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: this paper does not state a specific Theia3D software version or markerless camera hardware in its methods. This is one of the earliest published evaluations of Theia3D in this collection, published in 2022, reflecting the technology and available validation evidence at that time; several since-published studies, including others summarized on this hub, have since evaluated Theia3D's accuracy at the hip and in frontal- and transverse-plane motion more extensively and in larger samples.
Study Overview
Design
A short report combining the authors' own exploratory pilot comparison, three members of the research team performing three movements, each captured simultaneously by marker-based and markerless (Theia3D) motion capture, with their broader practical perspective on adopting markerless technology into existing movement science work.
Participants
Three members of the research team at the University of Delaware, selected to represent a range of body types and skin tones, performing the movements themselves rather than recruiting outside participants.
Data Collection
Each team member performed walking, an unweighted squat, and a single-leg forward hop, with each movement captured three times by Theia3D without reflective markers, then an additional three times with both systems recording simultaneously after markers were applied for the marker-based system. This design allowed the researchers to isolate how much of any difference in the markerless results was attributable specifically to the clothing and marker straps required for the marker-based system.
Analysis
The analysis was primarily descriptive, intended to share exploratory data for others to examine rather than to test formal statistical hypotheses. Waveform correlations and average differences between the markerless and marker-based systems were calculated for hip, knee, and ankle joint angles across all three movements and three planes of motion, and segment lengths generated by each system's underlying body model were compared across trials.
Key Findings
- Knee flexion-extension angles were the most comparable between the two systems across all three movements, with average differences of just 1 to 3 degrees, which the authors attribute in part to the knee having more exposed skin, giving the underlying computer-vision model a clearer, more consistent view of the joint.
- Hip and ankle sagittal-plane angles showed strong waveform correlation with marker-based motion capture but considerably larger systematic differences, notably 10 to 22 degrees at the hip and 3 to 8 degrees at the ankle, with the markerless system consistently reporting smaller angles than the marker-based system at all three joints.
- Side-to-side (frontal-plane) and rotational (transverse-plane) motion showed much weaker correlation between the two systems across all three joints and movements, leading the authors to decline drawing firm conclusions about markerless accuracy in these specific planes from their small, exploratory sample.
- Body-segment lengths generated by the markerless system varied more between trials than those from the marker-based system, and wearing the reflective markers and associated straps changed markerless-derived thigh segment lengths by up to about 4 centimeters, roughly 10% of thigh length, underscoring the importance of controlling for clothing when using markerless capture.
- Squat depth was noticeably shallower when participants wore the marker straps required for the marker-based system than when captured markerless without them, illustrating that marker-based capture itself can restrict natural movement in ways that markerless capture does not.
What This Means for Movement Science Researchers and Early-Adopter Teams
For research and movement-assessment teams considering markerless motion capture, this early, hands-on perspective offers a useful complement to more formal validation research: beyond the specific accuracy numbers, which have since been explored much more thoroughly in larger studies, it highlights practical, day-to-day considerations that matter for real-world adoption, including substantially larger file sizes, longer processing times, and the inability to review or correct a trial once it has been collected with markerless capture, unlike the real-time troubleshooting marker-based systems allow during a session. The authors' core message, that researchers and movement specialists should understand and weigh the specific trade-offs of each system for their own particular use case rather than assuming either system is a universal default, remains broadly relevant advice for teams evaluating markerless technology today, even as the specific accuracy figures reported here have been superseded by more extensive, better-powered validation research since this report was published.
The Role of Theia3D in Movement Analysis
Theia3D was the markerless motion capture system this research team trialed directly against their existing marker-based system, as one of the earliest hands-on evaluations of the technology from outside Theia Markerless itself. The authors' own framing, as scientists eager to adopt new technology but appropriately cautious about trusting it before understanding its specific strengths and limitations, reflects a useful, transparent approach to evaluating a new measurement tool, and their shared raw dataset was offered as a resource for other researchers to explore independently.
Study Limitations and Future Directions
The authors are explicit that their study was exploratory and descriptive rather than a formal validation study, using only three participants, members of their own research team rather than a broader recruited sample, and that their findings regarding frontal- and transverse-plane motion in particular should not be treated as definitive given the small sample size. They note that no independent, gold-standard reference such as biplanar fluoroscopy was used to establish which system, marker-based or markerless, more accurately reflected true underlying joint motion, meaning their comparison could only establish agreement between the two systems rather than confirm which was correct. The authors call for further research with larger sample sizes to more confidently characterize markerless capture's accuracy outside the sagittal plane, and encourage other researchers to explore their openly shared dataset directly.
Citation: Ito, N., Sigurðsson, H. B., Seymore, K. D., Arhos, E. K., Buchanan, T. S., Snyder-Mackler, L., & Silbernagel, K. G. (2022). Markerless motion capture: What clinician-scientists need to know right now. JSAMS Plus, 1, 100001. https://doi.org/10.1016/j.jsampl.2022.100001
Link: Read the full report
An Early, Hands-On Look at Markerless Motion Capture
This 2022 perspective from a University of Delaware research team shares their own hands-on pilot comparison of Theia3D against marker-based motion capture, finding close agreement at the knee but larger differences at the hip and outside the sagittal plane, alongside candid practical trade-offs in file size and processing time. If your work involves evaluating movement-assessment technology, talk to our team about where Theia3D stands today.