I wrote preset with expressions in AE cs4 and it was calculating the direction of layer's motion.
It worked perfectly in CS4.
In CS5 the resulting vector very often has a length of 0 - especially with Nulls that have tracking data applied.
Even though the Null IS moving around.
One workaround I found was to use a BakeTransforms script on the Null layer.
The motion doesn't change, but now for some reason position.vector gives a proper value.
After some testing it seems that if a Null gets tracking data applied, the position.velocity vector for that null is always invalid
Thanks, it worked but now I have a problem with border frames
(first or last) as with sampling at slightly different time ends up with no motion next to one of them .
I did a workaround by changing the time delta for this trick dependant on the time value but still it's not full proof.
Any idea why velocity at the exact frame time gives a zero vector?
The speed value is correct so there is motion in these frames (it's not an issue of frame easing)
And again - this totally worked in CS4 so what gives?
That's what I needed, totally forgot that position has two different easing modes (temporal and spatial).
But still I'm totally positive I didn't have to do it in CS4 - probably the tracker was changed or something