Study at a glance

In 14 sport science students sprinting across three phases of acceleration up to top speed, Theia3D-based markerless motion capture showed reasonable agreement with marker-based motion capture for most lower-limb and trunk angles, but the authors describe the overall evidence as mixed, with pelvis angle reliability poor to moderate and larger discrepancies appearing at top-speed ground contact, leading them to conclude that no clear, definitive answer could yet be given about markerless motion capture's suitability for sprint kinematics assessment.

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: The paper does not report a specific Theia3D software version number. The authors describe this as an exploratory pilot study and are explicit that their results do not support a definitive conclusion either way about markerless motion capture's overall reliability for sprint kinematics.

Study Overview

Design

A concurrent, within-subjects comparison study evaluating the reliability and validity of markerless (Theia3D) motion capture against a reference marker-based system for lower-limb and trunk sagittal-plane kinematics across three phases of the sprint acceleration spectrum.

Participants

Fourteen sport science students (12 men, 2 women; mean age 29.1 years; height 1.79 m; mass 74.0 kg), physically active and free of recent lower-limb trauma, none of whom were sprint specialists.

Data Collection

After a sprint-specific warm-up, participants completed nine total sprints, three each at three defined phases of acceleration: an early acceleration phase captured within the first four steps from a standing start, a mid-acceleration phase at roughly half of each participant's eventual top speed, and a top-speed phase captured once the athlete reached an upright running posture at or near maximal velocity. Marker-based data were captured using 12 infrared cameras (200 Hz) with 45 reflective markers, and markerless data were captured simultaneously using eight Qualisys Miqus video cameras (85 Hz) processed through Theia3D, with six embedded force platforms recording ground reaction force to precisely identify touchdown and take-off for each stride.

Analysis

Sagittal-plane joint and segment angles for the ankle, knee, hip, pelvis, and trunk were computed for both systems using consistent biomechanical modeling in Visual3D. Agreement between systems was assessed across the full movement curve using Standard Error of Measurement (SEM) against an established reliability threshold, and using Functional Intraclass Correlation Coefficients (ICC); agreement at specific discrete points (touchdown and take-off) was separately assessed using Bland-Altman limits of agreement, and the effect of running speed on measurement differences was tested statistically.

Key Findings

  • Measurement error (SEM) stayed below the study's five-degree "reasonable agreement" threshold for all joints and segments across all three running phases, with one key exception: hip and pelvis angles, where this threshold was crossed, and foot angle specifically at top speed also showed a significant difference between systems.
  • Reliability (Functional ICC) was excellent for hip, knee, foot, shank, and thigh angles across all three running phases, but only moderate for pelvis and trunk angles at top speed, and poor for pelvis angle during the early- and mid-acceleration phases.
  • The authors identify an important nuance behind the pelvis result: pelvis angle showed an absolute measurement error similar to the hip, but because the pelvis moves through a much smaller overall range of motion during sprinting, that same absolute error produced a much weaker reliability statistic for the pelvis than for the hip.
  • Running speed had a statistically significant effect on measurement differences for knee, foot, and shank angles specifically, with larger differences at top speed compared to the slower running phases, while ankle, hip, and trunk angle differences were not significantly affected by speed.
  • Discrepancies between the two systems increased specifically at the moment of ground contact (touchdown) during top-speed running, a pattern the authors attribute to greater soft-tissue movement affecting the marker-based system as ground-reaction forces rise with running speed, and the authors conclude that their results do not allow a clear, definitive answer about markerless motion capture's overall suitability for sprint kinematics assessment, particularly regarding the pelvis.

What This Means for Movement Science Researchers and Sprint and Team-Sport Performance Teams

For researchers and performance staff working with sprint running, this study offers a specific, usefully qualified picture rather than a simple validation result: markerless motion capture showed strong, consistent reliability for the hip, knee, ankle, foot, shank, and thigh across the full acceleration spectrum, even up to top speed, supporting its use for the lower-limb measures most commonly relied on in sprint technique analysis. The pelvis finding offers an equally specific and practically important caution, not because the markerless system necessarily performed worse in absolute terms, but because the pelvis's smaller range of motion during sprinting makes agreement between any two motion-capture systems statistically harder to establish for that particular segment specifically, a nuance teams should keep in mind if they rely on pelvis angle data from sprint assessments. The authors' own candor about not being able to draw a clear, definitive conclusion is itself a useful, practically important finding: it signals that sprint-specific markerless validation, especially at genuinely maximal speeds, remains an active and still-developing area rather than a fully settled question.

The Role of Theia3D in Movement Analysis

Theia3D was the markerless motion-capture system directly evaluated against marker-based motion capture in this study, specifically across a range of sprint running speeds that extends meaningfully beyond what most prior markerless validation research has tested. The study's joint-by-joint, phase-by-phase approach, and its explicit, carefully qualified conclusion that results were mixed rather than uniformly favorable, offers a transparent and practically useful picture of where markerless sprint analysis currently stands: strong for most lower-limb measures, less settled for the pelvis specifically, and an area the authors themselves frame as an ongoing validation project rather than a closed question.

Study Limitations and Future Directions

The authors identify several specific limitations of their own pilot study. An imperfect synchronization between the two motion-capture systems, caused by their differing sampling frequencies and internal clocks, may have introduced some bias into the comparison, and the study's approach assumed, without being able to fully confirm, that both systems captured peak joint values at the exact same instant. Because participants were tested in only a single session, the between-session reliability of the markerless system for sprint running remains unknown. Certain outlier values observed in the data were attributed to a modeling issue in the Visual3D software used to process both systems' output, rather than necessarily reflecting a limitation of the markerless system itself. The authors also caution that their sample of sport science students, none of whom were sprint specialists, may not generalize to higher-performance athletic populations who could show meaningfully different sprint technique and reach higher top speeds. They identify testing more skilled sprint populations, exploring early acceleration kinematics from a standing or blocked start, improving inter-system synchronization, and continuing to re-assess markerless performance as the software is updated as important next steps.

Citation: Crespeau, S., Soares da Silva, E., Ravel, A., Rossi, J., & Morin, J.-B. (2026). Concurrent assessment of sprint running lower-limb and trunk kinematics using marker-based and markerless motion capture. Journal of Sports Sciences, 44(10), 1334-1347. https://doi.org/10.1080/02640414.2025.2528439

Link: Read the full study

How Reliable Is Markerless Motion Capture for Sprint Running Kinematics?

This study of 14 sport science students sprinting across three phases of acceleration finds that Theia3D markerless motion capture showed reasonable agreement with marker-based motion capture for most lower-limb and trunk angles, while the authors describe pelvis angle reliability as poor to moderate and conclude that a clear, definitive answer about overall suitability for sprint kinematics assessment is not yet possible. If your work involves sprint or team-sport performance research, talk to our team about applying Theia3D to your own studies.

Recent Posts