Experiment 03 · figure

Full & Empty (虚实)

Is the weight in tai chi a single conserved sum, continuously poured between the two legs?

Figure 1. The form in motion — the reconstructed skeleton (left), each leg shaded by its load, synced to the weight trace (right) as a playhead sweeps 26–40 s. The right leg goes full as the boundary dips; the left as it climbs. Same data, now in a body. · CMU Mocap 12_04 · taiji_research/02-form-in-motion
Weight share on each leg across a 148-second tai chi sequence, drawn as two stacked bands that always sum to 100 percent; the boundary between them pours up and down as weight shifts leg to leg, sitting near 50 percent during the even opening stance.
Figure 2. Share of body weight on each leg across one 148 s tai chi sequence (CMU 12_04). The two shares always sum to the whole; the boundary is simply where the weight is placed. Quasi-static estimate from center-of-mass geometry (Winter 2009 segment masses). Data: CMU Mocap 12_04 · Fig: taiji_research/full-and-empty

What it shows

The classics name the weighted leg full (實) and the unweighted leg empty (虚), and insist the two are one relation, not two things. Here that relation is measured. The two bands are the share of body weight on each leg; they always sum to the whole, so the figure draws a conserved quantity being placed, not created or destroyed.

Three things read straight off it. The opening holds near 50/50 for the first fourteen seconds — the even, undifferentiated stance, wuji before the weight has been given anywhere. Then the form begins and the boundary pours: roughly twenty deep, near-complete transfers from one leg to the other. And the transfers are near-total — not the shallow 55/45 of walking, but the full commitment that frees the empty leg to step. To root is to place fullness here so that emptiness — and therefore freedom to move — is available there.

Method

Real motion capture, with no model of “tai chi” baked in: the CMU 12_04 trial (a recorded tai chi form) → forward kinematics to 3D joint positions → a whole-body center of mass via Winter’s segment masses → a quasi-static estimate of how weight shares between the feet from where the center of mass sits over its base. The full derivation, provenance, and license live in the figure’s provenance.json sidecar and the taiji_research pipeline.

The estimate is quasi-static — it reads load from geometry, valid because tai chi moves slowly enough that accelerations are small. It is not measured ground-reaction force, and it is labeled as such wherever it appears.

Journal

Sources & further reading

  1. Carnegie Mellon University. CMU Graphics Lab Motion Capture Database — subject 12, trial 12_04 (tai chi)
  2. David A. Winter. Biomechanics and Motor Control of Human Movement (2009)