A Proof-of-Concept Comparison of Azure Kinect and Smartphone-based OpenCap for Driving Musculoskeletal Simulations: A Unified Framework with Foot-Ground Contact Modeling

Abstract

Background: Marker-based motion capture remains the gold standard for deriving lower-limb kinematics and kinetics, but its high cost, lengthy setup time, and large space requirements limit its widespread use. Markerless technologies, such as depth camera systems (e.g., Azure Kinect) and emerging smartphone-based pipelines (e.g., OpenCap), promise laboratory-grade motion capture without markers. However, the effects of their capture data on musculoskeletal multibody dynamics simulation outcomes remain insufficiently understood.

Methods: This study was aimed at developing a single lower-limb musculoskeletal model simultaneously driven by two markerless motion capture inputs (Azure Kinect and OpenCap), and benchmarking the resulting joint angles, ground-reaction forces, and joint contact forces against synchronous Vicon recordings. In gait trials, movements were simultaneously recorded with Azure Kinect, OpenCap, and a Vicon motion capture system. The collected data were processed and used as inputs to construct the musculoskeletal model, which was then combined with a foot-ground contact model to compute lower-limb joint angles, ground reaction forces (GRFs), and joint contact forces.

Results: The OpenCap-based model showed strong agreement with the Vicon-referenced model (ρ > 0.73) in hip flexion-extension (FE), knee FE, and ankle FE, with a root mean square error (RMSE) of 4.04° to 7.66°, Sprague and Geers magnitude error ( M) of −0.25 to −0.10, phase error ( P) of 0.08 to 0.25, and composite error ( C) of 0.16 to 0.39. Additionally, strong correlations (ρ > 0.77) in hip contact force, knee contact force, medial knee contact force, and ankle contact force were observed between OpenCap and Vicon, with an RMSE of 0.26 to 0.90 Body Weight (BW), M of −0.03 to 0.13, P of 0.05 to 0.08, and C of 0.08 to 0.17.

Conclusions: Overall, under the study conditions, the smartphone-based OpenCap preliminarily showed accuracy as a potential alternative to marker-based systems for estimating lower-limb biomechanics. However, given the small sample size and tasks restricted to walking, it is currently primarily suited for research settings or initial screening, rather than high-precision clinical diagnosis. Further studies in larger, more diverse cohorts and validation across dynamic activities are required to confirm and extend its applicability.

Keywords

Foot-ground contact / kinematics / kinetics / lower-limb biomechanics / markerless motion capture / musculoskeletal multibody dynamics

References

[1] Brockett CL, Chapman GJ. Biomechanics of the ankle. Orthop Trauma 2016; 30(3): 232-8. [PMID: 27594929 DOI: 10.1016/j.mporth.2016.04.015]

[2] Yang C, Wei L, Huang X, Tu L, Xu Y, et al. Comparison of lower limb kinematic and kinetic estimation during athlete jumping between markerless and marker-based motion capture systems. Sci Rep 2025; 15(1): 18552. [PMID: 40425708 DOI: 10.1038/s41598-025-02739-9]

[3] Aharonson V, Seedat N, Israeli-Korn S, Hassin-Baer S, Postema M, et al. Automated stage discrimination of Parkinson’s disease. BIOI 2020; 1(2): 55-63. [DOI: 10.15212/bioi-2020-0006]

[4] Xu D, Zhou H, Quan W, Ma X, Chon TE, et al. New insights optimize landing strategies to reduce lower limb injury risk. Cyborg Bionic Syst 2024; 5: 0126. [PMID: 38778877 DOI: 10.34133/cbsystems.0126]

[5] van der Veen SM, Bordeleau M, Pidcoe PE, France CR, Thomas JS. Agreement analysis between Vive and Vicon systems to monitor lumbar postural changes. Sensors 2019; 19(17): 3632. [PMID: 31438520 DOI: 10.3390/s19173632]

