Yellow: source observations. Pink / cyan: original / candidate projections. Red ×: no visible match. Dashed: weak joint proxies. Lines link source points to model points; contours are finite garment edges, not identified material points.
Excluded-frame checks
Feature
Original px
Candidate px
Matched
Native-pixel RMS includes missing-match penalties. Ear fold: absolute placement. Ear motion pairs: frame-to-frame displacement error (two excluded pairs). Image scales differ by camera. These are previously examined frames, not independent validation.
Camera checkLimits
Two training shoulder times cannot uniquely separate body and shirt motion. The original shirt uses three geometric slack modes. The independent sewn shirt uses four static proportions. Neither has measured fabric stiffness. Cameras, physical scale, contact and force remain unresolved.
Rest geometry, same orthographic view. Red is a relative normal diagnostic, not tissue injury or volumetric inversion. The full screen also checks motion and attachment at 193 sampled times.
Yellow dots: source boundaries. Red/cyan: model arms. Tap images for full size.
Arms against the video
Cyan/red: left/right forearm outlines. White dots: modeled joints. Shirt edges are not skin measurements.
Original
Elbow-plane orientation
Show provisional boundary measurements
Visibility limits
No replacement for the primary twin.
Body repair trials
Geometry diagnostics only. Neither trial replaces the twin or validates contact forces.
Body contact errors
Red marks intersecting model surfaces. Left: visible surfaces. Right: hidden crossings shown through the body. These are model errors, not observed tissue damage.
Three sampled poses only. Local folds and contacts between distinct body parts need separate repairs. No candidate replaces the primary twin.