If we wanted to animate a transition between the initial state and A_0 instead, would we still need to decompose A_0 into rotation and scaling or would taking the linear combination of A_0 and the identity matrix work?
If we just ignore X(t) in the transformation, how can we guarantee the composite transformation of Q(t) and P(t) together will generate a valid interpolation? Why can we just drop X(t) in this case?
If we drop X(t), wouldn't this result in wrong endpoints? Q(0) is not always equal to Q_0, and Q(t) is not always equal to Q_1.
What exactly does x(t) represent and is it still the same rotation as (1-t)Q_0 + tQ_1?
How did the shading on the cow change?
why can we throw away x?
How do people come up with more realistic ways of interpolating transformations? I'm probably lacking a lot of intuition but I don't understand how they decided to use a decomposition. Also, are there different ways of interpolating transformations used depending on the object being transformed?