[6] Fregly BJ, Besier TF, Lloyd DG, Delp SL, Banks SA, et al. Grand challenge competition to predict in vivo knee loads. J Orthop Res 2012; 30(4): 503-13. [PMID: 22161745 DOI: 10.1002/jor.22023]

[7] Illés T, Somoskeöy S. The EOS™ imaging system and its uses in daily orthopaedic practice. Int Orthop (SICOT) 2012; 36(7): 1325-31. [PMID: 22371113 DOI: 10.1007/s00264-012-1512-y]

[8] Kernkamp WA, Van De Velde SK, Hosseini A, Tsai TY, Li JS, et al. In vivo anterolateral ligament length change in the healthy knee during functional activities-a combined magnetic resonance and dual fluoroscopic imaging analysis. Arthroscopy 2017; 33(1): 133-9. [PMID: 27663034 DOI: 10.1016/j.arthro.2016.07.008]

[9] Marra MA, Vanheule V, Fluit R, Koopman BHFJM, Rasmussen J, et al. A subject-specific musculoskeletal modeling framework to predict in vivo mechanics of total knee arthroplasty. J Biomech Eng 2015; 137(2): 020904. [PMID: 25429519 DOI: 10.1115/1.4029258]

[10] Chen Z, Zhang Z, Wang L, Li D, Zhang Y, et al. Evaluation of a subject-specific musculoskeletal modelling framework for load prediction in total knee arthroplasty. Med Eng Phys 2016; 38(8): 708-16. [PMID: 27245748 DOI: 10.1016/j.medengphy.2016.04.010]

[11] Xu D, Zhou H, Jie T, Zhou Z, Yuan Y, et al. Data-driven deep learning for predicting ligament fatigue failure risk mechanisms. Int J Mech Sci 2025; 301: 110519. [DOI: 10.1016/j.ijmecsci.2025.110519]

[12] Kakavand R, Ahmadi R, Parsaei A, Brent Edwards W, Komeili A. Comparison of kinematics and kinetics between OpenCap and a marker-based motion capture system in cycling. Comput Biol Med 2025; 192(Pt A): 110295. [PMID: 40311466 DOI: 10.1016/j.compbiomed.2025.110295]

[13] Calderone J, Outerleys J, Diaz Dilernia F, Mann S, Wood G, et al. Measuring static and dynamic lower limb alignment in patients with advanced knee osteoarthritis using markerless motion capture. Gait Posture 2025; 119: 157-62. [PMID: 40101546 DOI: 10.1016/j.gaitpost.2025.02.022]

[14] Curreli C, Di Puccio F, Davico G, Modenese L, Viceconti M. Using musculoskeletal models to estimate in vivo total knee replacement kinematics and loads: effect of differences between models. Front Bioeng Biotechnol 2021; 9: 703508. [PMID: 34395407 DOI: 10.3389/fbioe.2021.703508]

[15] Topley M, Richards JG. A comparison of currently available optoelectronic motion capture systems. J Biomech 2020; 106: 109820. [PMID: 32517978 DOI: 10.1016/j.jbiomech.2020.109820]

[16] Napoli A, Glass S, Ward C, Tucker C, Obeid I. Performance analysis of a generalized motion capture system using microsoft kinect 2.0. Biomed Signal Process Control 2017; 38: 265-80. [DOI: 10.1016/j.bspc.2017.06.006]

[17] Eltoukhy M, Kuenze C, Andersen MS, Oh J, Signorile J. Prediction of ground reaction forces for Parkinson’s disease patients using a kinect-driven musculoskeletal gait analysis model. Med Eng Phys 2017; 50: 75-82. [PMID: 29102274 DOI: 10.1016/j.medengphy.2017.10.004]

[18] Albert JA, Owolabi V, Gebel A, Brahms CM, Granacher U, et al. Evaluation of the pose tracking performance of the Azure Kinect and Kinect v2 for gait analysis in comparison with a gold standard: a pilot study. Sensors 2020; 20(18): 5104. [PMID: 32911651 DOI: 10.3390/s20185104]

