Study at a glance
Summary
Disclaimer: The following summary details independent academic research. Theia3D is intended for research, sports performance and educational use only. It is not cleared as a medical device and is not intended to be used for the diagnosis, treatment, mitigation or prevention of any disease, injury or medical condition.
Editor's note: Theia3D version v2023.1.0.3161, Patch 9, as stated in the paper.
Study Overview
Design
A within-subject comparison of Theia3D markerless motion capture deployed in two different camera configurations and physical spaces, a traditional 8-camera laboratory setup and a constrained 10-camera hospital hallway setup, with each participant completing the same tasks at both sites on the same day.
Participants
Twenty-five healthy adults (15 women, 10 men; mean age 34 years) with no self-reported conditions affecting their gait.
Camera Setups
The laboratory setup used 8 Sony RX0 II cameras arranged in a circle approximately 2.7 meters apart, mounted at a height of approximately 1.9 meters. The constrained setup used 10 Sony RX0 II cameras spaced along an active hospital hallway (5 cameras ceiling-mounted at approximately 3 meters, 5 wall-mounted at approximately 1.9 meters) to maximize coverage of an approximately 7.6-meter walkway, with 2 additional cameras added to compensate for the hallway's narrower, curved geometry.
Data Collection
Participants completed two data collections on the same day, one at each site, in the same clothing and footwear, following identical protocols: 30 seconds of quiet standing, 60 seconds of self-selected walking, and a 5-repetition sit-to-stand task, each completed twice.
Analysis
All data were processed in Theia3D with an inverse kinematics model and filtered consistently across sites. Within-site and between-site agreement were assessed using Pearson correlations, root-mean-square error (RMSE), Bland-Altman limits of agreement (LOA), and intraclass correlation coefficients (ICC) for the hip-knee-ankle angle during standing, lower-limb joint angle waveforms during walking, and peak trunk flexion during sit-to-stand.
Key Findings
- Standing and walking outcomes showed excellent agreement between the laboratory and constrained hallway setups. ICC values were 0.97 or higher, RMSE stayed under 2.3 degrees, and mean differences stayed under 1.1 degrees for both tasks.
- Walking kinematic waveforms correlated strongly across all three planes of motion at both sites. Within-site Pearson correlations were 0.9 or higher for the sagittal, frontal, and transverse planes at the hip, knee, and ankle, with between-site correlations of 0.85 or higher; between-site RMSE was under 2 degrees for every joint except hip flexion/extension, which reached 2.3 degrees.
- The sit-to-stand task showed more trial-to-trial and between-site variability than standing or walking, though reliability remained high. Limits of agreement for peak trunk flexion ranged from approximately 12 to 20 degrees, wider than the other two tasks, while ICC values stayed above 0.96.
- A small systematic bias appeared toward higher joint angles at the laboratory site for the knee, and at the constrained site for the ankle, a pattern the authors attribute to the narrower capture volume and higher camera placement used at the constrained site.
- Average walking speed and quiet-standing alignment were nearly identical between sites, with no statistically significant difference in gait speed, supporting that participants moved comparably in both environments despite the difference in available space.
What This Means for Movement Science Researchers and Teams
For researchers and movement science teams considering markerless gait analysis outside a dedicated laboratory, this study offers direct, encouraging evidence: with sufficient camera coverage and overlap, even a hallway-constrained setup can produce lower-limb kinematic data essentially comparable to a traditional laboratory arrangement for standing and walking tasks. The authors do flag that lower camera placement, while initially appealing as a space-saving option, appeared to introduce small differences compared to a more traditional elevated setup in this study, so teams adapting markerless capture to constrained spaces should prioritize camera coverage and placement consistency over simply fitting more cameras into less space.
The Role of Theia3D in Movement Analysis
Theia3D was the markerless system evaluated in both configurations in this study, generating lower-limb joint kinematics from standing, walking, and sit-to-stand trials collected in a traditional laboratory and in a space-constrained hospital hallway. This let the researchers isolate how much of any difference in output was attributable to the camera configuration and physical space itself, independent of any change in the movement task.
Study Limitations and Future Directions
The authors note that the hospital hallway used in this study had architectural features, including wider recessed sections, that provided a central capture area at or above 3 meters wide, likely improving camera placement, overlap, and line-of-sight; narrower hallways, such as those around 2 meters wide, may not offer the same advantages and could yield different results. All participants were healthy, generally younger adults with no current gait abnormalities, so these findings may not generalize to populations with significant gait or morphological differences, although the authors note that related comparisons between markerless and marker-based capture in those populations have generally shown high agreement. Sit-to-stand speed was not standardized, which may have contributed to the variability observed in that task, and the authors recommend multiple trials for that measure given a possible learning effect. One author, Outerleys, is employed by Theia Markerless Inc. and provided guidance on study design and reviewed the manuscript, but was not involved in data collection, analysis, or interpretation of results; the authors disclose that Theia Markerless Inc. had no influence on the data collection, analysis, or interpretation of results.
Citation: Ruder, M., Di Bacco, V. E., Outerleys, J., Malek, M., Laende, E., Gafoor, F., Madden, K., Adili, A., Deluzio, K., & Kobsar, D. (2026). Evaluating the agreement of markerless motion capture for joint angle estimation in a constrained hallway setting compared with a traditional laboratory setting. Journal of Applied Biomechanics, Ahead of Print. https://doi.org/10.1123/jab.2025-0265
Link: Read the full study
Can Markerless Capture Deliver Lab-Quality Data Outside the Lab?
This study suggests that, with thoughtful camera placement, a constrained space like a hallway can produce lower-limb kinematic data essentially comparable to a dedicated laboratory setup. If your work involves bringing gait analysis into nontraditional spaces and you want to understand what that setup should look like, talk to our team about deploying markerless capture in your own environment.