When using the 90/±90 angles, have you tried measuring the exact same point to point locations? If so, what is the measurement?
The short answer to your question is, yes, centroid to centroid; however, other things come into play too (which plane is picked first [it becomes the datum feature], the form of the planes [are they truly parallel to each other, perfect vectors, etc], the alignment, etc.).
Planes, unlike points, extend to infinity. If one plane is skewed, the 3D measurement will evaluate perpendicular to the 1st plane to the closest intersection point of the 2nd.
Try turning your graphics on inside the dimension and look at where it intersects the 2nd plane.
When using the 90/±90 angles, have you tried measuring the exact same point to point locations? If so, what is the measurement?
The short answer to your question is, yes, centroid to centroid; however, other things come into play too (which plane is picked first [it becomes the datum feature], the form of the planes [are they truly parallel to each other, perfect vectors, etc], the alignment, etc.).
Planes, unlike points, extend to infinity. If one plane is skewed, the 3D measurement will evaluate perpendicular to the 1st plane to the closest intersection point of the 2nd.
Try turning your graphics on inside the dimension and look at where it intersects the 2nd plane.