[19] Skals S, Rasmussen KP, Bendtsen KM, Yang J, Andersen MS . A musculoskeletal model driven by dual Microsoft Kinect Sensor data. Multibody Syst Dyn 2017; 41: 297-316. [DOI: 10.1007/s11044-017-9573-8]

[20] Asadi F, Arjmand N. Marker-less versus marker-based driven musculoskeletal models of the spine during static load-handling activities. J Biomech 2020; 112: 110043. [PMID: 32950760 DOI: 10.1016/j.jbiomech.2020.110043]

[21] Uhlrich SD, Falisse A, Kidziński Ł, Muccini J, Ko M, et al. OpenCap: human movement dynamics from smartphone videos. PLoS Comput Biol 2023; 19(10): e1011462. [PMID: 37856442 DOI: 10.1371/journal.pcbi.1011462]

[22] Cerfoglio S, Ferraris C, Vismara L, Amprimo G, Priano L, et al. Kinect-based assessment of lower limbs during gait in post-stroke hemiplegic patients: a narrative review. Sensors 2022; 22(13): 4910. [PMID: 35808426 DOI: 10.3390/s22134910]

[23] Lima Y, Collings T, Hall M, Bourne M, Diamond L. Assessing lower-limb kinematics via OpenCap during dynamic tasks relevant to anterior cruciate ligament injury: a validity study. J Sci Med Sport 2023; 26(Suppl 2): S105. [DOI: 10.1016/j.jsams.2023.08.123]

[24] Peng Y, Zhang Z, Gao Y, Chen Z, Xin H, et al. Concurrent prediction of ground reaction forces and moments and tibiofemoral contact forces during walking using musculoskeletal modelling. Med Eng Phys 2018; 52: 31-40. [PMID: 29269224 DOI: 10.1016/j.medengphy.2017.11.008]

[25] Jung Y, Jung M, Lee K, Koo S. Ground reaction force estimation using an insole-type pressure mat and joint kinematics during walking. J Biomech 2014; 47(11): 2693-9. [PMID: 24917473 DOI: 10.1016/j.jbiomech.2014.05.007]

[26] Fluit R, Andersen MS, Kolk S, Verdonschot N, Koopman HFJM. Prediction of ground reaction forces and moments during various activities of daily living. J Biomech 2014; 47(10): 2321-9. [PMID: 24835471 DOI: 10.1016/j.jbiomech.2014.04.030]

[27] Ripic Z, Kuenze C, Andersen MS, Theodorakos I, Signorile J, et al. Ground reaction force and joint moment estimation during gait using an Azure Kinect-driven musculoskeletal modeling approach. Gait Posture 2022; 95: 49-55. [PMID: 35428024 DOI: 10.1016/j.gaitpost.2022.04.005]

[28] Horsak B, Eichmann A, Lauer K, Prock K, Krondorfer P, et al. Concurrent validity of smartphone-based markerless motion capturing to quantify lower-limb joint kinematics in healthy and pathological gait. J Biomech 2023; 159: 111801. [PMID: 37738945 DOI: 10.1016/j.jbiomech.2023.111801]

[29] Peng Y, Wang W, Wang L, Zhou H, Chen Z, et al. Smartphone videos-driven musculoskeletal multibody dynamics modelling workflow to estimate the lower limb joint contact forces and ground reaction forces. Med Biol Eng Comput 2024; 62(12): 3841-53. [PMID: 39046692 DOI: 10.1007/s11517-024-03171-3]

[30] Cao Z, Hidalgo G, Simon T, Wei SE, Sheikh Y. OpenPose: realtime multi-person 2D pose estimation using part affinity fields. IEEE Trans Pattern Anal Mach Intell 2021; 43(1): 172-86. [PMID: 31331883 DOI: 10.1109/TPAMI.2019.2929257]

[31] Michaud B, Begon M. ezc3d: an easy C3D file I/O cross-platform solution for C++, Python and MATLAB. J Open Source Softw 2021; 6(58): 2911. [DOI: 10.21105/joss.02911]

