Markerless motion analysis makes step rate, angles and trunk position measurable — force and impulse never: the honest reach of start video analysis.
Straight talk: video analysis of the skeleton start is not a makeshift tool but an established method — yet it only carries half of the markers that make up your push. A markerless motion analysis system was developed, evaluated and applied in the field specifically for the push-start technique of elite skeleton athletes: step rates, angles and trunk position can be captured without markers on the body [1]. So technique in skeleton is objectively assessable, not just a coach's feeling.
The other half is invisible. The start analysis works with quantities no camera delivers: peak horizontal force around 827 ± 217 newtons and a mean horizontal impulse around 264 ± 68 newton-seconds [2]; plus peak watts and power drop from a ten-second test [3]. An image shows you how you move — not how hard you push.
And before you rebuild your training around a video, some humility belongs in the room: the relationship between push time and finish time varies across tracks between roughly 0.57 and −0.14, averaging about 0.21 [4]. A better start video is not automatically a better finish time.
Markerless analysis was built in skeleton to capture step rates, angles and trunk position in the field [1]. Those three are your legitimate questions to the image. Anything beyond that — "looks powerful", "seems explosive" — is interpretation and does not belong in a protocol. A video without a question formulated beforehand produces opinions, not findings.
The value of an analysis lies in comparability across weeks. Fixed camera position, fixed height, fixed frame rate, a scale in frame — and the same section of the approach. If you stand wider in autumn than in spring, you are measuring your tripod spot, not your push.
The image carries two quantities particularly well. First step frequency: the skeleton start demands maintaining maximum power at the highest possible step frequency [3] — and frequency is simply countable on video. Second the step-length profile: the push-off step sits at around 1.60 ± 0.14 metres, the following step at around 2.09 ± 0.18 [2]. The start opens up. Whether your steps follow that pattern or stay uniform is the single most informative observation a video delivers.
What you do not see: horizontal force (around 827 ± 217 newtons) and horizontal impulse (around 264 ± 68 newton-seconds) [2] — and force and power variables demonstrably contribute significantly to start performance [2]. Equally invisible: the power drop across ten seconds, which shows how long you hold output [3]. Two athletes can look identical and push completely differently. Inferring force from an image is guessing.
Even a cleanly improved push is no guaranteed time gain: the correlation with finish time ranges by track from roughly 0.57 to −0.14, averaging about 0.21 [4]. On some tracks a better start pays clearly, on others hardly at all. Use video to make your push more efficient — not to shift your entire training budget there.
Video analysis tempts you into repetition. Start simulations are maximal sprints with one-handed guidance and high force delivered forwards — only record as many attempts as you can run fresh. The twelfth attempt for the perfect image is the one where the Achilles tendon and lower back pay.
And one professional limit applies to the interpretation: the kinetic quantities that really carry your start performance [2] belong in the hands of performance diagnostics and your coach. Nobody should derive technique changes for the ice track from a self-shot video alone — the image is a diagnostic tool, not a clearance.
Set yourself a fixed check block of three numbers and compare only those across the season: steps to the 20-metre mark, ratio of push-off step to following step, trunk angle at the third step. Three numbers, the same camera, the same spot — nothing more is needed. That is exactly where markerless analysis has its value: it makes technique repeatedly measurable in the field [1], and repeatable beats elaborate.
Step rates, joint and trunk angles and trunk position — markerless motion analysis was developed and field-applied in elite skeleton for exactly that [1]. Particularly readable are step frequency [3] and the step-length profile from a push-off step around 1.60 metres to a following step around 2.09 metres [2].
Everything kinetic. Peak horizontal force around 827 ± 217 newtons and horizontal impulse around 264 ± 68 newton-seconds come from force measurement, not from the image [2] — and force and power variables contribute significantly to start performance [2]. The ten-second power drop is likewise only visible in a power test [3].
For step frequency and rough trunk position often yes, provided frame rate, position and framing stay constant. For angles and step lengths you need a high frame rate, a perspective perpendicular to the running direction and a scale in frame — otherwise you measure perspective instead of technique. That is experience, not a study standard.
Not against a norm — no published target value for dive timing exists. All that is documented is that the start is a hybrid of sprint acceleration and dive [2]. So work in before-and-after comparison with yourself and couple the image to splits at 5, 10 and 20 metres [3] instead of measuring against an invented reference.
It is the most direct video indicator of the central start requirement: maintaining maximum power at the highest possible step frequency [3]. If your frequency drops in the later approach while step lengths stay the same, that is a fatigue or technique signal — and one of the few findings a video alone delivers reliably [1].
What Separates Elite Starters: The Biomechanics of the Fastest Skeleton Push