Exercise 2: What's Correct and What's Not About the "Quaternions Eliminate Gimbal Lock Everywhere" Claim — Possible Solution ==================================================================== WHAT'S CORRECT ------------------------------ It's correct that switching the engine's INTERNAL orientation representation and composition logic to quaternions genuinely eliminates the numerical problems this course verified with Euler angles - Chapter 6 showed that composing sequential yaw/pitch/roll rotations breaks down at pitch=90 degrees, losing a degree of freedom in a way that depends on the specific Euler-angle parameterization being used. This chapter verified that quaternion composition never routes through that parameterization at all, staying numerically smooth and well-behaved through the exact orientation that broke the Euler-angle approach. So for anything the engine computes internally by composing or interpolating orientations - character animation, camera movement, physics - the developer's claim is accurate. WHAT'S NOT CORRECT ------------------------------ The claim overreaches specifically regarding the UI that displays yaw/pitch/roll to the player. This chapter verified directly that converting a quaternion back into Euler angles for display purposes still exhibits the identical symptom: two different original yaw/roll inputs (30/10 and 40/20, both with pitch=90) produced not just the same underlying quaternion, but the same EXTRACTED Euler angles (180, 90, 180) when converted back for display. If this UI needs to show the player numerical yaw/pitch/roll values, it would still experience the same loss-of-independent-information symptom at that exact orientation - the displayed yaw and roll values would still become coupled/redundant there, regardless of the fact that the engine's internal math never touched an Euler-angle representation. WHY THIS DISTINCTION MATTERS ------------------------------ The underlying reason is that gimbal lock, in its fullest sense, is not really a "bug" that a smarter representation eliminates - it's a description of a genuine mathematical fact: at that specific orientation, the (yaw, roll) decomposition itself doesn't carry independent information, no matter how the composition was computed internally. Quaternions fix the COMPUTATIONAL symptoms (numerical instability, coordinate singularities during composition/ interpolation) but cannot restore information that was never actually present in the orientation to begin with. WHY THIS WORKS AS AN ANSWER ------------------------------ The answer separates the developer's claim into the part this chapter's own evidence directly supports (internal composition and interpolation) and the part it directly contradicts (Euler-angle display/extraction), using this chapter's own specific verified 30/10 vs. 40/20 finding as concrete evidence for the second part, rather than treating the claim as entirely right or entirely wrong.