[32] Pellikaan P, Van Der Krogt MM, Carbone V, Fluit R, Vigneron LM, et al. Evaluation of a morphing based method to estimate muscle attachment sites of the lower extremity. J Biomech 2014; 47(5): 1144-50. [PMID: 24418197 DOI: 10.1016/j.jbiomech.2013.12.010]

[33] Andersen MS, Damsgaard M, MacWilliams B, Rasmussen J. A computationally efficient optimisation-based method for parameter identification of kinematically determinate and over-determinate biomechanical systems. Comput Methods Biomech Biomed Engin 2010; 13(2): 171-83. [PMID: 19693717 DOI: 10.1080/10255840903067080]

[34] Richards RE, Andersen MS, Harlaar J, Van Den Noort JC. Relationship between knee joint contact forces and external knee joint moments in patients with medial knee osteoarthritis: effects of gait modifications. Osteoarthritis Cartilage 2018; 26(9): 1203-14. [PMID: 29715509 DOI: 10.1016/j.joca.2018.04.011]

[35] Skals S, Jung MK, Damsgaard M, Andersen MS . Prediction of ground reaction forces and moments during sports-related movements. Multibody Syst Dyn 2017; 39: 175-95. [DOI: 10.1007/s11044-016-9537-4]

[36] Sprague MA, Geers TL. Spectral elements and field separation for an acoustic fluid subject to cavitation. J Comput Phys 2003; 184(1): 149-62. [DOI: 10.1016/S0021-9991(02)00024-4]

[37] Taylor R. Interpretation of the correlation coefficient: a basic review. J Diagn Med Sonogr 1990; 6(1): 35-9. [DOI: 10.1177/875647939000600106]

[38] Wang Y, Sun J, Li J, Zhao D. Gait recognition based on 3D skeleton joints captured by kinect. In: 2016 IEEE International Conference on Image Processing (ICIP), Phoenix, AZ, USA. IEEE; 2016. pp. 3151-5. [DOI: 10.1109/ICIP.2016.7532940]

[39] Kitaoka HB, Crevoisier XM, Hansen D, Katajarvi B, Harbst K, et al. Foot and ankle kinematics and ground reaction forces during ambulation. Foot Ankle Int 2006; 27(10): 808-13. [PMID: 17054883 DOI: 10.1177/107110070602701010]

[40] Alalem N, Gasparutto X, Rose-Dulcina K, DiGiovanni P, Hannouche D, et al. Validity of a single inertial measurement unit to measure hip range of motion during gait in patients undergoing total hip arthroplasty. Sensors 2025; 25(1): 3363. [PMID: 40968908 DOI: 10.3390/s25113363]

[41] McGinley JL, Baker R, Wolfe R, Morris ME. The reliability of three-dimensional kinematic gait measurements: a systematic review. Gait Posture 2009; 29(3): 360-9. [PMID: 19013070 DOI: 10.1016/j.gaitpost.2008.09.003]

[42] Guzik A, Drużbicki M, Wolan-Nieroda A, Turolla A, Kiper P. Estimating minimal clinically important differences for knee range of motion after stroke. J Clin Med 2020; 9(10): 3305. [PMID: 33076214 DOI: 10.3390/jcm9103305]

[43] Xu J, Wen Z, Augustine S, Sharir R, De Bleecker C, et al. Inter-session reliability of markerless lower extremity joint kinematics and kinetics for return-to-sport screening tasks. J Sports Sci 2025; 43(21): 2576-90. [PMID: 40757785 DOI: 10.1080/02640414.2025.2541490]

[44] Kamal KC, Kamal AM, Kamal D, Fugaru O, Matei D, et al. Gait analysis as a measure of physical performance in older adults with bilateral knee osteoarthritis. Medicina 2025; 61(12): 2118. [PMID: 41470120 DOI: 10.3390/medicina61122118]