just a guess,
had some complex positioning to achieve as well, and the only safe method for me was to matrix.decompose the object, then use the outputs (so the quaternion in you case).
@rookie, have you checked that these differences aren’t an abstraction of 0 under eulers constant…?
If a number is followed by e-N this is typically a sign that the number is a numerical representation of a number divided by a power which results in numbers minutely close to 0 but not absolutely 0, this can look confusing in the console because some of those numbers can start with positive numbers, eg… 5.294685e-56
EDIT: a good way to check this is if a value contains “e-” you should be able to practically assume that it’s 0, or a representation thereof…