Physical AI Lab A finite repeating panel has four zero-cost match hypotheses. An ideal Blender camera study separates correspondence ambiguity, local pattern changes and a wrong search interval.
Two ideal point contacts miss one spatial moment direction. Compare the rotated pair and an off-line third point, while keeping linear rank separate from friction feasibility.
Inside and outside origins can both produce a ray hit—and both can miss. A controlled Blender API experiment separates first surface intersection from origin containment.
An 18 mm deepest point does not provide 18 mm of depth across a finite flat tip. Compare first bottom contact with head seating in the complete stack.
A quarter-turn and a stationary wheel can be prescribed without a dwell lock. Check the finite pin route and what holds the indexed wheel.
Two cable legs do not guarantee constant 2:1 carriage travel. Compare a full stroke, changing wrap and the ideal route conditions behind the calculation.
A long marker baseline does not ensure a stable orientation frame. Compare transverse separation, a fixed displacement and undefined three-point cases.
Quaternion endpoint agreement does not determine a custom interpolation path. See exact midpoint cancellation and a finite tool comparison under NLERP.
A rotated child can lose its right angle under a stretched parent while its own scale stays at one. Check inherited world edges, not only local rotation fields.
A flat plane can bend in reconstruction even with exact distances. Match axial Z to an unnormalized ray and range to a unit ray, then test an off-axis point.