion, and hip flexion compared to the compliant and commode conditions. L = left, R = right, DF = dorsiflexion, and FL = forward lean.
Applsci 16 08906 g002
Table 1. Demographic information (N = 50).
Table 1. Demographic information (N = 50).
Characteristic Category Mean (SD)n
Age 25.6 (3.84)-
Height (cm) 172.31 (10.89)-
Weight (kg) 79.11 (17.66)-
BMI (kg/m2) 26.49 (4.60)-
Gender Male -32
Female-18
Race Asian -4
Black/African American-3
Hispanic/Latino-5
Indian-1
White-35
White/Hispanic/Latino-2
Table 2. Linear mixed model omnibus effects.
Table 2. Linear mixed model omnibus effects.
EffectFdfp
Device16.121, 2613<0.001
Surface78.982, 2613<0.001
Joint0.008, 26131.000
Device × Surface3.462, 26130.032
Device × Joint47.348, 2613<0.001
Surface × Joint7.2616, 2613<0.001
Note. The non-significant main effect of joint reflects the use of within-joint z-score standardizations completed prior to analysis. ROM values are standardized separately for each joint because absolute ROM values differed substantially across biomechanical measures, making direct comparisons of raw joint angles inappropriate. Overall mean differences between joints are removed, allowing the model to evaluate relative differences across devices and surfaces independent of differences in absolute ROM magnitude. Model-level effect sizes are indexed using marginal and conditional R2 values due to the use of a linear mixed-effects modeling framework.
Table 3. Selected significant post hoc comparisons for interaction effects.
Table 3. Selected significant post hoc comparisons for interaction effects.
InteractionComparisonMean Difference95% CIp
Device × SurfaceDARI Hard vs. Kinotek Hard0.192[0.081, 0.303]0.010
DARI Soft vs. Kinotek Soft0.191[0.080, 0.302]0.011
Device × JointLeft Hip Adduction1.341[1.069, 1.613]<0.001
Right Hip Adduction1.130[0.858, 1.401]<0.001
Trunk/Spine FL 1.119[0.847, 1.390]<0.001
Left Hip Flexion−0.338[−0.610, −0.067]0.015
Right Hip Flexion−0.314[−0.586, −0.043]0.023
Surface × JointH vs. C: Left Ankle DF0.765[0.530, 1.001]<0.001
H vs. C: Right Ankle DF0.741[0.506, 0.976]<0.001
H vs. C: Left Hip Flexion0.472[0.236, 0.707]0.030
H vs. C: Left Knee Flexion0.725[0.688, 1.159]<0.001
H vs. C: Right Knee Flexion0.702[0.685, 1.115]<0.001
H vs. S: Left Ankle DF0.521[0.285, 0.756]<0.001
H vs. S: Right Ankle DF0.495[0.260, 0.731]<0.001
H vs. S: Left Knee Flexion0.629[0.394, 0.865]<0.001
H vs. S: Right Knee Flexion0.613[0.378, 0.848]<0.001
Note. Only statistically significant post hoc comparisons relevant to significant interaction effects discussed in text are displayed. Positive values indicate larger zROM values for the first condition listed. Confidence intervals are calculated using Wald-based estimates from the linear mixed-effects model. DF = dorsiflexion, FL = forward lean, H vs. C = hard vs. commode, and H vs. S = hard vs. soft.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Voltmer, C.; Imperio, C.; Apostol, M. Measurement Differences Between Markerless Motion Capture Systems During Sit-to-Stand Performance: Implications for Clinical Interpretation and Physical Therapy Education. Appl. Sci. 2026, 16, 8906. https://doi.org/10.3390/app16188906

AMA Style

Voltmer C, Imperio C, Apostol M. Measurement Differences Between Markerless Motion Capture Systems During Sit-to-Stand Performance: Implications for Clinical Interpretation and Physical Therapy Education. Applied Sciences. 2026; 16(18):8906. https://doi.org/10.3390/app16188906

Chicago/Turabian Style

Voltmer, Christopher, Casey Imperio, and Mark Apostol. 2026. "Measurement Differences Between Markerless Motion Capture Systems During Sit-to-Stand Performance: Implications for Clinical Interpretation and Physical Therapy Education" Applied Sciences 16, no. 18: 8906. https://doi.org/10.3390/app16188906

APA Style

Voltmer, C., Imperio, C., & Apostol, M. (2026). Measurement Differences Between Markerless Motion Capture Systems During Sit-to-Stand Performance: Implications for Clinical Interpretation and Physical Therapy Education. Applied Sciences, 16(18), 8906. https://doi.org/10.3390/app16188906

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop