Clip missing for this beat — not a substitute video.
1st half NEAR → 2nd half FAR
1st half FAR → 2nd half NEAR
Hypothesis: on both named contacts the smash is a timed whip — hip, then shoulder, then elbow, then wrist — not one arm heave. Method: one broadcast camera, 2D joints from the YOLO+RF-DETR cache, 03:06–03:36 / 186–216s PRE-switch. Markerless here means we read the picture. It is not a lab.
Mid-hip image-y vs pre-plant baseline ÷ local court-width scale at feet. Shoulder-line vs image horizontal. Same whip ms. Not force-plate Vert, not 3D IR, not °/s.
One camera, this Game 3 clip. Everyday kit. No markers, no wand, no capture volume.
shetty_shi_g3_detections.json.gz · 7426 frames · identity remap at 218s. COCO-style joints on the image.
Peak 2D joint speed around plant / jump-smash. Hip → shoulder → elbow → wrist. Projection-biased. Not °/s.
Same cache, same window. Degrees are image-plane proxies — not 3D X-factor, not lab IR. Do not rank raw jump lift across the net: Shetty is near.
.c3d export · Snap Score · Hawk-Eye2D peak timing · hip/shoulder can share a frame · not °/s · not lab kinematics
The jump is a hip lifting on the image. The feet stay a court-plane km/h story — same live overlay recipe, not a force plate.
Jump proxy is not metres of height. Feet km/h is court-plane from the existing cache, ESTIMATED at live overlay grain. Teal is always Shetty.
Start 17–6. Never concede five in a row while leading by 11. Repeat ×5.
Start mini-games at 3–3. Win the next 3 of 4 before leaving. Repeat ×5.
Score-tied Pulse drills on this Game 3 board. Visual court cards kept. BWF Coach Education (open kinetic chain) · King et al. 2020 (jump smash lab — not this camera).
BWF framework
· Coaching Framework PDF
· CE pathway Aug 2026
· King 2020
Same card language as the cream landing — dark Amicro, this 21–19 only. Karthik / Kavya / 17.4 km/h stay on the demo site.
Hip → shoulder → elbow → wrist. Same whip as Kinetic chain above. We time plant-pauses on this camera. We do not invent 4.2 rad/s or “2 of 7 attacking shots.”
Shetty 49 / 51 left–right on court. Even. The landing’s BH lunge −6% is not this pack. This 30s: Shetty 62 m vs Shi 78 m.
Capture frame-by-frame technique breakdowns like smashes, lunges and landings,
using biomechanical data
See how the player reaches and positions their body at the net.
See the jump and arm movement behind the smash.
See the compact swing that takes the shuttle short.
See how the player lands and begins recovery.
This sample shows how Deltaplay turns match data into insights for coaches. Imagine the same perspective on your own athletes.
Sample report · Includes measured, estimated, and illustrative content, as labeled.
Coverage = occupancy bins with a foot sample (8×14 grid, GPU cache, remap 218s). Not landing 100% / 48%. Avg km/h is full-game court-plane — slot spikes ~43 unused. A weekly academy hub would list many athletes; this file does not.
Identity remapped at 218s. Shetty more samples far (after switch). Shi more near. Same bar language as the academy week chart — different data.
Axes we have: metres, avg km/h, points, cell coverage, longest run. Dummy “U-17 elite BWF” spider from the landing is not drawn.
Same 30s rally as Physical Story. Three short annotated beats — not a 30-film library. Skeleton plus hip-line / shoulder-line on the picture, then a metric stack: hip sway · shoulder turn · trunk. Inspired by a student swing-review layout. Not a golf clone.
Hip sway is 2D hip-mid minus ankle-mid on the same frame (else foot_px), scaled by local court-width ppm. Shoulder turn is the shoulder-line vs image horizontal. Trunk is shoulder-line yaw minus hip-line yaw. Teal is Shetty. Magenta is Shi. Switch at 218s is unused.
Example technique visuals showing the parts of the game we coach toward. Photo is this Game 3 clip. Arrows, X, pocket and T are coaching overlays — not measured on this rally, not 142°.
Reaching low and wide, then recovering fast to centre.
Explosive jump and rotation.
Resetting to T / centre between shots.
Photo: this Game 3 clip. Arrows / X / pocket / T are coaching — illustrative, not measured 142°. Not Ridge academy stills.
Example academy view — sample data. 3 coaches, 11 athletes. This is what your weekly academy dashboard looks like; these are not real students. The Overview tab is the real 21–19 match. Every figure here is DEMO.
Open-high-low-close as group energy. Same candle grammar as Mono Charts — recreated here, no Mono tab required.
Sample data — not real students.
W1–W4 close climbs 52 → 65. Last week is the 7-day chart to the left.
One coach, many kids. Ribbon thickness follows this-week coverage. Hover a band for name + last-week Δ. Not serve-hold.
Sample data — not real students.
| # | Athlete | Coach | Coverage | Peak km/h | Rallies | Δ cov |
|---|
Sorted by this-week cell coverage. Sample data — not real students.
How much each student moved vs last week. Sparkline is a 4-week coverage series (W1→W4). Before/after is last week → this week.
| Pt | Journey | Board | Server | Won | Conf |
|---|
No shuttle landing scatter. No invented court-thirds / shot-mix. Those stay DEMO on Mono Charts.
Smash and recovery cards live on Overview → What to work on next (Pulse drills from this board). This tab stays the three score-shape games — not a second drill list.
Hero figures are stylized character portraits from stills — not broadcast video frames, not smash meshes.
This file does not overwrite coach_report.html.