Motion capture and video analysis of the shot in archery research

What a motion capture system records

A motion capture system does not see the body: it sees points. Several infrared cameras track reflective markers fixed to the skin or the bow, and from their positions in three dimensions come angles, distances, speeds, the duration of each phase and the way two joints are coordinated. In 1990, at Loughborough, an optical scanner followed just three markers, on the head, on the elbow of the drawing arm and on the bow's pressure button, at 30 samples per second, with a standard deviation of 0.45 mm on the lateral axis when nothing moved (nine British archers). The Korean, Chinese, Polish and Malaysian studies of the last twenty years use between five and twelve infrared cameras, at 100 to 200 frames per second, and up to thirty-nine markers. Some recent studies replace the cameras with inertial sensors fixed to the body, as many as 17 in the thirteen university students of an American study: they need no laboratory, and they are covered in the section on wearable sensors.

Alongside these set-ups there is ordinary video, analysed with free software such as Kinovea. A camera running at 30 frames per second beside the archer, with three markers, was enough to time the phases of the shot in sixteen archers and to measure an angle on the image plane; it is not enough for the release, which lasts a few thousandths of a second and needs high-speed video. How the same angle changes from one instrument to another is explained in the section on joint angles below.

How the studies have used it

The first use is comparing levels. In a Japanese study from 1997, archers with at least three years of practice opened the angle between the shoulders and the string elbow to about 155 degrees, and beginners to about 140: this is the entry on the string elbow in experienced archers and beginners, which rests on means read off a graph, with no standard deviations. The same design, with far finer instruments, returns in 28 Chinese archers filmed by eight cameras at 200 Hz, where one of the three components of the coupled movement of shoulder and elbow separates the levels, and in 14 Polish national-team archers compared with 14 people who had never shot.

The second use, closer to the questions an archer actually asks, compares the good and poor arrows of the same archer. In 2007 the archer then ranked second in the world shot 60 arrows at 70 metres outdoors in front of seven infrared cameras running at 200 frames per second; a year later eight university archers in Seoul did the same. The cleanest design is that of six archers from a national squad, with ninety arrows analysed per archer at 18 metres: in the arrows scoring 9 and 10, shoulders, elbows and wrists vary so that their changes cancel out and the arrow keeps its direction, with an index of 0.4 to 0.6 at the release against 0.1 to 0.3 in the arrows scoring 6 to 8. The entry on joints compensating for one another sets out the detail.

The third use measures consistency, and here the findings do not all point the same way. Among high-school kyudo archers, the position of the elbow changed between two arrows by 0.706 mm in the most regular student and by 36.274 mm in a beginner; in the nine British archers of 1990, by contrast, consistency of posture at the release did not separate the elite from good county-level archers. The entry on variability in experienced archers holds both findings together. Finally, recorded movement is used as input for models: the draw of a single archer is one of the bases of the entry on shoulder load during the draw, where the muscles are calculated rather than measured.

Where the measurement stops

The first limit is the setting. Infrared cameras need a laboratory, and the shot gets shorter to fit inside one: 5 metres in the Polish comparison, 6 for the ten Malaysian traditional archers, 8 for the nine Korean women archers, 12 in the Italian study from 1994, 18 for the six national-squad archers, and 20 for the Korean para archers, chosen to set up a three-dimensional analysis with twelve cameras. In the Polish comparison the beginners did not release: for safety they stopped at full draw and let down.

The second limit is the plane. A single camera measures on the image plane, and when the movement leaves it the figure changes without warning. In 2010 three women from the German national team shot 12 arrows each at 30 metres while three systems measured aiming sway at the same time: an optoelectronic device on the bow, 3D video with two cameras and 2D video with one camera diagonally behind the archer. Vertically, the 2D video agreed with the device on the bow, with correlations from 0.69 to 0.97; horizontally they ranged from -0.31 to 0.25, none of them significant, and the archer with the poorest agreement was the one who rotated the bow most. This is the entry on the camera and aiming sway.

The third limit lies in the arithmetic. Few archers and many measurements make high correlations easy: with five Indian archers a correlation is significant only above 0.878, and fifteen angles out of eighteen cleared that bar, with scores ranging from 9.9 to 10. In the eight Seoul university archers, the line between shoulder speed in the follow-through and score, with an R squared of 0.874, is fitted to four points, the means of the arrows scoring 7, 8, 9 and 10; this is one reason the entry on a quick release and follow-through is contested. Two arrows per archer, as in the kyudo study, say little about consistency.

What it can tell an archer or a coach

The most solid finding is that movement explains little of the score. In the archer then ranked second in the world, five shoulder and trunk parameters over 60 arrows explain 17.9% of the arrow's score; in the eight French elite archers, shoulder and elbow angles and chest movement do not separate the arrows scoring 8, 9 and 10 in any archer. This is the entry on movement parameters and score. A mean angle therefore describes how an archer shoots, not where the next arrow will land.

The second concerns individual technique. In 1994 the recordings of the twelve Italian archers showed each of them reaching full draw in their own way and repeating it from one arrow to the next, and the registry rates as refuted the idea of a single correct elite technique. For a coach, then, how far an angle sits from a reference value matters less than whether it stays the same from arrow to arrow, and whether the joints compensate for one another when something changes. What falls with experience, the entry on variability concludes, is the variability of whatever matters for where the arrow goes.

For anyone filming themselves on a phone, finally, a fixed camera, side-on or from above, measures the angles and timings that lie in its plane well and those that leave it poorly; draw and aiming times can be counted even at 30 frames per second, the release cannot. How to use video in training is covered in the article on video analysis and self-coaching.

How design and sample are weighed is explained in How we read a study, the history of consistency measurements in the line on movement variability and individual technique, and the practical side in the article is there one correct archery technique.

@misc{campagna2026_enresearchmethodsmotioncapturevideo,
  author = {Campagna, Matteo},
  title = {Motion capture and video kinematics},
  year = {2026},
  howpublished = {Research, matteocampagna.com},
  url = {https://matteocampagna.com/en/research/methods/motion-capture-video},
  note = {accessed 2026-10-16}
}